diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17863 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 14.895626822157434,
+  "eval_steps": 500,
+  "global_step": 1440,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.013811383928571397,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4093.0,
+      "completions/mean_length": 604.600341796875,
+      "completions/mean_terminated_length": 555.7039184570312,
+      "completions/min_length": 2.0,
+      "completions/min_terminated_length": 2.0,
+      "epoch": 0.009329446064139942,
+      "grad_norm": 0.1397363245487213,
+      "learning_rate": 1e-06,
+      "loss": -0.048,
+      "num_tokens": 18561316.0,
+      "reward": 0.4968262016773224,
+      "reward_std": 0.2623644471168518,
+      "rewards/simpleverify_reward/mean": 0.496826171875,
+      "rewards/simpleverify_reward/std": 0.4999985992908478,
+      "step": 1
+    },
+    {
+      "clip_ratio/high_max": 0.0024105336706270464,
+      "clip_ratio/high_mean": 0.0011643338402791414,
+      "clip_ratio/low_mean": 0.0006752321896783542,
+      "clip_ratio/low_min": 4.17824294345337e-05,
+      "clip_ratio/region_mean": 0.0018395660154055804,
+      "epoch": 0.018658892128279883,
+      "grad_norm": 0.16730327904224396,
+      "learning_rate": 1e-06,
+      "loss": -0.0597,
+      "step": 2
+    },
+    {
+      "clip_ratio/high_max": 0.0023721187353658024,
+      "clip_ratio/high_mean": 0.0010058305524580646,
+      "clip_ratio/low_mean": 0.0006981411743254284,
+      "clip_ratio/low_min": 7.500308038288495e-05,
+      "clip_ratio/region_mean": 0.0017039716913131997,
+      "epoch": 0.027988338192419825,
+      "grad_norm": 0.1430075615644455,
+      "learning_rate": 1e-06,
+      "loss": -0.0098,
+      "step": 3
+    },
+    {
+      "clip_ratio/high_max": 0.00247967795439763,
+      "clip_ratio/high_mean": 0.0011456298452685587,
+      "clip_ratio/low_mean": 0.0007949689115775982,
+      "clip_ratio/low_min": 5.361851071938872e-05,
+      "clip_ratio/region_mean": 0.001940598725923337,
+      "epoch": 0.037317784256559766,
+      "grad_norm": 0.12244745343923569,
+      "learning_rate": 1e-06,
+      "loss": -0.0126,
+      "step": 4
+    },
+    {
+      "clip_ratio/high_max": 0.002327150032215286,
+      "clip_ratio/high_mean": 0.001072493227184168,
+      "clip_ratio/low_mean": 0.000989349700830644,
+      "clip_ratio/low_min": 0.00019277110914117657,
+      "clip_ratio/region_mean": 0.0020618429261958227,
+      "epoch": 0.04664723032069971,
+      "grad_norm": 0.12621773779392242,
+      "learning_rate": 1e-06,
+      "loss": 0.0301,
+      "step": 5
+    },
+    {
+      "clip_ratio/high_max": 0.003110960067715496,
+      "clip_ratio/high_mean": 0.0013379114570852835,
+      "clip_ratio/low_mean": 0.0014299873473646585,
+      "clip_ratio/low_min": 0.0001926458553498378,
+      "clip_ratio/region_mean": 0.0027678988117258996,
+      "epoch": 0.05597667638483965,
+      "grad_norm": 0.14269208908081055,
+      "learning_rate": 1e-06,
+      "loss": 0.0275,
+      "step": 6
+    },
+    {
+      "clip_ratio/high_max": 0.002738219467573799,
+      "clip_ratio/high_mean": 0.0013427227968350053,
+      "clip_ratio/low_mean": 0.0014824369209236465,
+      "clip_ratio/low_min": 0.00021844370439794147,
+      "clip_ratio/region_mean": 0.0028251596813788638,
+      "epoch": 0.0653061224489796,
+      "grad_norm": 0.1470305621623993,
+      "learning_rate": 1e-06,
+      "loss": 0.0004,
+      "step": 7
+    },
+    {
+      "clip_ratio/high_max": 0.002615711866383208,
+      "clip_ratio/high_mean": 0.0013420677387330215,
+      "clip_ratio/low_mean": 0.0012677105605689576,
+      "clip_ratio/low_min": 1.3812154975312296e-05,
+      "clip_ratio/region_mean": 0.0026097782974829897,
+      "epoch": 0.07463556851311953,
+      "grad_norm": 0.14808005094528198,
+      "learning_rate": 1e-06,
+      "loss": -0.0093,
+      "step": 8
+    },
+    {
+      "clip_ratio/high_max": 0.003145890757878078,
+      "clip_ratio/high_mean": 0.001340907871053787,
+      "clip_ratio/low_mean": 0.0014410102194233332,
+      "clip_ratio/low_min": 0.00028794246645702515,
+      "clip_ratio/region_mean": 0.0027819181050290354,
+      "epoch": 0.08396501457725948,
+      "grad_norm": 0.2074601799249649,
+      "learning_rate": 1e-06,
+      "loss": 0.0545,
+      "step": 9
+    },
+    {
+      "clip_ratio/high_max": 0.0027386108340579085,
+      "clip_ratio/high_mean": 0.0013568961912824307,
+      "clip_ratio/low_mean": 0.0011387128906790167,
+      "clip_ratio/low_min": 7.724008537479676e-05,
+      "clip_ratio/region_mean": 0.0024956090419436805,
+      "epoch": 0.09329446064139942,
+      "grad_norm": 0.12807653844356537,
+      "learning_rate": 1e-06,
+      "loss": 0.0025,
+      "step": 10
+    },
+    {
+      "clip_ratio/high_max": 0.0025042851120815612,
+      "clip_ratio/high_mean": 0.0010798836501635378,
+      "clip_ratio/low_mean": 0.0011460282512416597,
+      "clip_ratio/low_min": 0.00019961351790698245,
+      "clip_ratio/region_mean": 0.0022259119286900386,
+      "epoch": 0.10262390670553936,
+      "grad_norm": 0.12754510343074799,
+      "learning_rate": 1e-06,
+      "loss": 0.0195,
+      "step": 11
+    },
+    {
+      "clip_ratio/high_max": 0.0024633969951537438,
+      "clip_ratio/high_mean": 0.0011423286305216607,
+      "clip_ratio/low_mean": 0.0011185748808202334,
+      "clip_ratio/low_min": 9.579515335644828e-05,
+      "clip_ratio/region_mean": 0.002260903529531788,
+      "epoch": 0.1119533527696793,
+      "grad_norm": 0.12464673817157745,
+      "learning_rate": 1e-06,
+      "loss": -0.0152,
+      "step": 12
+    },
+    {
+      "clip_ratio/high_max": 0.0028245924768270925,
+      "clip_ratio/high_mean": 0.0012071984747308306,
+      "clip_ratio/low_mean": 0.0009489665426372085,
+      "clip_ratio/low_min": 0.0002143360579793807,
+      "clip_ratio/region_mean": 0.0021561651010415517,
+      "epoch": 0.12128279883381925,
+      "grad_norm": 0.12157316505908966,
+      "learning_rate": 1e-06,
+      "loss": -0.055,
+      "step": 13
+    },
+    {
+      "clip_ratio/high_max": 0.0021413482863863464,
+      "clip_ratio/high_mean": 0.0010344708316551987,
+      "clip_ratio/low_mean": 0.0009623366968298797,
+      "clip_ratio/low_min": 0.00021045880657766247,
+      "clip_ratio/region_mean": 0.0019968075284850784,
+      "epoch": 0.1306122448979592,
+      "grad_norm": 0.13257591426372528,
+      "learning_rate": 1e-06,
+      "loss": 0.0125,
+      "step": 14
+    },
+    {
+      "clip_ratio/high_max": 0.0022448954623541795,
+      "clip_ratio/high_mean": 0.0010418586498417426,
+      "clip_ratio/low_mean": 0.0009500270971329883,
+      "clip_ratio/low_min": 4.0889111915021203e-05,
+      "clip_ratio/region_mean": 0.00199188577971654,
+      "epoch": 0.13994169096209913,
+      "grad_norm": 0.12219729274511337,
+      "learning_rate": 1e-06,
+      "loss": 0.0268,
+      "step": 15
+    },
+    {
+      "clip_ratio/high_max": 0.0023817530236556195,
+      "clip_ratio/high_mean": 0.001076888573152246,
+      "clip_ratio/low_mean": 0.0010004579180531437,
+      "clip_ratio/low_min": 0.00018702058241615305,
+      "clip_ratio/region_mean": 0.0020773464857484214,
+      "epoch": 0.14927113702623906,
+      "grad_norm": 0.12375976145267487,
+      "learning_rate": 1e-06,
+      "loss": -0.0197,
+      "step": 16
+    },
+    {
+      "clip_ratio/high_max": 0.002389628625678597,
+      "clip_ratio/high_mean": 0.0011601394489844097,
+      "clip_ratio/low_mean": 0.000944834671827266,
+      "clip_ratio/low_min": 0.00019620048169599613,
+      "clip_ratio/region_mean": 0.002104974104440771,
+      "epoch": 0.158600583090379,
+      "grad_norm": 0.13005965948104858,
+      "learning_rate": 1e-06,
+      "loss": 0.0155,
+      "step": 17
+    },
+    {
+      "clip_ratio/high_max": 0.002444075100356713,
+      "clip_ratio/high_mean": 0.0010035866580437869,
+      "clip_ratio/low_mean": 0.0011323059152346104,
+      "clip_ratio/low_min": 0.0002343704682061798,
+      "clip_ratio/region_mean": 0.0021358925296226516,
+      "epoch": 0.16793002915451896,
+      "grad_norm": 0.11367172002792358,
+      "learning_rate": 1e-06,
+      "loss": 0.0525,
+      "step": 18
+    },
+    {
+      "clip_ratio/high_max": 0.0028036852600052953,
+      "clip_ratio/high_mean": 0.00110845327435527,
+      "clip_ratio/low_mean": 0.0010357285573263653,
+      "clip_ratio/low_min": 0.00016398551724705612,
+      "clip_ratio/region_mean": 0.0021441818826133385,
+      "epoch": 0.1772594752186589,
+      "grad_norm": 0.12272893637418747,
+      "learning_rate": 1e-06,
+      "loss": 0.0291,
+      "step": 19
+    },
+    {
+      "clip_ratio/high_max": 0.002739102936175186,
+      "clip_ratio/high_mean": 0.0012600917070813011,
+      "clip_ratio/low_mean": 0.0009373848279210506,
+      "clip_ratio/low_min": 4.1304773731098976e-05,
+      "clip_ratio/region_mean": 0.0021974765477352776,
+      "epoch": 0.18658892128279883,
+      "grad_norm": 0.12404225766658783,
+      "learning_rate": 1e-06,
+      "loss": -0.0447,
+      "step": 20
+    },
+    {
+      "clip_ratio/high_max": 0.002547814503486734,
+      "clip_ratio/high_mean": 0.0011831440642708912,
+      "clip_ratio/low_mean": 0.0011259725797572173,
+      "clip_ratio/low_min": 0.00013308052984939422,
+      "clip_ratio/region_mean": 0.0023091166876838543,
+      "epoch": 0.19591836734693877,
+      "grad_norm": 0.13162827491760254,
+      "learning_rate": 1e-06,
+      "loss": 0.0289,
+      "step": 21
+    },
+    {
+      "clip_ratio/high_max": 0.0024744910915615037,
+      "clip_ratio/high_mean": 0.001208832789416192,
+      "clip_ratio/low_mean": 0.0011467382500995882,
+      "clip_ratio/low_min": 0.00014194786672305781,
+      "clip_ratio/region_mean": 0.002355571064981632,
+      "epoch": 0.20524781341107873,
+      "grad_norm": 0.11738279461860657,
+      "learning_rate": 1e-06,
+      "loss": -0.0238,
+      "step": 22
+    },
+    {
+      "clip_ratio/high_max": 0.0028532593059935607,
+      "clip_ratio/high_mean": 0.0012577767956827302,
+      "clip_ratio/low_mean": 0.001016283655189909,
+      "clip_ratio/low_min": 7.209397517726757e-05,
+      "clip_ratio/region_mean": 0.0022740604690625332,
+      "epoch": 0.21457725947521866,
+      "grad_norm": 0.12402624636888504,
+      "learning_rate": 1e-06,
+      "loss": -0.0227,
+      "step": 23
+    },
+    {
+      "clip_ratio/high_max": 0.0025150998117169365,
+      "clip_ratio/high_mean": 0.001034343171340879,
+      "clip_ratio/low_mean": 0.0011613182941800915,
+      "clip_ratio/low_min": 0.00020887598293484189,
+      "clip_ratio/region_mean": 0.0021956614291411825,
+      "epoch": 0.2239067055393586,
+      "grad_norm": 0.13170954585075378,
+      "learning_rate": 1e-06,
+      "loss": -0.014,
+      "step": 24
+    },
+    {
+      "clip_ratio/high_max": 0.0026861606020247564,
+      "clip_ratio/high_mean": 0.0012759202218148857,
+      "clip_ratio/low_mean": 0.0012954406629432924,
+      "clip_ratio/low_min": 0.00022114499461167725,
+      "clip_ratio/region_mean": 0.0025713608774822205,
+      "epoch": 0.23323615160349853,
+      "grad_norm": 0.12683427333831787,
+      "learning_rate": 1e-06,
+      "loss": 0.0117,
+      "step": 25
+    },
+    {
+      "clip_ratio/high_max": 0.0024499935971107334,
+      "clip_ratio/high_mean": 0.0011090274092566688,
+      "clip_ratio/low_mean": 0.001383901217195671,
+      "clip_ratio/low_min": 0.0003855242366626044,
+      "clip_ratio/region_mean": 0.0024929286373662762,
+      "epoch": 0.2425655976676385,
+      "grad_norm": 0.12441050261259079,
+      "learning_rate": 1e-06,
+      "loss": 0.017,
+      "step": 26
+    },
+    {
+      "clip_ratio/high_max": 0.0027391781550250016,
+      "clip_ratio/high_mean": 0.001410774842952378,
+      "clip_ratio/low_mean": 0.0013861599618394393,
+      "clip_ratio/low_min": 0.0002453724773658905,
+      "clip_ratio/region_mean": 0.002796934793877881,
+      "epoch": 0.2518950437317784,
+      "grad_norm": 0.13218040764331818,
+      "learning_rate": 1e-06,
+      "loss": 0.0027,
+      "step": 27
+    },
+    {
+      "clip_ratio/high_max": 0.0027316369305481203,
+      "clip_ratio/high_mean": 0.0012518592266133055,
+      "clip_ratio/low_mean": 0.0013879930957045872,
+      "clip_ratio/low_min": 0.0002513652771085617,
+      "clip_ratio/region_mean": 0.0026398523114039563,
+      "epoch": 0.2612244897959184,
+      "grad_norm": 0.12012336403131485,
+      "learning_rate": 1e-06,
+      "loss": 0.022,
+      "step": 28
+    },
+    {
+      "clip_ratio/high_max": 0.00247526651219232,
+      "clip_ratio/high_mean": 0.0012796688424714375,
+      "clip_ratio/low_mean": 0.001452852968213847,
+      "clip_ratio/low_min": 0.00023619042349309893,
+      "clip_ratio/region_mean": 0.002732521796133369,
+      "epoch": 0.2705539358600583,
+      "grad_norm": 0.12793003022670746,
+      "learning_rate": 1e-06,
+      "loss": -0.0111,
+      "step": 29
+    },
+    {
+      "clip_ratio/high_max": 0.002717217183089815,
+      "clip_ratio/high_mean": 0.0012803186491510132,
+      "clip_ratio/low_mean": 0.001327859441516921,
+      "clip_ratio/low_min": 0.00014889834528730717,
+      "clip_ratio/region_mean": 0.002608178161608521,
+      "epoch": 0.27988338192419826,
+      "grad_norm": 0.12293572723865509,
+      "learning_rate": 1e-06,
+      "loss": -0.0498,
+      "step": 30
+    },
+    {
+      "clip_ratio/high_max": 0.002836730760463979,
+      "clip_ratio/high_mean": 0.0012897639244329184,
+      "clip_ratio/low_mean": 0.001453427266824292,
+      "clip_ratio/low_min": 0.0002986752597280429,
+      "clip_ratio/region_mean": 0.0027431912603788078,
+      "epoch": 0.2892128279883382,
+      "grad_norm": 0.12500274181365967,
+      "learning_rate": 1e-06,
+      "loss": 0.0282,
+      "step": 31
+    },
+    {
+      "clip_ratio/high_max": 0.002470735300448723,
+      "clip_ratio/high_mean": 0.0011536918573256116,
+      "clip_ratio/low_mean": 0.0013952195295132697,
+      "clip_ratio/low_min": 0.00020339445291028824,
+      "clip_ratio/region_mean": 0.0025489114195806906,
+      "epoch": 0.29854227405247813,
+      "grad_norm": 0.13123559951782227,
+      "learning_rate": 1e-06,
+      "loss": 0.0281,
+      "step": 32
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015973772321428603,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4071.0,
+      "completions/mean_length": 605.833984375,
+      "completions/mean_terminated_length": 549.1778564453125,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 0.30787172011661806,
+      "grad_norm": 0.1392851620912552,
+      "learning_rate": 1e-06,
+      "loss": 0.0085,
+      "num_tokens": 36893228.0,
+      "reward": 0.5213449001312256,
+      "reward_std": 0.2461894452571869,
+      "rewards/simpleverify_reward/mean": 0.5213448405265808,
+      "rewards/simpleverify_reward/std": 0.4995529055595398,
+      "step": 33
+    },
+    {
+      "clip_ratio/high_max": 0.002154866528144339,
+      "clip_ratio/high_mean": 0.0009558827587170526,
+      "clip_ratio/low_mean": 0.0006220669692993397,
+      "clip_ratio/low_min": 4.9355419832863845e-05,
+      "clip_ratio/region_mean": 0.0015779497516632546,
+      "epoch": 0.317201166180758,
+      "grad_norm": 0.14256572723388672,
+      "learning_rate": 1e-06,
+      "loss": -0.0135,
+      "step": 34
+    },
+    {
+      "clip_ratio/high_max": 0.002520695641578641,
+      "clip_ratio/high_mean": 0.0010729252098826692,
+      "clip_ratio/low_mean": 0.0006731027169735171,
+      "clip_ratio/low_min": 5.905377838644199e-05,
+      "clip_ratio/region_mean": 0.0017460279559600167,
+      "epoch": 0.32653061224489793,
+      "grad_norm": 0.14906616508960724,
+      "learning_rate": 1e-06,
+      "loss": 0.0183,
+      "step": 35
+    },
+    {
+      "clip_ratio/high_max": 0.0023729700624244288,
+      "clip_ratio/high_mean": 0.0010100933868670836,
+      "clip_ratio/low_mean": 0.000690194572598557,
+      "clip_ratio/low_min": 6.126456173660699e-05,
+      "clip_ratio/region_mean": 0.0017002879903884605,
+      "epoch": 0.3358600583090379,
+      "grad_norm": 0.13483555614948273,
+      "learning_rate": 1e-06,
+      "loss": -0.0105,
+      "step": 36
+    },
+    {
+      "clip_ratio/high_max": 0.002428283325571101,
+      "clip_ratio/high_mean": 0.0010372213218943216,
+      "clip_ratio/low_mean": 0.0006968513553147204,
+      "clip_ratio/low_min": 8.627349325251998e-05,
+      "clip_ratio/region_mean": 0.0017340726844849996,
+      "epoch": 0.34518950437317786,
+      "grad_norm": 0.12697209417819977,
+      "learning_rate": 1e-06,
+      "loss": 0.035,
+      "step": 37
+    },
+    {
+      "clip_ratio/high_max": 0.002426805476716254,
+      "clip_ratio/high_mean": 0.0010459239347255789,
+      "clip_ratio/low_mean": 0.0008609316391812172,
+      "clip_ratio/low_min": 2.8912171728734393e-05,
+      "clip_ratio/region_mean": 0.0019068555993726477,
+      "epoch": 0.3545189504373178,
+      "grad_norm": 0.12941433489322662,
+      "learning_rate": 1e-06,
+      "loss": -0.059,
+      "step": 38
+    },
+    {
+      "clip_ratio/high_max": 0.002500117028830573,
+      "clip_ratio/high_mean": 0.0009735620005812962,
+      "clip_ratio/low_mean": 0.0010004887990362477,
+      "clip_ratio/low_min": 0.00021374362040660344,
+      "clip_ratio/region_mean": 0.001974050741409883,
+      "epoch": 0.3638483965014577,
+      "grad_norm": 0.12562952935695648,
+      "learning_rate": 1e-06,
+      "loss": 0.004,
+      "step": 39
+    },
+    {
+      "clip_ratio/high_max": 0.002515172745916061,
+      "clip_ratio/high_mean": 0.0010752747330116108,
+      "clip_ratio/low_mean": 0.0009461420777370222,
+      "clip_ratio/low_min": 0.00011673462176986504,
+      "clip_ratio/region_mean": 0.0020214168616803363,
+      "epoch": 0.37317784256559766,
+      "grad_norm": 0.13119079172611237,
+      "learning_rate": 1e-06,
+      "loss": 0.0267,
+      "step": 40
+    },
+    {
+      "clip_ratio/high_max": 0.002653695162734948,
+      "clip_ratio/high_mean": 0.0011458707340352703,
+      "clip_ratio/low_mean": 0.0010781829259940423,
+      "clip_ratio/low_min": 0.0001908422373162466,
+      "clip_ratio/region_mean": 0.0022240536200115457,
+      "epoch": 0.3825072886297376,
+      "grad_norm": 0.1273055523633957,
+      "learning_rate": 1e-06,
+      "loss": 0.0333,
+      "step": 41
+    },
+    {
+      "clip_ratio/high_max": 0.002564333553891629,
+      "clip_ratio/high_mean": 0.001194845957797952,
+      "clip_ratio/low_mean": 0.0013181191789044533,
+      "clip_ratio/low_min": 0.0002340795017516939,
+      "clip_ratio/region_mean": 0.0025129651476163417,
+      "epoch": 0.39183673469387753,
+      "grad_norm": 0.13553135097026825,
+      "learning_rate": 1e-06,
+      "loss": 0.0488,
+      "step": 42
+    },
+    {
+      "clip_ratio/high_max": 0.002800189373374451,
+      "clip_ratio/high_mean": 0.0012944095906277653,
+      "clip_ratio/low_mean": 0.0011908328633580822,
+      "clip_ratio/low_min": 0.00015302571773645468,
+      "clip_ratio/region_mean": 0.002485242410330102,
+      "epoch": 0.40116618075801747,
+      "grad_norm": 0.1388290971517563,
+      "learning_rate": 1e-06,
+      "loss": -0.0287,
+      "step": 43
+    },
+    {
+      "clip_ratio/high_max": 0.0025861069443635643,
+      "clip_ratio/high_mean": 0.0011626795130723622,
+      "clip_ratio/low_mean": 0.0012623116344911978,
+      "clip_ratio/low_min": 0.00012071824676240794,
+      "clip_ratio/region_mean": 0.0024249911366496235,
+      "epoch": 0.41049562682215746,
+      "grad_norm": 0.13290996849536896,
+      "learning_rate": 1e-06,
+      "loss": 0.0106,
+      "step": 44
+    },
+    {
+      "clip_ratio/high_max": 0.002650840317073744,
+      "clip_ratio/high_mean": 0.0012656280014198273,
+      "clip_ratio/low_mean": 0.0010785593658511061,
+      "clip_ratio/low_min": 8.578704819228733e-05,
+      "clip_ratio/region_mean": 0.0023441873709089123,
+      "epoch": 0.4198250728862974,
+      "grad_norm": 0.12145407497882843,
+      "learning_rate": 1e-06,
+      "loss": -0.0204,
+      "step": 45
+    },
+    {
+      "clip_ratio/high_max": 0.0027479734053486027,
+      "clip_ratio/high_mean": 0.001259499247680651,
+      "clip_ratio/low_mean": 0.0013141406761860708,
+      "clip_ratio/low_min": 0.00020944891184626613,
+      "clip_ratio/region_mean": 0.0025736399620654993,
+      "epoch": 0.4291545189504373,
+      "grad_norm": 0.1341453492641449,
+      "learning_rate": 1e-06,
+      "loss": 0.0245,
+      "step": 46
+    },
+    {
+      "clip_ratio/high_max": 0.002799942667479627,
+      "clip_ratio/high_mean": 0.001302520242461469,
+      "clip_ratio/low_mean": 0.001102146818084293,
+      "clip_ratio/low_min": 0.00012456459808163345,
+      "clip_ratio/region_mean": 0.002404667073278688,
+      "epoch": 0.43848396501457726,
+      "grad_norm": 0.13179446756839752,
+      "learning_rate": 1e-06,
+      "loss": -0.0716,
+      "step": 47
+    },
+    {
+      "clip_ratio/high_max": 0.0019659979589050636,
+      "clip_ratio/high_mean": 0.0009074569661606802,
+      "clip_ratio/low_mean": 0.0010306584172212752,
+      "clip_ratio/low_min": 0.0001766020905051846,
+      "clip_ratio/region_mean": 0.0019381153761059977,
+      "epoch": 0.4478134110787172,
+      "grad_norm": 0.11361198127269745,
+      "learning_rate": 1e-06,
+      "loss": 0.0251,
+      "step": 48
+    },
+    {
+      "clip_ratio/high_max": 0.002903555490775034,
+      "clip_ratio/high_mean": 0.001160112125944579,
+      "clip_ratio/low_mean": 0.0010674140103219543,
+      "clip_ratio/low_min": 0.00013766964366368484,
+      "clip_ratio/region_mean": 0.002227526143542491,
+      "epoch": 0.45714285714285713,
+      "grad_norm": 0.12459854781627655,
+      "learning_rate": 1e-06,
+      "loss": -0.0059,
+      "step": 49
+    },
+    {
+      "clip_ratio/high_max": 0.0021034674427937716,
+      "clip_ratio/high_mean": 0.0009915491064020898,
+      "clip_ratio/low_mean": 0.0010019534202001523,
+      "clip_ratio/low_min": 8.00598563728272e-05,
+      "clip_ratio/region_mean": 0.001993502512050327,
+      "epoch": 0.46647230320699706,
+      "grad_norm": 0.1285279095172882,
+      "learning_rate": 1e-06,
+      "loss": 0.045,
+      "step": 50
+    },
+    {
+      "clip_ratio/high_max": 0.002185424615163356,
+      "clip_ratio/high_mean": 0.0010555866901995614,
+      "clip_ratio/low_mean": 0.0010579091213003267,
+      "clip_ratio/low_min": 0.00010131898125109728,
+      "clip_ratio/region_mean": 0.0021134958078619093,
+      "epoch": 0.47580174927113705,
+      "grad_norm": 0.12838125228881836,
+      "learning_rate": 1e-06,
+      "loss": -0.0143,
+      "step": 51
+    },
+    {
+      "clip_ratio/high_max": 0.00245238406205317,
+      "clip_ratio/high_mean": 0.0010763966020022053,
+      "clip_ratio/low_mean": 0.0010526140322326683,
+      "clip_ratio/low_min": 0.00015946100575092714,
+      "clip_ratio/region_mean": 0.002129010666976683,
+      "epoch": 0.485131195335277,
+      "grad_norm": 0.12845982611179352,
+      "learning_rate": 1e-06,
+      "loss": 0.001,
+      "step": 52
+    },
+    {
+      "clip_ratio/high_max": 0.0025971759750973433,
+      "clip_ratio/high_mean": 0.0010989311995217577,
+      "clip_ratio/low_mean": 0.000848133853651234,
+      "clip_ratio/low_min": 4.5184184273239225e-05,
+      "clip_ratio/region_mean": 0.0019470650586299598,
+      "epoch": 0.4944606413994169,
+      "grad_norm": 0.12556153535842896,
+      "learning_rate": 1e-06,
+      "loss": 0.0073,
+      "step": 53
+    },
+    {
+      "clip_ratio/high_max": 0.0018824939033947885,
+      "clip_ratio/high_mean": 0.0008808384827716509,
+      "clip_ratio/low_mean": 0.0011637852112471592,
+      "clip_ratio/low_min": 0.0001357701694360003,
+      "clip_ratio/region_mean": 0.0020446236740099266,
+      "epoch": 0.5037900874635568,
+      "grad_norm": 0.12901367247104645,
+      "learning_rate": 1e-06,
+      "loss": 0.0621,
+      "step": 54
+    },
+    {
+      "clip_ratio/high_max": 0.0025989841306000017,
+      "clip_ratio/high_mean": 0.0010859335525310598,
+      "clip_ratio/low_mean": 0.0011131775827379897,
+      "clip_ratio/low_min": 0.00014724131960974773,
+      "clip_ratio/region_mean": 0.00219911116437288,
+      "epoch": 0.5131195335276968,
+      "grad_norm": 0.13426685333251953,
+      "learning_rate": 1e-06,
+      "loss": 0.0392,
+      "step": 55
+    },
+    {
+      "clip_ratio/high_max": 0.0026810664203367196,
+      "clip_ratio/high_mean": 0.001214465457451297,
+      "clip_ratio/low_mean": 0.0010145335327251814,
+      "clip_ratio/low_min": 0.00021548767927015433,
+      "clip_ratio/region_mean": 0.002228998993814457,
+      "epoch": 0.5224489795918368,
+      "grad_norm": 0.13487224280834198,
+      "learning_rate": 1e-06,
+      "loss": -0.0119,
+      "step": 56
+    },
+    {
+      "clip_ratio/high_max": 0.0027388824746594764,
+      "clip_ratio/high_mean": 0.0012344783390290104,
+      "clip_ratio/low_mean": 0.0011397648631827906,
+      "clip_ratio/low_min": 6.230076996871503e-05,
+      "clip_ratio/region_mean": 0.002374243202211801,
+      "epoch": 0.5317784256559767,
+      "grad_norm": 0.1314925104379654,
+      "learning_rate": 1e-06,
+      "loss": -0.0286,
+      "step": 57
+    },
+    {
+      "clip_ratio/high_max": 0.0025781981530599296,
+      "clip_ratio/high_mean": 0.0012049272190779448,
+      "clip_ratio/low_mean": 0.0013043091021245345,
+      "clip_ratio/low_min": 0.00019696603885677177,
+      "clip_ratio/region_mean": 0.002509236292098649,
+      "epoch": 0.5411078717201167,
+      "grad_norm": 0.12213936448097229,
+      "learning_rate": 1e-06,
+      "loss": -0.0069,
+      "step": 58
+    },
+    {
+      "clip_ratio/high_max": 0.002602761422167532,
+      "clip_ratio/high_mean": 0.001203188716317527,
+      "clip_ratio/low_mean": 0.001175208184577059,
+      "clip_ratio/low_min": 6.338120056170737e-05,
+      "clip_ratio/region_mean": 0.0023783968717907555,
+      "epoch": 0.5504373177842565,
+      "grad_norm": 0.1302884817123413,
+      "learning_rate": 1e-06,
+      "loss": 0.003,
+      "step": 59
+    },
+    {
+      "clip_ratio/high_max": 0.00260514669935219,
+      "clip_ratio/high_mean": 0.0012445204774849117,
+      "clip_ratio/low_mean": 0.001062608298525447,
+      "clip_ratio/low_min": 0.0001771707375155529,
+      "clip_ratio/region_mean": 0.0023071288014762104,
+      "epoch": 0.5597667638483965,
+      "grad_norm": 0.12155602127313614,
+      "learning_rate": 1e-06,
+      "loss": -0.0112,
+      "step": 60
+    },
+    {
+      "clip_ratio/high_max": 0.0026845285610761493,
+      "clip_ratio/high_mean": 0.0012388419709168375,
+      "clip_ratio/low_mean": 0.0011018094155588187,
+      "clip_ratio/low_min": 0.00012176146992715076,
+      "clip_ratio/region_mean": 0.00234065131371608,
+      "epoch": 0.5690962099125364,
+      "grad_norm": 0.11204208433628082,
+      "learning_rate": 1e-06,
+      "loss": -0.0361,
+      "step": 61
+    },
+    {
+      "clip_ratio/high_max": 0.0023121083868318237,
+      "clip_ratio/high_mean": 0.0011512128767208196,
+      "clip_ratio/low_mean": 0.0011560534112504683,
+      "clip_ratio/low_min": 0.00012153263014624827,
+      "clip_ratio/region_mean": 0.0023072662224876694,
+      "epoch": 0.5784256559766764,
+      "grad_norm": 0.12549570202827454,
+      "learning_rate": 1e-06,
+      "loss": 0.0251,
+      "step": 62
+    },
+    {
+      "clip_ratio/high_max": 0.0026555501099210232,
+      "clip_ratio/high_mean": 0.0012307019424042664,
+      "clip_ratio/low_mean": 0.0012997674675716553,
+      "clip_ratio/low_min": 0.0002600036677904427,
+      "clip_ratio/region_mean": 0.002530469442717731,
+      "epoch": 0.5877551020408164,
+      "grad_norm": 0.13226014375686646,
+      "learning_rate": 1e-06,
+      "loss": 0.0108,
+      "step": 63
+    },
+    {
+      "clip_ratio/high_max": 0.002542233094573021,
+      "clip_ratio/high_mean": 0.0011151730759593192,
+      "clip_ratio/low_mean": 0.001162676668172935,
+      "clip_ratio/low_min": 9.582073016645154e-05,
+      "clip_ratio/region_mean": 0.0022778497732360847,
+      "epoch": 0.5970845481049563,
+      "grad_norm": 0.1157546266913414,
+      "learning_rate": 1e-06,
+      "loss": -0.0076,
+      "step": 64
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.014927455357142905,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4057.0,
+      "completions/mean_length": 616.0746459960938,
+      "completions/mean_terminated_length": 563.341064453125,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 0.6064139941690962,
+      "grad_norm": 0.13193269073963165,
+      "learning_rate": 1e-06,
+      "loss": 0.0043,
+      "num_tokens": 55696457.0,
+      "reward": 0.5386091470718384,
+      "reward_std": 0.23196251690387726,
+      "rewards/simpleverify_reward/mean": 0.5386090874671936,
+      "rewards/simpleverify_reward/std": 0.498515784740448,
+      "step": 65
+    },
+    {
+      "clip_ratio/high_max": 0.002328980182937812,
+      "clip_ratio/high_mean": 0.0009799439176276792,
+      "clip_ratio/low_mean": 0.0006488139661087189,
+      "clip_ratio/low_min": 5.3973290050635114e-05,
+      "clip_ratio/region_mean": 0.001628757905564271,
+      "epoch": 0.6157434402332361,
+      "grad_norm": 0.1179637759923935,
+      "learning_rate": 1e-06,
+      "loss": -0.0024,
+      "step": 66
+    },
+    {
+      "clip_ratio/high_max": 0.002294747951964382,
+      "clip_ratio/high_mean": 0.0009669599385233596,
+      "clip_ratio/low_mean": 0.0005805053824587958,
+      "clip_ratio/low_min": 2.3299446183955297e-05,
+      "clip_ratio/region_mean": 0.0015474653046112508,
+      "epoch": 0.6250728862973761,
+      "grad_norm": 0.12232931703329086,
+      "learning_rate": 1e-06,
+      "loss": -0.0121,
+      "step": 67
+    },
+    {
+      "clip_ratio/high_max": 0.0019414879061514512,
+      "clip_ratio/high_mean": 0.0008062621145654703,
+      "clip_ratio/low_mean": 0.0006693491923215333,
+      "clip_ratio/low_min": 9.827238864090759e-05,
+      "clip_ratio/region_mean": 0.0014756113305338658,
+      "epoch": 0.634402332361516,
+      "grad_norm": 0.12383954226970673,
+      "learning_rate": 1e-06,
+      "loss": 0.0716,
+      "step": 68
+    },
+    {
+      "clip_ratio/high_max": 0.002290696393174585,
+      "clip_ratio/high_mean": 0.0010502979894226883,
+      "clip_ratio/low_mean": 0.0006360281422530534,
+      "clip_ratio/low_min": 3.04950572171947e-05,
+      "clip_ratio/region_mean": 0.0016863261116668582,
+      "epoch": 0.643731778425656,
+      "grad_norm": 0.12718434631824493,
+      "learning_rate": 1e-06,
+      "loss": 0.0009,
+      "step": 69
+    },
+    {
+      "clip_ratio/high_max": 0.0023595283746544737,
+      "clip_ratio/high_mean": 0.0010088914568768814,
+      "clip_ratio/low_mean": 0.000681449559124303,
+      "clip_ratio/low_min": 6.201896030688658e-05,
+      "clip_ratio/region_mean": 0.001690341036010068,
+      "epoch": 0.6530612244897959,
+      "grad_norm": 0.12831735610961914,
+      "learning_rate": 1e-06,
+      "loss": -0.0054,
+      "step": 70
+    },
+    {
+      "clip_ratio/high_max": 0.002613749631564133,
+      "clip_ratio/high_mean": 0.0012710145165328868,
+      "clip_ratio/low_mean": 0.0006204042911122087,
+      "clip_ratio/low_min": 3.946598280890612e-05,
+      "clip_ratio/region_mean": 0.001891418818559032,
+      "epoch": 0.6623906705539359,
+      "grad_norm": 0.11891865730285645,
+      "learning_rate": 1e-06,
+      "loss": -0.0785,
+      "step": 71
+    },
+    {
+      "clip_ratio/high_max": 0.0019938925761380233,
+      "clip_ratio/high_mean": 0.0008738418837310746,
+      "clip_ratio/low_mean": 0.0008569636047468521,
+      "clip_ratio/low_min": 0.00013989033141115215,
+      "clip_ratio/region_mean": 0.0017308055175817572,
+      "epoch": 0.6717201166180758,
+      "grad_norm": 0.12106315791606903,
+      "learning_rate": 1e-06,
+      "loss": -0.0,
+      "step": 72
+    },
+    {
+      "clip_ratio/high_max": 0.0020695106395578478,
+      "clip_ratio/high_mean": 0.0009604874248907436,
+      "clip_ratio/low_mean": 0.000847232684463961,
+      "clip_ratio/low_min": 9.820732611842686e-05,
+      "clip_ratio/region_mean": 0.0018077200875268318,
+      "epoch": 0.6810495626822157,
+      "grad_norm": 0.11928822845220566,
+      "learning_rate": 1e-06,
+      "loss": -0.0151,
+      "step": 73
+    },
+    {
+      "clip_ratio/high_max": 0.00259551964700222,
+      "clip_ratio/high_mean": 0.0011883105362358037,
+      "clip_ratio/low_mean": 0.0009380183073517401,
+      "clip_ratio/low_min": 0.0001481371818954358,
+      "clip_ratio/region_mean": 0.0021263288144837134,
+      "epoch": 0.6903790087463557,
+      "grad_norm": 0.11544797569513321,
+      "learning_rate": 1e-06,
+      "loss": -0.0434,
+      "step": 74
+    },
+    {
+      "clip_ratio/high_max": 0.002432831723126583,
+      "clip_ratio/high_mean": 0.0010206585338892182,
+      "clip_ratio/low_mean": 0.0010139798687305301,
+      "clip_ratio/low_min": 0.00013609098277811427,
+      "clip_ratio/region_mean": 0.0020346384335425682,
+      "epoch": 0.6997084548104956,
+      "grad_norm": 0.11292950809001923,
+      "learning_rate": 1e-06,
+      "loss": 0.0182,
+      "step": 75
+    },
+    {
+      "clip_ratio/high_max": 0.002513583305699285,
+      "clip_ratio/high_mean": 0.0010161203063034918,
+      "clip_ratio/low_mean": 0.0011661909265967552,
+      "clip_ratio/low_min": 0.00014336886943056015,
+      "clip_ratio/region_mean": 0.0021823112765559927,
+      "epoch": 0.7090379008746356,
+      "grad_norm": 0.12492461502552032,
+      "learning_rate": 1e-06,
+      "loss": 0.0654,
+      "step": 76
+    },
+    {
+      "clip_ratio/high_max": 0.0023220856455736794,
+      "clip_ratio/high_mean": 0.0011234527482884005,
+      "clip_ratio/low_mean": 0.0011635668743110728,
+      "clip_ratio/low_min": 0.00017433294124202803,
+      "clip_ratio/region_mean": 0.002287019589857664,
+      "epoch": 0.7183673469387755,
+      "grad_norm": 0.12138062715530396,
+      "learning_rate": 1e-06,
+      "loss": 0.0237,
+      "step": 77
+    },
+    {
+      "clip_ratio/high_max": 0.0021567578660324216,
+      "clip_ratio/high_mean": 0.0010381691681686789,
+      "clip_ratio/low_mean": 0.0011469302444311325,
+      "clip_ratio/low_min": 0.0001505892769273487,
+      "clip_ratio/region_mean": 0.0021850994307897054,
+      "epoch": 0.7276967930029155,
+      "grad_norm": 0.14336355030536652,
+      "learning_rate": 1e-06,
+      "loss": 0.0249,
+      "step": 78
+    },
+    {
+      "clip_ratio/high_max": 0.002793032857880462,
+      "clip_ratio/high_mean": 0.0012763544618792366,
+      "clip_ratio/low_mean": 0.0010872337788896402,
+      "clip_ratio/low_min": 6.249236321309581e-05,
+      "clip_ratio/region_mean": 0.0023635882171220146,
+      "epoch": 0.7370262390670554,
+      "grad_norm": 0.13491053879261017,
+      "learning_rate": 1e-06,
+      "loss": -0.0055,
+      "step": 79
+    },
+    {
+      "clip_ratio/high_max": 0.0023751558182993904,
+      "clip_ratio/high_mean": 0.001017768619931303,
+      "clip_ratio/low_mean": 0.0011549389055289794,
+      "clip_ratio/low_min": 9.652048993302742e-05,
+      "clip_ratio/region_mean": 0.0021727075218223035,
+      "epoch": 0.7463556851311953,
+      "grad_norm": 0.12230775505304337,
+      "learning_rate": 1e-06,
+      "loss": 0.0404,
+      "step": 80
+    },
+    {
+      "clip_ratio/high_max": 0.002409681343124248,
+      "clip_ratio/high_mean": 0.0011061313925893046,
+      "clip_ratio/low_mean": 0.0011409010257921182,
+      "clip_ratio/low_min": 3.7721911212429404e-05,
+      "clip_ratio/region_mean": 0.0022470324402092956,
+      "epoch": 0.7556851311953353,
+      "grad_norm": 0.11919881403446198,
+      "learning_rate": 1e-06,
+      "loss": -0.0187,
+      "step": 81
+    },
+    {
+      "clip_ratio/high_max": 0.0023957524172146805,
+      "clip_ratio/high_mean": 0.0011061043751396937,
+      "clip_ratio/low_mean": 0.0012182273931102827,
+      "clip_ratio/low_min": 0.0001539431141281966,
+      "clip_ratio/region_mean": 0.002324331711861305,
+      "epoch": 0.7650145772594752,
+      "grad_norm": 0.12190832942724228,
+      "learning_rate": 1e-06,
+      "loss": -0.0079,
+      "step": 82
+    },
+    {
+      "clip_ratio/high_max": 0.0023902300745248795,
+      "clip_ratio/high_mean": 0.0011275278434368374,
+      "clip_ratio/low_mean": 0.0009957713482435793,
+      "clip_ratio/low_min": 0.00010872465827560518,
+      "clip_ratio/region_mean": 0.002123299171216786,
+      "epoch": 0.7743440233236152,
+      "grad_norm": 0.11692412942647934,
+      "learning_rate": 1e-06,
+      "loss": -0.0226,
+      "step": 83
+    },
+    {
+      "clip_ratio/high_max": 0.0024930099170887843,
+      "clip_ratio/high_mean": 0.0010037593092420138,
+      "clip_ratio/low_mean": 0.0011639502772595733,
+      "clip_ratio/low_min": 0.00019019318006030517,
+      "clip_ratio/region_mean": 0.00216770960832946,
+      "epoch": 0.7836734693877551,
+      "grad_norm": 0.1270606964826584,
+      "learning_rate": 1e-06,
+      "loss": 0.0167,
+      "step": 84
+    },
+    {
+      "clip_ratio/high_max": 0.002333626318431925,
+      "clip_ratio/high_mean": 0.0010789451935124816,
+      "clip_ratio/low_mean": 0.0011098564245912712,
+      "clip_ratio/low_min": 0.00016086923460534308,
+      "clip_ratio/region_mean": 0.002188801568991039,
+      "epoch": 0.793002915451895,
+      "grad_norm": 0.13270266354084015,
+      "learning_rate": 1e-06,
+      "loss": 0.0276,
+      "step": 85
+    },
+    {
+      "clip_ratio/high_max": 0.002080409445625264,
+      "clip_ratio/high_mean": 0.0009758403903106228,
+      "clip_ratio/low_mean": 0.001007801060040947,
+      "clip_ratio/low_min": 0.00010497249058971647,
+      "clip_ratio/region_mean": 0.0019836414576275274,
+      "epoch": 0.8023323615160349,
+      "grad_norm": 0.12897315621376038,
+      "learning_rate": 1e-06,
+      "loss": 0.0203,
+      "step": 86
+    },
+    {
+      "clip_ratio/high_max": 0.0024895572860259563,
+      "clip_ratio/high_mean": 0.0010984627297148108,
+      "clip_ratio/low_mean": 0.0011370503343641758,
+      "clip_ratio/low_min": 0.00016351762496924493,
+      "clip_ratio/region_mean": 0.002235513020423241,
+      "epoch": 0.8116618075801749,
+      "grad_norm": 0.12234677374362946,
+      "learning_rate": 1e-06,
+      "loss": -0.0097,
+      "step": 87
+    },
+    {
+      "clip_ratio/high_max": 0.002670867368578911,
+      "clip_ratio/high_mean": 0.0011942313103645574,
+      "clip_ratio/low_mean": 0.0009526853827992454,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.002146916667697951,
+      "epoch": 0.8209912536443149,
+      "grad_norm": 0.12400404363870621,
+      "learning_rate": 1e-06,
+      "loss": -0.0188,
+      "step": 88
+    },
+    {
+      "clip_ratio/high_max": 0.0024363375050597824,
+      "clip_ratio/high_mean": 0.0011551420502655674,
+      "clip_ratio/low_mean": 0.001058746944181621,
+      "clip_ratio/low_min": 0.00020158013012405718,
+      "clip_ratio/region_mean": 0.002213888983533252,
+      "epoch": 0.8303206997084548,
+      "grad_norm": 0.12443777173757553,
+      "learning_rate": 1e-06,
+      "loss": -0.0388,
+      "step": 89
+    },
+    {
+      "clip_ratio/high_max": 0.0022709333352395333,
+      "clip_ratio/high_mean": 0.001171580715890741,
+      "clip_ratio/low_mean": 0.0010937659062619787,
+      "clip_ratio/low_min": 2.5980607460951433e-05,
+      "clip_ratio/region_mean": 0.002265346636704635,
+      "epoch": 0.8396501457725948,
+      "grad_norm": 0.11693009734153748,
+      "learning_rate": 1e-06,
+      "loss": -0.0212,
+      "step": 90
+    },
+    {
+      "clip_ratio/high_max": 0.0024615596048533916,
+      "clip_ratio/high_mean": 0.001110722660087049,
+      "clip_ratio/low_mean": 0.001268210409762105,
+      "clip_ratio/low_min": 6.843203846074175e-05,
+      "clip_ratio/region_mean": 0.002378933015279472,
+      "epoch": 0.8489795918367347,
+      "grad_norm": 0.12528090178966522,
+      "learning_rate": 1e-06,
+      "loss": 0.0024,
+      "step": 91
+    },
+    {
+      "clip_ratio/high_max": 0.002672517905011773,
+      "clip_ratio/high_mean": 0.001037338446622016,
+      "clip_ratio/low_mean": 0.0012742921826429665,
+      "clip_ratio/low_min": 0.0001510250622231979,
+      "clip_ratio/region_mean": 0.002311630640178919,
+      "epoch": 0.8583090379008746,
+      "grad_norm": 0.1276797354221344,
+      "learning_rate": 1e-06,
+      "loss": 0.0289,
+      "step": 92
+    },
+    {
+      "clip_ratio/high_max": 0.0025825381017057225,
+      "clip_ratio/high_mean": 0.0010888647302635945,
+      "clip_ratio/low_mean": 0.0015090270280779805,
+      "clip_ratio/low_min": 0.0001346614990325179,
+      "clip_ratio/region_mean": 0.002597891798359342,
+      "epoch": 0.8676384839650145,
+      "grad_norm": 0.12030808627605438,
+      "learning_rate": 1e-06,
+      "loss": 0.0248,
+      "step": 93
+    },
+    {
+      "clip_ratio/high_max": 0.002218307043222012,
+      "clip_ratio/high_mean": 0.0010351078635721933,
+      "clip_ratio/low_mean": 0.0012474247851059772,
+      "clip_ratio/low_min": 0.00011207708666916005,
+      "clip_ratio/region_mean": 0.0022825326450401917,
+      "epoch": 0.8769679300291545,
+      "grad_norm": 0.11305058002471924,
+      "learning_rate": 1e-06,
+      "loss": 0.0442,
+      "step": 94
+    },
+    {
+      "clip_ratio/high_max": 0.0023669503134442493,
+      "clip_ratio/high_mean": 0.0010745935978775378,
+      "clip_ratio/low_mean": 0.001287968538235873,
+      "clip_ratio/low_min": 0.00017820822722569574,
+      "clip_ratio/region_mean": 0.0023625621179235168,
+      "epoch": 0.8862973760932945,
+      "grad_norm": 0.12017299979925156,
+      "learning_rate": 1e-06,
+      "loss": -0.0124,
+      "step": 95
+    },
+    {
+      "clip_ratio/high_max": 0.0025466951410635374,
+      "clip_ratio/high_mean": 0.00105575274938019,
+      "clip_ratio/low_mean": 0.001242694845132064,
+      "clip_ratio/low_min": 4.785998498846311e-05,
+      "clip_ratio/region_mean": 0.0022984475508565083,
+      "epoch": 0.8956268221574344,
+      "grad_norm": 0.13566216826438904,
+      "learning_rate": 1e-06,
+      "loss": 0.0247,
+      "step": 96
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015973772321428603,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4094.0,
+      "completions/mean_length": 608.2108154296875,
+      "completions/mean_terminated_length": 551.5933227539062,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 1.00932944606414,
+      "grad_norm": 0.12515154480934143,
+      "learning_rate": 1e-06,
+      "loss": 0.01,
+      "num_tokens": 74133166.0,
+      "reward": 0.5551409125328064,
+      "reward_std": 0.2201182246208191,
+      "rewards/simpleverify_reward/mean": 0.5551409125328064,
+      "rewards/simpleverify_reward/std": 0.49695885181427,
+      "step": 97
+    },
+    {
+      "clip_ratio/high_max": 0.0022979457571636885,
+      "clip_ratio/high_mean": 0.0009395265988132451,
+      "clip_ratio/low_mean": 0.0005775715308118379,
+      "clip_ratio/low_min": 1.569563028169796e-05,
+      "clip_ratio/region_mean": 0.0015170981350820512,
+      "epoch": 1.01865889212828,
+      "grad_norm": 0.1306575983762741,
+      "learning_rate": 1e-06,
+      "loss": 0.0205,
+      "step": 98
+    },
+    {
+      "clip_ratio/high_max": 0.002436228205624502,
+      "clip_ratio/high_mean": 0.0009995441268983996,
+      "clip_ratio/low_mean": 0.0005524815983335429,
+      "clip_ratio/low_min": 5.513196447282098e-05,
+      "clip_ratio/region_mean": 0.0015520257002208382,
+      "epoch": 1.0279883381924197,
+      "grad_norm": 0.12664695084095,
+      "learning_rate": 1e-06,
+      "loss": -0.0136,
+      "step": 99
+    },
+    {
+      "clip_ratio/high_max": 0.002102709284372395,
+      "clip_ratio/high_mean": 0.0008546321787434863,
+      "clip_ratio/low_mean": 0.0005916793070355197,
+      "clip_ratio/low_min": 3.160142841807101e-05,
+      "clip_ratio/region_mean": 0.0014463114894169848,
+      "epoch": 1.0373177842565597,
+      "grad_norm": 0.12008006870746613,
+      "learning_rate": 1e-06,
+      "loss": 0.0151,
+      "step": 100
+    },
+    {
+      "clip_ratio/high_max": 0.0021189190629229415,
+      "clip_ratio/high_mean": 0.0008503895369358361,
+      "clip_ratio/low_mean": 0.0005945331467955839,
+      "clip_ratio/low_min": 3.941907471016748e-05,
+      "clip_ratio/region_mean": 0.0014449226728174835,
+      "epoch": 1.0466472303206997,
+      "grad_norm": 0.13553902506828308,
+      "learning_rate": 1e-06,
+      "loss": 0.0655,
+      "step": 101
+    },
+    {
+      "clip_ratio/high_max": 0.0021607013914035633,
+      "clip_ratio/high_mean": 0.0008976444769359659,
+      "clip_ratio/low_mean": 0.0006076663830754114,
+      "clip_ratio/low_min": 2.47099651460303e-05,
+      "clip_ratio/region_mean": 0.0015053108218126,
+      "epoch": 1.0559766763848397,
+      "grad_norm": 0.11779744178056717,
+      "learning_rate": 1e-06,
+      "loss": 0.0305,
+      "step": 102
+    },
+    {
+      "clip_ratio/high_max": 0.002528972690925002,
+      "clip_ratio/high_mean": 0.0009709200603538193,
+      "clip_ratio/low_mean": 0.0008465050232189242,
+      "clip_ratio/low_min": 0.00010089717852679314,
+      "clip_ratio/region_mean": 0.001817425072658807,
+      "epoch": 1.0653061224489795,
+      "grad_norm": 0.1320830136537552,
+      "learning_rate": 1e-06,
+      "loss": 0.0292,
+      "step": 103
+    },
+    {
+      "clip_ratio/high_max": 0.0023619561616214924,
+      "clip_ratio/high_mean": 0.0010557245223026257,
+      "clip_ratio/low_mean": 0.0007360655290540308,
+      "clip_ratio/low_min": 8.46889788590488e-05,
+      "clip_ratio/region_mean": 0.0017917900295287836,
+      "epoch": 1.0746355685131195,
+      "grad_norm": 0.12015572190284729,
+      "learning_rate": 1e-06,
+      "loss": -0.0462,
+      "step": 104
+    },
+    {
+      "clip_ratio/high_max": 0.002650026151968632,
+      "clip_ratio/high_mean": 0.001030772862577578,
+      "clip_ratio/low_mean": 0.0007596392151754117,
+      "clip_ratio/low_min": 1.6914749721763656e-05,
+      "clip_ratio/region_mean": 0.0017904120977618732,
+      "epoch": 1.0839650145772595,
+      "grad_norm": 0.12176768481731415,
+      "learning_rate": 1e-06,
+      "loss": -0.0513,
+      "step": 105
+    },
+    {
+      "clip_ratio/high_max": 0.001983782567549497,
+      "clip_ratio/high_mean": 0.0009059975091076922,
+      "clip_ratio/low_mean": 0.0009869285931927152,
+      "clip_ratio/low_min": 9.539515849610325e-05,
+      "clip_ratio/region_mean": 0.00189292603317881,
+      "epoch": 1.0932944606413995,
+      "grad_norm": 0.12015856802463531,
+      "learning_rate": 1e-06,
+      "loss": 0.0096,
+      "step": 106
+    },
+    {
+      "clip_ratio/high_max": 0.0022477297970908694,
+      "clip_ratio/high_mean": 0.0010037471292889677,
+      "clip_ratio/low_mean": 0.0009222691951435991,
+      "clip_ratio/low_min": 0.00017914639465743676,
+      "clip_ratio/region_mean": 0.001926016280776821,
+      "epoch": 1.1026239067055394,
+      "grad_norm": 0.1231522411108017,
+      "learning_rate": 1e-06,
+      "loss": -0.0026,
+      "step": 107
+    },
+    {
+      "clip_ratio/high_max": 0.002572487028373871,
+      "clip_ratio/high_mean": 0.0011812613702204544,
+      "clip_ratio/low_mean": 0.0010373023178544827,
+      "clip_ratio/low_min": 7.952180021675304e-05,
+      "clip_ratio/region_mean": 0.0022185637135407887,
+      "epoch": 1.1119533527696792,
+      "grad_norm": 0.12579306960105896,
+      "learning_rate": 1e-06,
+      "loss": -0.0262,
+      "step": 108
+    },
+    {
+      "clip_ratio/high_max": 0.0023505327699240297,
+      "clip_ratio/high_mean": 0.0010666778580343816,
+      "clip_ratio/low_mean": 0.0009655847279645968,
+      "clip_ratio/low_min": 4.217499645164935e-05,
+      "clip_ratio/region_mean": 0.002032262593274936,
+      "epoch": 1.1212827988338192,
+      "grad_norm": 0.12550635635852814,
+      "learning_rate": 1e-06,
+      "loss": 0.0071,
+      "step": 109
+    },
+    {
+      "clip_ratio/high_max": 0.0021835168881807476,
+      "clip_ratio/high_mean": 0.001035374612911255,
+      "clip_ratio/low_mean": 0.0010122447183675831,
+      "clip_ratio/low_min": 5.1894465286750346e-05,
+      "clip_ratio/region_mean": 0.0020476193458307534,
+      "epoch": 1.1306122448979592,
+      "grad_norm": 0.12222646176815033,
+      "learning_rate": 1e-06,
+      "loss": -0.0129,
+      "step": 110
+    },
+    {
+      "clip_ratio/high_max": 0.0022354327084030956,
+      "clip_ratio/high_mean": 0.0010449666770000476,
+      "clip_ratio/low_mean": 0.0011392355554562528,
+      "clip_ratio/low_min": 0.00027197795679967385,
+      "clip_ratio/region_mean": 0.002184202225180343,
+      "epoch": 1.1399416909620992,
+      "grad_norm": 0.12912093102931976,
+      "learning_rate": 1e-06,
+      "loss": 0.0143,
+      "step": 111
+    },
+    {
+      "clip_ratio/high_max": 0.0025578335698810406,
+      "clip_ratio/high_mean": 0.0010727523331297562,
+      "clip_ratio/low_mean": 0.001113414178689709,
+      "clip_ratio/low_min": 7.72684725234285e-05,
+      "clip_ratio/region_mean": 0.0021861665445612743,
+      "epoch": 1.149271137026239,
+      "grad_norm": 0.12282103300094604,
+      "learning_rate": 1e-06,
+      "loss": 0.0258,
+      "step": 112
+    },
+    {
+      "clip_ratio/high_max": 0.0025094529046327807,
+      "clip_ratio/high_mean": 0.0010429745743749663,
+      "clip_ratio/low_mean": 0.001107412033888977,
+      "clip_ratio/low_min": 0.00014175496471580118,
+      "clip_ratio/region_mean": 0.0021503865646081977,
+      "epoch": 1.158600583090379,
+      "grad_norm": 0.12653689086437225,
+      "learning_rate": 1e-06,
+      "loss": 0.0325,
+      "step": 113
+    },
+    {
+      "clip_ratio/high_max": 0.00234408591495594,
+      "clip_ratio/high_mean": 0.001093566283088876,
+      "clip_ratio/low_mean": 0.0009334662954643136,
+      "clip_ratio/low_min": 9.101541945710778e-05,
+      "clip_ratio/region_mean": 0.002027032518526539,
+      "epoch": 1.167930029154519,
+      "grad_norm": 0.1202814131975174,
+      "learning_rate": 1e-06,
+      "loss": -0.0067,
+      "step": 114
+    },
+    {
+      "clip_ratio/high_max": 0.002573394274804741,
+      "clip_ratio/high_mean": 0.0011231584248889703,
+      "clip_ratio/low_mean": 0.0011573454685276374,
+      "clip_ratio/low_min": 0.00014929425742593594,
+      "clip_ratio/region_mean": 0.002280503911606502,
+      "epoch": 1.177259475218659,
+      "grad_norm": 0.13579046726226807,
+      "learning_rate": 1e-06,
+      "loss": 0.0051,
+      "step": 115
+    },
+    {
+      "clip_ratio/high_max": 0.0018269551655976102,
+      "clip_ratio/high_mean": 0.0008515936824551318,
+      "clip_ratio/low_mean": 0.0009331661731266649,
+      "clip_ratio/low_min": 0.00018841306246031309,
+      "clip_ratio/region_mean": 0.0017847598646767437,
+      "epoch": 1.186588921282799,
+      "grad_norm": 0.10874009877443314,
+      "learning_rate": 1e-06,
+      "loss": -0.0038,
+      "step": 116
+    },
+    {
+      "clip_ratio/high_max": 0.002437371775158681,
+      "clip_ratio/high_mean": 0.0010400869177829009,
+      "clip_ratio/low_mean": 0.0009712751398183173,
+      "clip_ratio/low_min": 9.253769621864194e-05,
+      "clip_ratio/region_mean": 0.002011362048506271,
+      "epoch": 1.1959183673469387,
+      "grad_norm": 0.1319679468870163,
+      "learning_rate": 1e-06,
+      "loss": -0.0193,
+      "step": 117
+    },
+    {
+      "clip_ratio/high_max": 0.0021046325637144037,
+      "clip_ratio/high_mean": 0.000982396406470798,
+      "clip_ratio/low_mean": 0.001065889733581571,
+      "clip_ratio/low_min": 8.660030835017096e-05,
+      "clip_ratio/region_mean": 0.0020482861291384324,
+      "epoch": 1.2052478134110787,
+      "grad_norm": 0.12338002026081085,
+      "learning_rate": 1e-06,
+      "loss": 0.0261,
+      "step": 118
+    },
+    {
+      "clip_ratio/high_max": 0.00250383437378332,
+      "clip_ratio/high_mean": 0.001061306054907618,
+      "clip_ratio/low_mean": 0.001043384545482695,
+      "clip_ratio/low_min": 5.890324700885685e-05,
+      "clip_ratio/region_mean": 0.0021046906913397834,
+      "epoch": 1.2145772594752187,
+      "grad_norm": 0.12722188234329224,
+      "learning_rate": 1e-06,
+      "loss": -0.0103,
+      "step": 119
+    },
+    {
+      "clip_ratio/high_max": 0.0023383466686937027,
+      "clip_ratio/high_mean": 0.0011093730536231305,
+      "clip_ratio/low_mean": 0.0010856035423785215,
+      "clip_ratio/low_min": 0.00018352546248934232,
+      "clip_ratio/region_mean": 0.0021949766305624507,
+      "epoch": 1.2239067055393587,
+      "grad_norm": 0.13765819370746613,
+      "learning_rate": 1e-06,
+      "loss": -0.0071,
+      "step": 120
+    },
+    {
+      "clip_ratio/high_max": 0.00228412381693488,
+      "clip_ratio/high_mean": 0.0010354996848036535,
+      "clip_ratio/low_mean": 0.0011145055323140696,
+      "clip_ratio/low_min": 0.00011972517313552089,
+      "clip_ratio/region_mean": 0.002150005275325384,
+      "epoch": 1.2332361516034984,
+      "grad_norm": 0.12663955986499786,
+      "learning_rate": 1e-06,
+      "loss": 0.0045,
+      "step": 121
+    },
+    {
+      "clip_ratio/high_max": 0.002686190706299385,
+      "clip_ratio/high_mean": 0.0012359067841316573,
+      "clip_ratio/low_mean": 0.0011155900865560398,
+      "clip_ratio/low_min": 4.857255407841876e-05,
+      "clip_ratio/region_mean": 0.002351496856135782,
+      "epoch": 1.2425655976676384,
+      "grad_norm": 0.13473036885261536,
+      "learning_rate": 1e-06,
+      "loss": 0.0353,
+      "step": 122
+    },
+    {
+      "clip_ratio/high_max": 0.0022190124800545163,
+      "clip_ratio/high_mean": 0.0009583649407431949,
+      "clip_ratio/low_mean": 0.0012937010324094445,
+      "clip_ratio/low_min": 0.0002862225883291103,
+      "clip_ratio/region_mean": 0.00225206594041083,
+      "epoch": 1.2518950437317784,
+      "grad_norm": 0.12195934355258942,
+      "learning_rate": 1e-06,
+      "loss": 0.0541,
+      "step": 123
+    },
+    {
+      "clip_ratio/high_max": 0.002577113380539231,
+      "clip_ratio/high_mean": 0.0011613152055360842,
+      "clip_ratio/low_mean": 0.0010268922815157566,
+      "clip_ratio/low_min": 0.00016467454406665638,
+      "clip_ratio/region_mean": 0.0021882075307075866,
+      "epoch": 1.2612244897959184,
+      "grad_norm": 0.12652873992919922,
+      "learning_rate": 1e-06,
+      "loss": -0.0178,
+      "step": 124
+    },
+    {
+      "clip_ratio/high_max": 0.002815144332998898,
+      "clip_ratio/high_mean": 0.0012095953352400102,
+      "clip_ratio/low_mean": 0.0011194726648682263,
+      "clip_ratio/low_min": 0.00016523562953807414,
+      "clip_ratio/region_mean": 0.0023290679382625967,
+      "epoch": 1.2705539358600584,
+      "grad_norm": 0.14093166589736938,
+      "learning_rate": 1e-06,
+      "loss": 0.0056,
+      "step": 125
+    },
+    {
+      "clip_ratio/high_max": 0.002683347811398562,
+      "clip_ratio/high_mean": 0.0011711218357959297,
+      "clip_ratio/low_mean": 0.0010713923256844282,
+      "clip_ratio/low_min": 9.582078564562835e-05,
+      "clip_ratio/region_mean": 0.0022425140923587605,
+      "epoch": 1.2798833819241984,
+      "grad_norm": 0.126437246799469,
+      "learning_rate": 1e-06,
+      "loss": -0.0557,
+      "step": 126
+    },
+    {
+      "clip_ratio/high_max": 0.0026961914845742285,
+      "clip_ratio/high_mean": 0.0011760315064748283,
+      "clip_ratio/low_mean": 0.0011478004598757252,
+      "clip_ratio/low_min": 0.00014482657752523664,
+      "clip_ratio/region_mean": 0.002323831991816405,
+      "epoch": 1.2892128279883381,
+      "grad_norm": 0.12861402332782745,
+      "learning_rate": 1e-06,
+      "loss": 0.012,
+      "step": 127
+    },
+    {
+      "clip_ratio/high_max": 0.0026040639204438776,
+      "clip_ratio/high_mean": 0.001162967051641317,
+      "clip_ratio/low_mean": 0.0010441786362207495,
+      "clip_ratio/low_min": 0.00019241467089159414,
+      "clip_ratio/region_mean": 0.002207145713327918,
+      "epoch": 1.2985422740524781,
+      "grad_norm": 0.11767813563346863,
+      "learning_rate": 1e-06,
+      "loss": -0.0296,
+      "step": 128
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0185546875,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4092.0,
+      "completions/mean_length": 629.6763916015625,
+      "completions/mean_terminated_length": 564.1438598632812,
+      "completions/min_length": 11.0,
+      "completions/min_terminated_length": 11.0,
+      "epoch": 1.3078717201166181,
+      "grad_norm": 0.12247433513402939,
+      "learning_rate": 1e-06,
+      "loss": -0.0098,
+      "num_tokens": 92890087.0,
+      "reward": 0.5526646375656128,
+      "reward_std": 0.20904073119163513,
+      "rewards/simpleverify_reward/mean": 0.5526646375656128,
+      "rewards/simpleverify_reward/std": 0.49722740054130554,
+      "step": 129
+    },
+    {
+      "clip_ratio/high_max": 0.0019482800671539735,
+      "clip_ratio/high_mean": 0.0008655789733893471,
+      "clip_ratio/low_mean": 0.0005754319145125919,
+      "clip_ratio/low_min": 1.1814744539151434e-05,
+      "clip_ratio/region_mean": 0.0014410108924494125,
+      "epoch": 1.3172011661807579,
+      "grad_norm": 0.12304199486970901,
+      "learning_rate": 1e-06,
+      "loss": -0.0123,
+      "step": 130
+    },
+    {
+      "clip_ratio/high_max": 0.0020373328152345493,
+      "clip_ratio/high_mean": 0.0008123245115712052,
+      "clip_ratio/low_mean": 0.0005806810022477293,
+      "clip_ratio/low_min": 1.0511267646506894e-05,
+      "clip_ratio/region_mean": 0.001393005524732871,
+      "epoch": 1.3265306122448979,
+      "grad_norm": 0.12376153469085693,
+      "learning_rate": 1e-06,
+      "loss": 0.0191,
+      "step": 131
+    },
+    {
+      "clip_ratio/high_max": 0.002150004576833453,
+      "clip_ratio/high_mean": 0.0009839979193202453,
+      "clip_ratio/low_mean": 0.0006266257241804851,
+      "clip_ratio/low_min": 3.0013732612133026e-05,
+      "clip_ratio/region_mean": 0.0016106235707411543,
+      "epoch": 1.3358600583090379,
+      "grad_norm": 0.11202864348888397,
+      "learning_rate": 1e-06,
+      "loss": -0.0337,
+      "step": 132
+    },
+    {
+      "clip_ratio/high_max": 0.0023354754666797817,
+      "clip_ratio/high_mean": 0.0010488585885468638,
+      "clip_ratio/low_mean": 0.0007199145620688796,
+      "clip_ratio/low_min": 4.286346666049212e-05,
+      "clip_ratio/region_mean": 0.0017687731706246268,
+      "epoch": 1.3451895043731779,
+      "grad_norm": 0.1314229518175125,
+      "learning_rate": 1e-06,
+      "loss": -0.0123,
+      "step": 133
+    },
+    {
+      "clip_ratio/high_max": 0.002094859824865125,
+      "clip_ratio/high_mean": 0.0009514483353996184,
+      "clip_ratio/low_mean": 0.0009187509349430911,
+      "clip_ratio/low_min": 7.496398575312924e-05,
+      "clip_ratio/region_mean": 0.0018701992739806883,
+      "epoch": 1.3545189504373178,
+      "grad_norm": 0.13386139273643494,
+      "learning_rate": 1e-06,
+      "loss": 0.0007,
+      "step": 134
+    },
+    {
+      "clip_ratio/high_max": 0.0017033482108672615,
+      "clip_ratio/high_mean": 0.0007622610100952443,
+      "clip_ratio/low_mean": 0.0007751339417154668,
+      "clip_ratio/low_min": 8.541487386537483e-05,
+      "clip_ratio/region_mean": 0.0015373949499917217,
+      "epoch": 1.3638483965014578,
+      "grad_norm": 0.1234249547123909,
+      "learning_rate": 1e-06,
+      "loss": 0.0476,
+      "step": 135
+    },
+    {
+      "clip_ratio/high_max": 0.0021852707795915194,
+      "clip_ratio/high_mean": 0.0010210138134425506,
+      "clip_ratio/low_mean": 0.0007802446561981924,
+      "clip_ratio/low_min": 2.5228006052202545e-05,
+      "clip_ratio/region_mean": 0.001801258469640743,
+      "epoch": 1.3731778425655976,
+      "grad_norm": 0.12055256217718124,
+      "learning_rate": 1e-06,
+      "loss": -0.015,
+      "step": 136
+    },
+    {
+      "clip_ratio/high_max": 0.0022463522109319456,
+      "clip_ratio/high_mean": 0.0009425262169315829,
+      "clip_ratio/low_mean": 0.0009136901262536412,
+      "clip_ratio/low_min": 8.54077770782169e-05,
+      "clip_ratio/region_mean": 0.001856216367741581,
+      "epoch": 1.3825072886297376,
+      "grad_norm": 0.13432520627975464,
+      "learning_rate": 1e-06,
+      "loss": 0.0201,
+      "step": 137
+    },
+    {
+      "clip_ratio/high_max": 0.0023435970360878855,
+      "clip_ratio/high_mean": 0.0009898163880279753,
+      "clip_ratio/low_mean": 0.0008015044604690047,
+      "clip_ratio/low_min": 1.6099947970360518e-05,
+      "clip_ratio/region_mean": 0.0017913208430400118,
+      "epoch": 1.3918367346938776,
+      "grad_norm": 0.1121196523308754,
+      "learning_rate": 1e-06,
+      "loss": 0.0019,
+      "step": 138
+    },
+    {
+      "clip_ratio/high_max": 0.002259192791825626,
+      "clip_ratio/high_mean": 0.0009618518561183009,
+      "clip_ratio/low_mean": 0.0008269448953797109,
+      "clip_ratio/low_min": 0.00010682068659662036,
+      "clip_ratio/region_mean": 0.001788796769687906,
+      "epoch": 1.4011661807580174,
+      "grad_norm": 0.11160842329263687,
+      "learning_rate": 1e-06,
+      "loss": -0.0077,
+      "step": 139
+    },
+    {
+      "clip_ratio/high_max": 0.001967262251127977,
+      "clip_ratio/high_mean": 0.0009652902335801627,
+      "clip_ratio/low_mean": 0.000870268569997279,
+      "clip_ratio/low_min": 7.185075901361415e-05,
+      "clip_ratio/region_mean": 0.001835558796301484,
+      "epoch": 1.4104956268221573,
+      "grad_norm": 0.12078763544559479,
+      "learning_rate": 1e-06,
+      "loss": 0.0019,
+      "step": 140
+    },
+    {
+      "clip_ratio/high_max": 0.0021865140297450125,
+      "clip_ratio/high_mean": 0.0008542850664525758,
+      "clip_ratio/low_mean": 0.0009709040750749409,
+      "clip_ratio/low_min": 7.92704704508651e-05,
+      "clip_ratio/region_mean": 0.0018251891888212413,
+      "epoch": 1.4198250728862973,
+      "grad_norm": 0.12320893257856369,
+      "learning_rate": 1e-06,
+      "loss": 0.0114,
+      "step": 141
+    },
+    {
+      "clip_ratio/high_max": 0.001995073282159865,
+      "clip_ratio/high_mean": 0.0009057185598067008,
+      "clip_ratio/low_mean": 0.0009829190021264367,
+      "clip_ratio/low_min": 0.00010751451736723538,
+      "clip_ratio/region_mean": 0.0018886375473812222,
+      "epoch": 1.4291545189504373,
+      "grad_norm": 0.1212090477347374,
+      "learning_rate": 1e-06,
+      "loss": 0.0207,
+      "step": 142
+    },
+    {
+      "clip_ratio/high_max": 0.002361187922360841,
+      "clip_ratio/high_mean": 0.000939879784709774,
+      "clip_ratio/low_mean": 0.0010385727482571383,
+      "clip_ratio/low_min": 0.00013409857638180256,
+      "clip_ratio/region_mean": 0.001978452506591566,
+      "epoch": 1.4384839650145773,
+      "grad_norm": 0.1244337186217308,
+      "learning_rate": 1e-06,
+      "loss": 0.0468,
+      "step": 143
+    },
+    {
+      "clip_ratio/high_max": 0.002408552565611899,
+      "clip_ratio/high_mean": 0.0010358501604059711,
+      "clip_ratio/low_mean": 0.0010924177477136254,
+      "clip_ratio/low_min": 0.00013807321647618664,
+      "clip_ratio/region_mean": 0.0021282679299474694,
+      "epoch": 1.4478134110787173,
+      "grad_norm": 0.1246391236782074,
+      "learning_rate": 1e-06,
+      "loss": -0.0085,
+      "step": 144
+    },
+    {
+      "clip_ratio/high_max": 0.0023346201924141496,
+      "clip_ratio/high_mean": 0.0009564228494127747,
+      "clip_ratio/low_mean": 0.0009136104872595752,
+      "clip_ratio/low_min": 7.062887016218156e-05,
+      "clip_ratio/region_mean": 0.0018700333093875088,
+      "epoch": 1.457142857142857,
+      "grad_norm": 0.1140151098370552,
+      "learning_rate": 1e-06,
+      "loss": 0.0227,
+      "step": 145
+    },
+    {
+      "clip_ratio/high_max": 0.0025495362642686814,
+      "clip_ratio/high_mean": 0.0009596319250704255,
+      "clip_ratio/low_mean": 0.0009141252467088634,
+      "clip_ratio/low_min": 1.5096617971721571e-05,
+      "clip_ratio/region_mean": 0.0018737571299425326,
+      "epoch": 1.466472303206997,
+      "grad_norm": 0.12359322607517242,
+      "learning_rate": 1e-06,
+      "loss": 0.0032,
+      "step": 146
+    },
+    {
+      "clip_ratio/high_max": 0.002412621382973157,
+      "clip_ratio/high_mean": 0.0009882448630378349,
+      "clip_ratio/low_mean": 0.0010921952780336142,
+      "clip_ratio/low_min": 0.0001224439729412552,
+      "clip_ratio/region_mean": 0.0020804400628549047,
+      "epoch": 1.475801749271137,
+      "grad_norm": 0.11045132577419281,
+      "learning_rate": 1e-06,
+      "loss": -0.0131,
+      "step": 147
+    },
+    {
+      "clip_ratio/high_max": 0.0023295222272281535,
+      "clip_ratio/high_mean": 0.0009796738995646592,
+      "clip_ratio/low_mean": 0.0009771978675416904,
+      "clip_ratio/low_min": 6.716169991705101e-05,
+      "clip_ratio/region_mean": 0.0019568717543734238,
+      "epoch": 1.485131195335277,
+      "grad_norm": 0.11553800851106644,
+      "learning_rate": 1e-06,
+      "loss": 0.0044,
+      "step": 148
+    },
+    {
+      "clip_ratio/high_max": 0.002373781768255867,
+      "clip_ratio/high_mean": 0.0009506058595434297,
+      "clip_ratio/low_mean": 0.0011260525534453336,
+      "clip_ratio/low_min": 0.00017674227183306357,
+      "clip_ratio/region_mean": 0.0020766584493685514,
+      "epoch": 1.4944606413994168,
+      "grad_norm": 0.11694473773241043,
+      "learning_rate": 1e-06,
+      "loss": 0.0176,
+      "step": 149
+    },
+    {
+      "clip_ratio/high_max": 0.0024190633703256026,
+      "clip_ratio/high_mean": 0.0009742487691255519,
+      "clip_ratio/low_mean": 0.0010879485234909225,
+      "clip_ratio/low_min": 0.0001114521337512997,
+      "clip_ratio/region_mean": 0.0020621973380912095,
+      "epoch": 1.5037900874635568,
+      "grad_norm": 0.11433441936969757,
+      "learning_rate": 1e-06,
+      "loss": -0.008,
+      "step": 150
+    },
+    {
+      "clip_ratio/high_max": 0.0023994471193873324,
+      "clip_ratio/high_mean": 0.0010328274074709043,
+      "clip_ratio/low_mean": 0.0011116435671283398,
+      "clip_ratio/low_min": 7.758178071526345e-05,
+      "clip_ratio/region_mean": 0.0021444709709612653,
+      "epoch": 1.5131195335276968,
+      "grad_norm": 0.11905031651258469,
+      "learning_rate": 1e-06,
+      "loss": -0.0183,
+      "step": 151
+    },
+    {
+      "clip_ratio/high_max": 0.002360008511459455,
+      "clip_ratio/high_mean": 0.001005455069389427,
+      "clip_ratio/low_mean": 0.0013339525867195334,
+      "clip_ratio/low_min": 0.0002080884496535873,
+      "clip_ratio/region_mean": 0.0023394076415570453,
+      "epoch": 1.5224489795918368,
+      "grad_norm": 0.1792532503604889,
+      "learning_rate": 1e-06,
+      "loss": 0.003,
+      "step": 152
+    },
+    {
+      "clip_ratio/high_max": 0.002323104050447,
+      "clip_ratio/high_mean": 0.0010053261012217263,
+      "clip_ratio/low_mean": 0.0010526574060349958,
+      "clip_ratio/low_min": 5.6324959587072954e-05,
+      "clip_ratio/region_mean": 0.002057983489066828,
+      "epoch": 1.5317784256559768,
+      "grad_norm": 0.12376930564641953,
+      "learning_rate": 1e-06,
+      "loss": 0.0337,
+      "step": 153
+    },
+    {
+      "clip_ratio/high_max": 0.002311307114723604,
+      "clip_ratio/high_mean": 0.0009813068609219044,
+      "clip_ratio/low_mean": 0.0011149177771585528,
+      "clip_ratio/low_min": 0.00012089210304111475,
+      "clip_ratio/region_mean": 0.002096224641718436,
+      "epoch": 1.5411078717201168,
+      "grad_norm": 0.12247707694768906,
+      "learning_rate": 1e-06,
+      "loss": 0.0177,
+      "step": 154
+    },
+    {
+      "clip_ratio/high_max": 0.0023144585284171626,
+      "clip_ratio/high_mean": 0.0010975206787406933,
+      "clip_ratio/low_mean": 0.001081908972992096,
+      "clip_ratio/low_min": 0.00012427501314959954,
+      "clip_ratio/region_mean": 0.0021794296917505562,
+      "epoch": 1.5504373177842565,
+      "grad_norm": 0.1175316721200943,
+      "learning_rate": 1e-06,
+      "loss": -0.0327,
+      "step": 155
+    },
+    {
+      "clip_ratio/high_max": 0.0021606523951049894,
+      "clip_ratio/high_mean": 0.0009900249970087316,
+      "clip_ratio/low_mean": 0.001106171705032466,
+      "clip_ratio/low_min": 8.075089408521308e-05,
+      "clip_ratio/region_mean": 0.002096196672937367,
+      "epoch": 1.5597667638483965,
+      "grad_norm": 0.12338127195835114,
+      "learning_rate": 1e-06,
+      "loss": -0.0043,
+      "step": 156
+    },
+    {
+      "clip_ratio/high_max": 0.0019886453155777417,
+      "clip_ratio/high_mean": 0.0008289978686661925,
+      "clip_ratio/low_mean": 0.0014164040258037858,
+      "clip_ratio/low_min": 0.00022766150232200744,
+      "clip_ratio/region_mean": 0.002245401898107957,
+      "epoch": 1.5690962099125363,
+      "grad_norm": 0.12142599374055862,
+      "learning_rate": 1e-06,
+      "loss": 0.0823,
+      "step": 157
+    },
+    {
+      "clip_ratio/high_max": 0.002212329964095261,
+      "clip_ratio/high_mean": 0.0010394248220109148,
+      "clip_ratio/low_mean": 0.0010776175986393355,
+      "clip_ratio/low_min": 4.852474376093596e-05,
+      "clip_ratio/region_mean": 0.002117042415193282,
+      "epoch": 1.5784256559766763,
+      "grad_norm": 0.10937978327274323,
+      "learning_rate": 1e-06,
+      "loss": -0.0099,
+      "step": 158
+    },
+    {
+      "clip_ratio/high_max": 0.0022363809766829945,
+      "clip_ratio/high_mean": 0.0009165700976154767,
+      "clip_ratio/low_mean": 0.001409736189089017,
+      "clip_ratio/low_min": 0.00010802586075442377,
+      "clip_ratio/region_mean": 0.002326306239410769,
+      "epoch": 1.5877551020408163,
+      "grad_norm": 0.12906044721603394,
+      "learning_rate": 1e-06,
+      "loss": 0.0243,
+      "step": 159
+    },
+    {
+      "clip_ratio/high_max": 0.0022373147803591564,
+      "clip_ratio/high_mean": 0.0009547759400447831,
+      "clip_ratio/low_mean": 0.001126741235566442,
+      "clip_ratio/low_min": 5.19391751367948e-05,
+      "clip_ratio/region_mean": 0.0020815171737922356,
+      "epoch": 1.5970845481049563,
+      "grad_norm": 0.11226746439933777,
+      "learning_rate": 1e-06,
+      "loss": -0.0214,
+      "step": 160
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0193568638392857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4089.0,
+      "completions/mean_length": 630.4431762695312,
+      "completions/mean_terminated_length": 562.0366821289062,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 1.6064139941690962,
+      "grad_norm": 0.13052918016910553,
+      "learning_rate": 1e-06,
+      "loss": 0.0433,
+      "num_tokens": 111551057.0,
+      "reward": 0.559884250164032,
+      "reward_std": 0.2066090852022171,
+      "rewards/simpleverify_reward/mean": 0.5598841905593872,
+      "rewards/simpleverify_reward/std": 0.4964096248149872,
+      "step": 161
+    },
+    {
+      "clip_ratio/high_max": 0.0019473583088256419,
+      "clip_ratio/high_mean": 0.0007786882524669636,
+      "clip_ratio/low_mean": 0.0006195795758685563,
+      "clip_ratio/low_min": 3.413532340346137e-05,
+      "clip_ratio/region_mean": 0.0013982678392494563,
+      "epoch": 1.6157434402332362,
+      "grad_norm": 0.11692752689123154,
+      "learning_rate": 1e-06,
+      "loss": 0.0253,
+      "step": 162
+    },
+    {
+      "clip_ratio/high_max": 0.001971633617358748,
+      "clip_ratio/high_mean": 0.0008421267611993244,
+      "clip_ratio/low_mean": 0.000644899269900634,
+      "clip_ratio/low_min": 3.6869254472549073e-05,
+      "clip_ratio/region_mean": 0.0014870260274619795,
+      "epoch": 1.6250728862973762,
+      "grad_norm": 0.10882455110549927,
+      "learning_rate": 1e-06,
+      "loss": 0.0236,
+      "step": 163
+    },
+    {
+      "clip_ratio/high_max": 0.0020929321253788657,
+      "clip_ratio/high_mean": 0.0009205508504237514,
+      "clip_ratio/low_mean": 0.0005239095698925667,
+      "clip_ratio/low_min": 8.321128916577436e-06,
+      "clip_ratio/region_mean": 0.0014444604312302545,
+      "epoch": 1.634402332361516,
+      "grad_norm": 0.11553318798542023,
+      "learning_rate": 1e-06,
+      "loss": -0.0412,
+      "step": 164
+    },
+    {
+      "clip_ratio/high_max": 0.00216324757275288,
+      "clip_ratio/high_mean": 0.0009188975582219427,
+      "clip_ratio/low_mean": 0.0006150362187327119,
+      "clip_ratio/low_min": 8.641618114779703e-05,
+      "clip_ratio/region_mean": 0.0015339337915065698,
+      "epoch": 1.643731778425656,
+      "grad_norm": 0.1298108547925949,
+      "learning_rate": 1e-06,
+      "loss": 0.0091,
+      "step": 165
+    },
+    {
+      "clip_ratio/high_max": 0.0020830659559578635,
+      "clip_ratio/high_mean": 0.0008959771093941526,
+      "clip_ratio/low_mean": 0.0008042422614380484,
+      "clip_ratio/low_min": 0.00014302170984592522,
+      "clip_ratio/region_mean": 0.001700219352642307,
+      "epoch": 1.6530612244897958,
+      "grad_norm": 0.12637987732887268,
+      "learning_rate": 1e-06,
+      "loss": 0.0232,
+      "step": 166
+    },
+    {
+      "clip_ratio/high_max": 0.00219865715916967,
+      "clip_ratio/high_mean": 0.0009272287534258794,
+      "clip_ratio/low_mean": 0.0006882043526275083,
+      "clip_ratio/low_min": 2.6838769372261595e-05,
+      "clip_ratio/region_mean": 0.0016154331242432818,
+      "epoch": 1.6623906705539357,
+      "grad_norm": 0.14283263683319092,
+      "learning_rate": 1e-06,
+      "loss": 0.0111,
+      "step": 167
+    },
+    {
+      "clip_ratio/high_max": 0.0024597456213086843,
+      "clip_ratio/high_mean": 0.0010091853146150243,
+      "clip_ratio/low_mean": 0.0007544972431787755,
+      "clip_ratio/low_min": 2.8480291803134605e-05,
+      "clip_ratio/region_mean": 0.0017636826160014607,
+      "epoch": 1.6717201166180757,
+      "grad_norm": 0.11524340510368347,
+      "learning_rate": 1e-06,
+      "loss": 0.01,
+      "step": 168
+    },
+    {
+      "clip_ratio/high_max": 0.002631057708640583,
+      "clip_ratio/high_mean": 0.001049726277415175,
+      "clip_ratio/low_mean": 0.0007909286287031136,
+      "clip_ratio/low_min": 9.947924900188809e-05,
+      "clip_ratio/region_mean": 0.001840654898842331,
+      "epoch": 1.6810495626822157,
+      "grad_norm": 0.11886724829673767,
+      "learning_rate": 1e-06,
+      "loss": -0.0221,
+      "step": 169
+    },
+    {
+      "clip_ratio/high_max": 0.0023531218539574184,
+      "clip_ratio/high_mean": 0.0008936539852584247,
+      "clip_ratio/low_mean": 0.0009439257082703989,
+      "clip_ratio/low_min": 0.00010847598605323583,
+      "clip_ratio/region_mean": 0.0018375797080807388,
+      "epoch": 1.6903790087463557,
+      "grad_norm": 0.1280161738395691,
+      "learning_rate": 1e-06,
+      "loss": 0.0174,
+      "step": 170
+    },
+    {
+      "clip_ratio/high_max": 0.0024234568045358174,
+      "clip_ratio/high_mean": 0.0009716022650536615,
+      "clip_ratio/low_mean": 0.0008536897275916999,
+      "clip_ratio/low_min": 7.567417560494505e-05,
+      "clip_ratio/region_mean": 0.0018252919617225416,
+      "epoch": 1.6997084548104957,
+      "grad_norm": 0.11074575036764145,
+      "learning_rate": 1e-06,
+      "loss": 0.0065,
+      "step": 171
+    },
+    {
+      "clip_ratio/high_max": 0.002372522772930097,
+      "clip_ratio/high_mean": 0.000934289766519214,
+      "clip_ratio/low_mean": 0.0010841503899428062,
+      "clip_ratio/low_min": 0.000123855254969385,
+      "clip_ratio/region_mean": 0.002018440172832925,
+      "epoch": 1.7090379008746357,
+      "grad_norm": 0.11330542713403702,
+      "learning_rate": 1e-06,
+      "loss": 0.0198,
+      "step": 172
+    },
+    {
+      "clip_ratio/high_max": 0.002437242215819424,
+      "clip_ratio/high_mean": 0.0010677119189494988,
+      "clip_ratio/low_mean": 0.0009012334085127804,
+      "clip_ratio/low_min": 0.00010626952553138835,
+      "clip_ratio/region_mean": 0.0019689453547471203,
+      "epoch": 1.7183673469387755,
+      "grad_norm": 0.12956194579601288,
+      "learning_rate": 1e-06,
+      "loss": -0.0177,
+      "step": 173
+    },
+    {
+      "clip_ratio/high_max": 0.002794365653244313,
+      "clip_ratio/high_mean": 0.001146391859947471,
+      "clip_ratio/low_mean": 0.0009963155862351414,
+      "clip_ratio/low_min": 0.00014578002082998864,
+      "clip_ratio/region_mean": 0.002142707380698994,
+      "epoch": 1.7276967930029155,
+      "grad_norm": 0.1224517673254013,
+      "learning_rate": 1e-06,
+      "loss": -0.005,
+      "step": 174
+    },
+    {
+      "clip_ratio/high_max": 0.0023935651261126623,
+      "clip_ratio/high_mean": 0.0010778728283185046,
+      "clip_ratio/low_mean": 0.0008526510118826991,
+      "clip_ratio/low_min": 4.828855071536964e-05,
+      "clip_ratio/region_mean": 0.0019305238674860448,
+      "epoch": 1.7370262390670554,
+      "grad_norm": 0.11898007243871689,
+      "learning_rate": 1e-06,
+      "loss": -0.0189,
+      "step": 175
+    },
+    {
+      "clip_ratio/high_max": 0.0022502077554236166,
+      "clip_ratio/high_mean": 0.0009356172595289536,
+      "clip_ratio/low_mean": 0.000968181702774018,
+      "clip_ratio/low_min": 0.00015928049015201395,
+      "clip_ratio/region_mean": 0.0019037989986827597,
+      "epoch": 1.7463556851311952,
+      "grad_norm": 0.13075068593025208,
+      "learning_rate": 1e-06,
+      "loss": -0.0001,
+      "step": 176
+    },
+    {
+      "clip_ratio/high_max": 0.002857508196029812,
+      "clip_ratio/high_mean": 0.0010823418851941824,
+      "clip_ratio/low_mean": 0.0009438116758246906,
+      "clip_ratio/low_min": 7.394163912977092e-05,
+      "clip_ratio/region_mean": 0.002026153582846746,
+      "epoch": 1.7556851311953352,
+      "grad_norm": 0.1229390874505043,
+      "learning_rate": 1e-06,
+      "loss": -0.0202,
+      "step": 177
+    },
+    {
+      "clip_ratio/high_max": 0.002277478364703711,
+      "clip_ratio/high_mean": 0.000946376841966412,
+      "clip_ratio/low_mean": 0.0009406296576344175,
+      "clip_ratio/low_min": 8.676519428263418e-05,
+      "clip_ratio/region_mean": 0.0018870064959628507,
+      "epoch": 1.7650145772594752,
+      "grad_norm": 0.13334843516349792,
+      "learning_rate": 1e-06,
+      "loss": 0.0081,
+      "step": 178
+    },
+    {
+      "clip_ratio/high_max": 0.002178639391786419,
+      "clip_ratio/high_mean": 0.0009134401789197,
+      "clip_ratio/low_mean": 0.0009079107312572887,
+      "clip_ratio/low_min": 0.0001339919417659985,
+      "clip_ratio/region_mean": 0.0018213508883491158,
+      "epoch": 1.7743440233236152,
+      "grad_norm": 0.11784674227237701,
+      "learning_rate": 1e-06,
+      "loss": 0.0013,
+      "step": 179
+    },
+    {
+      "clip_ratio/high_max": 0.0022050546249374747,
+      "clip_ratio/high_mean": 0.0009802271133594331,
+      "clip_ratio/low_mean": 0.0009644347010180354,
+      "clip_ratio/low_min": 5.919880186411319e-05,
+      "clip_ratio/region_mean": 0.001944661773450207,
+      "epoch": 1.7836734693877552,
+      "grad_norm": 0.12492590397596359,
+      "learning_rate": 1e-06,
+      "loss": -0.0113,
+      "step": 180
+    },
+    {
+      "clip_ratio/high_max": 0.002152851040591486,
+      "clip_ratio/high_mean": 0.0009666321675467771,
+      "clip_ratio/low_mean": 0.001131426943175029,
+      "clip_ratio/low_min": 7.19606687198393e-05,
+      "clip_ratio/region_mean": 0.0020980590998078696,
+      "epoch": 1.7930029154518952,
+      "grad_norm": 0.12422338128089905,
+      "learning_rate": 1e-06,
+      "loss": 0.0284,
+      "step": 181
+    },
+    {
+      "clip_ratio/high_max": 0.0021346145804272965,
+      "clip_ratio/high_mean": 0.0010018986722570844,
+      "clip_ratio/low_mean": 0.0009776542992767645,
+      "clip_ratio/low_min": 8.275802974822e-05,
+      "clip_ratio/region_mean": 0.0019795529296970926,
+      "epoch": 1.802332361516035,
+      "grad_norm": 0.12853233516216278,
+      "learning_rate": 1e-06,
+      "loss": -0.0186,
+      "step": 182
+    },
+    {
+      "clip_ratio/high_max": 0.002379450190346688,
+      "clip_ratio/high_mean": 0.0010064417983812746,
+      "clip_ratio/low_mean": 0.0010600036548567005,
+      "clip_ratio/low_min": 7.705206371610984e-05,
+      "clip_ratio/region_mean": 0.0020664455150836147,
+      "epoch": 1.811661807580175,
+      "grad_norm": 0.12457560747861862,
+      "learning_rate": 1e-06,
+      "loss": 0.0193,
+      "step": 183
+    },
+    {
+      "clip_ratio/high_max": 0.0025077843893086538,
+      "clip_ratio/high_mean": 0.0011306777996651363,
+      "clip_ratio/low_mean": 0.001088579920178745,
+      "clip_ratio/low_min": 8.311457168019842e-05,
+      "clip_ratio/region_mean": 0.0022192577525856905,
+      "epoch": 1.820991253644315,
+      "grad_norm": 0.11980686336755753,
+      "learning_rate": 1e-06,
+      "loss": -0.021,
+      "step": 184
+    },
+    {
+      "clip_ratio/high_max": 0.002225047763204202,
+      "clip_ratio/high_mean": 0.0008986193679447751,
+      "clip_ratio/low_mean": 0.0010928749852610053,
+      "clip_ratio/low_min": 0.00013986617159389425,
+      "clip_ratio/region_mean": 0.001991494413232431,
+      "epoch": 1.8303206997084547,
+      "grad_norm": 0.1230274960398674,
+      "learning_rate": 1e-06,
+      "loss": 0.0392,
+      "step": 185
+    },
+    {
+      "clip_ratio/high_max": 0.002219788177171722,
+      "clip_ratio/high_mean": 0.0009121713774220552,
+      "clip_ratio/low_mean": 0.0012891650330857374,
+      "clip_ratio/low_min": 0.00010991991166520165,
+      "clip_ratio/region_mean": 0.0022013364141457714,
+      "epoch": 1.8396501457725947,
+      "grad_norm": 0.1255647838115692,
+      "learning_rate": 1e-06,
+      "loss": 0.0514,
+      "step": 186
+    },
+    {
+      "clip_ratio/high_max": 0.0024172653138521127,
+      "clip_ratio/high_mean": 0.001157182727183681,
+      "clip_ratio/low_mean": 0.0009896736228256486,
+      "clip_ratio/low_min": 3.2526670111110434e-05,
+      "clip_ratio/region_mean": 0.0021468563718372025,
+      "epoch": 1.8489795918367347,
+      "grad_norm": 0.14806382358074188,
+      "learning_rate": 1e-06,
+      "loss": -0.0618,
+      "step": 187
+    },
+    {
+      "clip_ratio/high_max": 0.00214373854760197,
+      "clip_ratio/high_mean": 0.0009269417587347561,
+      "clip_ratio/low_mean": 0.001136309394496493,
+      "clip_ratio/low_min": 0.0001882109063444659,
+      "clip_ratio/region_mean": 0.00206325115141226,
+      "epoch": 1.8583090379008746,
+      "grad_norm": 0.1317405104637146,
+      "learning_rate": 1e-06,
+      "loss": 0.0117,
+      "step": 188
+    },
+    {
+      "clip_ratio/high_max": 0.0022073624495533295,
+      "clip_ratio/high_mean": 0.000885869047124288,
+      "clip_ratio/low_mean": 0.0012851926330768038,
+      "clip_ratio/low_min": 0.00021439420379465446,
+      "clip_ratio/region_mean": 0.0021710616783821024,
+      "epoch": 1.8676384839650146,
+      "grad_norm": 0.12194634228944778,
+      "learning_rate": 1e-06,
+      "loss": 0.0226,
+      "step": 189
+    },
+    {
+      "clip_ratio/high_max": 0.002016449892835226,
+      "clip_ratio/high_mean": 0.000939275791097316,
+      "clip_ratio/low_mean": 0.0010900188663072186,
+      "clip_ratio/low_min": 0.00014371005454449914,
+      "clip_ratio/region_mean": 0.0020292946574045345,
+      "epoch": 1.8769679300291546,
+      "grad_norm": 0.1152808740735054,
+      "learning_rate": 1e-06,
+      "loss": -0.0167,
+      "step": 190
+    },
+    {
+      "clip_ratio/high_max": 0.002209578982728999,
+      "clip_ratio/high_mean": 0.0010171312333113747,
+      "clip_ratio/low_mean": 0.001362909661111189,
+      "clip_ratio/low_min": 0.00021611467036564136,
+      "clip_ratio/region_mean": 0.002380040859861765,
+      "epoch": 1.8862973760932946,
+      "grad_norm": 0.13314864039421082,
+      "learning_rate": 1e-06,
+      "loss": 0.0363,
+      "step": 191
+    },
+    {
+      "clip_ratio/high_max": 0.002335728539037518,
+      "clip_ratio/high_mean": 0.001085810299628065,
+      "clip_ratio/low_mean": 0.0011010053312929813,
+      "clip_ratio/low_min": 0.00011865987471537665,
+      "clip_ratio/region_mean": 0.0021868156763957813,
+      "epoch": 1.8956268221574344,
+      "grad_norm": 0.12692096829414368,
+      "learning_rate": 1e-06,
+      "loss": -0.027,
+      "step": 192
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.019810267857142905,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4025.0,
+      "completions/mean_length": 630.1661376953125,
+      "completions/mean_terminated_length": 560.119384765625,
+      "completions/min_length": 4.0,
+      "completions/min_terminated_length": 4.0,
+      "epoch": 2.00932944606414,
+      "grad_norm": 0.1092682033777237,
+      "learning_rate": 1e-06,
+      "loss": -0.0044,
+      "num_tokens": 130140299.0,
+      "reward": 0.579659640789032,
+      "reward_std": 0.19633613526821136,
+      "rewards/simpleverify_reward/mean": 0.5796595811843872,
+      "rewards/simpleverify_reward/std": 0.49362218379974365,
+      "step": 193
+    },
+    {
+      "clip_ratio/high_max": 0.0020895358466077596,
+      "clip_ratio/high_mean": 0.0008441342015430564,
+      "clip_ratio/low_mean": 0.0006324653386400314,
+      "clip_ratio/low_min": 3.0290034374047536e-05,
+      "clip_ratio/region_mean": 0.0014765995219931938,
+      "epoch": 2.01865889212828,
+      "grad_norm": 0.1230829507112503,
+      "learning_rate": 1e-06,
+      "loss": -0.0328,
+      "step": 194
+    },
+    {
+      "clip_ratio/high_max": 0.00207390798459528,
+      "clip_ratio/high_mean": 0.0008967022058641305,
+      "clip_ratio/low_mean": 0.0005146934408912784,
+      "clip_ratio/low_min": 1.3510591998056043e-05,
+      "clip_ratio/region_mean": 0.001411395671311766,
+      "epoch": 2.02798833819242,
+      "grad_norm": 0.1321178525686264,
+      "learning_rate": 1e-06,
+      "loss": -0.0412,
+      "step": 195
+    },
+    {
+      "clip_ratio/high_max": 0.0021295985134202056,
+      "clip_ratio/high_mean": 0.0009101082832785323,
+      "clip_ratio/low_mean": 0.0006119706013123505,
+      "clip_ratio/low_min": 3.116128482361091e-05,
+      "clip_ratio/region_mean": 0.0015220788918668404,
+      "epoch": 2.03731778425656,
+      "grad_norm": 0.13616017997264862,
+      "learning_rate": 1e-06,
+      "loss": 0.032,
+      "step": 196
+    },
+    {
+      "clip_ratio/high_max": 0.002133457033778541,
+      "clip_ratio/high_mean": 0.0008904576734494185,
+      "clip_ratio/low_mean": 0.0007426326465065358,
+      "clip_ratio/low_min": 5.287952535582008e-05,
+      "clip_ratio/region_mean": 0.0016330903490597848,
+      "epoch": 2.0466472303206995,
+      "grad_norm": 0.12524496018886566,
+      "learning_rate": 1e-06,
+      "loss": -0.0134,
+      "step": 197
+    },
+    {
+      "clip_ratio/high_max": 0.0019300127423775848,
+      "clip_ratio/high_mean": 0.0008277610777440714,
+      "clip_ratio/low_mean": 0.0005185083236938226,
+      "clip_ratio/low_min": 2.8496213417383842e-05,
+      "clip_ratio/region_mean": 0.0013462693896144629,
+      "epoch": 2.0559766763848395,
+      "grad_norm": 0.10855976492166519,
+      "learning_rate": 1e-06,
+      "loss": -0.0366,
+      "step": 198
+    },
+    {
+      "clip_ratio/high_max": 0.002347043984627817,
+      "clip_ratio/high_mean": 0.0009639352228987264,
+      "clip_ratio/low_mean": 0.0006995748717599781,
+      "clip_ratio/low_min": 9.55956120378687e-05,
+      "clip_ratio/region_mean": 0.0016635101201245561,
+      "epoch": 2.0653061224489795,
+      "grad_norm": 0.12525874376296997,
+      "learning_rate": 1e-06,
+      "loss": 0.0054,
+      "step": 199
+    },
+    {
+      "clip_ratio/high_max": 0.0022166262278915383,
+      "clip_ratio/high_mean": 0.0008933486005844316,
+      "clip_ratio/low_mean": 0.0008267901648650877,
+      "clip_ratio/low_min": 2.9530114261433482e-05,
+      "clip_ratio/region_mean": 0.0017201387527165934,
+      "epoch": 2.0746355685131195,
+      "grad_norm": 0.13545803725719452,
+      "learning_rate": 1e-06,
+      "loss": 0.0656,
+      "step": 200
+    },
+    {
+      "clip_ratio/high_max": 0.0021380551261245273,
+      "clip_ratio/high_mean": 0.0008750795986998128,
+      "clip_ratio/low_mean": 0.0007241222647280665,
+      "clip_ratio/low_min": 7.284074399649398e-05,
+      "clip_ratio/region_mean": 0.0015992018743418157,
+      "epoch": 2.0839650145772595,
+      "grad_norm": 0.13121257722377777,
+      "learning_rate": 1e-06,
+      "loss": 0.0272,
+      "step": 201
+    },
+    {
+      "clip_ratio/high_max": 0.0018240930658066645,
+      "clip_ratio/high_mean": 0.0008700213911652099,
+      "clip_ratio/low_mean": 0.0008648789225844666,
+      "clip_ratio/low_min": 5.121918093209388e-05,
+      "clip_ratio/region_mean": 0.001734900288283825,
+      "epoch": 2.0932944606413995,
+      "grad_norm": 0.12381664663553238,
+      "learning_rate": 1e-06,
+      "loss": 0.025,
+      "step": 202
+    },
+    {
+      "clip_ratio/high_max": 0.0019286003152956255,
+      "clip_ratio/high_mean": 0.0008171215304173529,
+      "clip_ratio/low_mean": 0.0007734704777249135,
+      "clip_ratio/low_min": 6.245358144951751e-05,
+      "clip_ratio/region_mean": 0.0015905919462966267,
+      "epoch": 2.1026239067055394,
+      "grad_norm": 0.11860410124063492,
+      "learning_rate": 1e-06,
+      "loss": -0.0229,
+      "step": 203
+    },
+    {
+      "clip_ratio/high_max": 0.002014702717133332,
+      "clip_ratio/high_mean": 0.0008024648914215504,
+      "clip_ratio/low_mean": 0.0009128984147537267,
+      "clip_ratio/low_min": 5.4807868764328305e-05,
+      "clip_ratio/region_mean": 0.0017153633452835493,
+      "epoch": 2.1119533527696794,
+      "grad_norm": 0.11724842339754105,
+      "learning_rate": 1e-06,
+      "loss": 0.0373,
+      "step": 204
+    },
+    {
+      "clip_ratio/high_max": 0.002323144584806869,
+      "clip_ratio/high_mean": 0.0009170987123070518,
+      "clip_ratio/low_mean": 0.0008524646545993164,
+      "clip_ratio/low_min": 5.6268603657372296e-05,
+      "clip_ratio/region_mean": 0.0017695634087431245,
+      "epoch": 2.1212827988338194,
+      "grad_norm": 0.10747351497411728,
+      "learning_rate": 1e-06,
+      "loss": -0.0159,
+      "step": 205
+    },
+    {
+      "clip_ratio/high_max": 0.002367728498938959,
+      "clip_ratio/high_mean": 0.0009401266124768881,
+      "clip_ratio/low_mean": 0.0009918116229528096,
+      "clip_ratio/low_min": 6.329270854621427e-05,
+      "clip_ratio/region_mean": 0.0019319382408866659,
+      "epoch": 2.130612244897959,
+      "grad_norm": 0.11633989214897156,
+      "learning_rate": 1e-06,
+      "loss": 0.0026,
+      "step": 206
+    },
+    {
+      "clip_ratio/high_max": 0.002109798457240686,
+      "clip_ratio/high_mean": 0.0008825089917081641,
+      "clip_ratio/low_mean": 0.0009929372818078264,
+      "clip_ratio/low_min": 5.883443282073131e-05,
+      "clip_ratio/region_mean": 0.0018754462616925593,
+      "epoch": 2.139941690962099,
+      "grad_norm": 0.12492082267999649,
+      "learning_rate": 1e-06,
+      "loss": -0.0004,
+      "step": 207
+    },
+    {
+      "clip_ratio/high_max": 0.0025152976595563814,
+      "clip_ratio/high_mean": 0.0010189184904447757,
+      "clip_ratio/low_mean": 0.0008982785839179996,
+      "clip_ratio/low_min": 7.195130820036866e-05,
+      "clip_ratio/region_mean": 0.0019171971071045846,
+      "epoch": 2.149271137026239,
+      "grad_norm": 0.1098606139421463,
+      "learning_rate": 1e-06,
+      "loss": -0.0135,
+      "step": 208
+    },
+    {
+      "clip_ratio/high_max": 0.0022413926890294533,
+      "clip_ratio/high_mean": 0.000899045386177022,
+      "clip_ratio/low_mean": 0.000838117346575018,
+      "clip_ratio/low_min": 0.0001263984631805215,
+      "clip_ratio/region_mean": 0.0017371627400279976,
+      "epoch": 2.158600583090379,
+      "grad_norm": 0.11806578189134598,
+      "learning_rate": 1e-06,
+      "loss": -0.03,
+      "step": 209
+    },
+    {
+      "clip_ratio/high_max": 0.0021019109262852,
+      "clip_ratio/high_mean": 0.0008178860243788222,
+      "clip_ratio/low_mean": 0.0009143050228885841,
+      "clip_ratio/low_min": 3.181555985065643e-05,
+      "clip_ratio/region_mean": 0.0017321909981546924,
+      "epoch": 2.167930029154519,
+      "grad_norm": 0.11653382331132889,
+      "learning_rate": 1e-06,
+      "loss": 0.0303,
+      "step": 210
+    },
+    {
+      "clip_ratio/high_max": 0.0023580261040478945,
+      "clip_ratio/high_mean": 0.0009531390278425533,
+      "clip_ratio/low_mean": 0.00113502951717237,
+      "clip_ratio/low_min": 8.734521907172166e-05,
+      "clip_ratio/region_mean": 0.00208816855592886,
+      "epoch": 2.177259475218659,
+      "grad_norm": 0.1219073086977005,
+      "learning_rate": 1e-06,
+      "loss": 0.0192,
+      "step": 211
+    },
+    {
+      "clip_ratio/high_max": 0.002315523350262083,
+      "clip_ratio/high_mean": 0.0009459816174057778,
+      "clip_ratio/low_mean": 0.0010800387535709888,
+      "clip_ratio/low_min": 0.00014688047667732462,
+      "clip_ratio/region_mean": 0.002026020381890703,
+      "epoch": 2.186588921282799,
+      "grad_norm": 0.127249076962471,
+      "learning_rate": 1e-06,
+      "loss": 0.0073,
+      "step": 212
+    },
+    {
+      "clip_ratio/high_max": 0.0023402224978781305,
+      "clip_ratio/high_mean": 0.0009386223282490391,
+      "clip_ratio/low_mean": 0.0011775325474445708,
+      "clip_ratio/low_min": 7.335049667744897e-05,
+      "clip_ratio/region_mean": 0.0021161549011594616,
+      "epoch": 2.195918367346939,
+      "grad_norm": 0.11784949153661728,
+      "learning_rate": 1e-06,
+      "loss": -0.0055,
+      "step": 213
+    },
+    {
+      "clip_ratio/high_max": 0.0022509518203150947,
+      "clip_ratio/high_mean": 0.0010360320047766436,
+      "clip_ratio/low_mean": 0.0010909208776865853,
+      "clip_ratio/low_min": 9.097713518713135e-05,
+      "clip_ratio/region_mean": 0.002126952924299985,
+      "epoch": 2.205247813411079,
+      "grad_norm": 0.12944884598255157,
+      "learning_rate": 1e-06,
+      "loss": 0.0245,
+      "step": 214
+    },
+    {
+      "clip_ratio/high_max": 0.002420110467937775,
+      "clip_ratio/high_mean": 0.0009371957312396262,
+      "clip_ratio/low_mean": 0.0010524861718295142,
+      "clip_ratio/low_min": 9.543488522467669e-05,
+      "clip_ratio/region_mean": 0.0019896818994311616,
+      "epoch": 2.2145772594752184,
+      "grad_norm": 0.13038748502731323,
+      "learning_rate": 1e-06,
+      "loss": -0.0012,
+      "step": 215
+    },
+    {
+      "clip_ratio/high_max": 0.002297401319083292,
+      "clip_ratio/high_mean": 0.0009000710924738087,
+      "clip_ratio/low_mean": 0.0010327846011932706,
+      "clip_ratio/low_min": 8.079760846158024e-05,
+      "clip_ratio/region_mean": 0.0019328556300024502,
+      "epoch": 2.2239067055393584,
+      "grad_norm": 0.11851400882005692,
+      "learning_rate": 1e-06,
+      "loss": 0.0023,
+      "step": 216
+    },
+    {
+      "clip_ratio/high_max": 0.00250308339309413,
+      "clip_ratio/high_mean": 0.0009396381992701208,
+      "clip_ratio/low_mean": 0.0010309597673767712,
+      "clip_ratio/low_min": 3.620628922362812e-05,
+      "clip_ratio/region_mean": 0.0019705979866557755,
+      "epoch": 2.2332361516034984,
+      "grad_norm": 0.1116507425904274,
+      "learning_rate": 1e-06,
+      "loss": 0.0264,
+      "step": 217
+    },
+    {
+      "clip_ratio/high_max": 0.0023662418461753987,
+      "clip_ratio/high_mean": 0.0010013960363721708,
+      "clip_ratio/low_mean": 0.0010428183886688203,
+      "clip_ratio/low_min": 2.7021183996112086e-05,
+      "clip_ratio/region_mean": 0.0020442144668777473,
+      "epoch": 2.2425655976676384,
+      "grad_norm": 0.12043320387601852,
+      "learning_rate": 1e-06,
+      "loss": 0.0043,
+      "step": 218
+    },
+    {
+      "clip_ratio/high_max": 0.002208399157098029,
+      "clip_ratio/high_mean": 0.0009531907053315081,
+      "clip_ratio/low_mean": 0.0010326984884159174,
+      "clip_ratio/low_min": 4.3137135435245e-05,
+      "clip_ratio/region_mean": 0.0019858891901094466,
+      "epoch": 2.2518950437317784,
+      "grad_norm": 0.11767147481441498,
+      "learning_rate": 1e-06,
+      "loss": -0.0127,
+      "step": 219
+    },
+    {
+      "clip_ratio/high_max": 0.0023339143881457858,
+      "clip_ratio/high_mean": 0.0009989120135287521,
+      "clip_ratio/low_mean": 0.0012673940327658784,
+      "clip_ratio/low_min": 8.345290279976325e-05,
+      "clip_ratio/region_mean": 0.002266306029923726,
+      "epoch": 2.2612244897959184,
+      "grad_norm": 0.13508707284927368,
+      "learning_rate": 1e-06,
+      "loss": -0.0055,
+      "step": 220
+    },
+    {
+      "clip_ratio/high_max": 0.0024647469690535218,
+      "clip_ratio/high_mean": 0.0010625124850776047,
+      "clip_ratio/low_mean": 0.0010962694032059517,
+      "clip_ratio/low_min": 0.0001967881407836103,
+      "clip_ratio/region_mean": 0.002158781891921535,
+      "epoch": 2.2705539358600584,
+      "grad_norm": 0.13242405652999878,
+      "learning_rate": 1e-06,
+      "loss": 0.0069,
+      "step": 221
+    },
+    {
+      "clip_ratio/high_max": 0.002287006995175034,
+      "clip_ratio/high_mean": 0.0010311212790838908,
+      "clip_ratio/low_mean": 0.0010112026575370692,
+      "clip_ratio/low_min": 7.01167264196556e-05,
+      "clip_ratio/region_mean": 0.002042323954810854,
+      "epoch": 2.2798833819241984,
+      "grad_norm": 0.11984199285507202,
+      "learning_rate": 1e-06,
+      "loss": -0.0283,
+      "step": 222
+    },
+    {
+      "clip_ratio/high_max": 0.002268641379487235,
+      "clip_ratio/high_mean": 0.0009501121567154769,
+      "clip_ratio/low_mean": 0.0011279360769549385,
+      "clip_ratio/low_min": 9.394236622028984e-05,
+      "clip_ratio/region_mean": 0.0020780483100679703,
+      "epoch": 2.2892128279883384,
+      "grad_norm": 0.11281422525644302,
+      "learning_rate": 1e-06,
+      "loss": -0.0141,
+      "step": 223
+    },
+    {
+      "clip_ratio/high_max": 0.0022647486330242828,
+      "clip_ratio/high_mean": 0.0010632389967213385,
+      "clip_ratio/low_mean": 0.0011535417233972112,
+      "clip_ratio/low_min": 5.966096432530321e-05,
+      "clip_ratio/region_mean": 0.0022167806964716874,
+      "epoch": 2.298542274052478,
+      "grad_norm": 0.12871526181697845,
+      "learning_rate": 1e-06,
+      "loss": 0.0013,
+      "step": 224
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.019566127232142905,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4087.0,
+      "completions/mean_length": 630.736328125,
+      "completions/mean_terminated_length": 561.5814208984375,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 2.307871720116618,
+      "grad_norm": 0.13329946994781494,
+      "learning_rate": 1e-06,
+      "loss": -0.0225,
+      "num_tokens": 148803354.0,
+      "reward": 0.5794503688812256,
+      "reward_std": 0.19710879027843475,
+      "rewards/simpleverify_reward/mean": 0.5794503092765808,
+      "rewards/simpleverify_reward/std": 0.49365589022636414,
+      "step": 225
+    },
+    {
+      "clip_ratio/high_max": 0.001931429149408359,
+      "clip_ratio/high_mean": 0.0008306541731144534,
+      "clip_ratio/low_mean": 0.0006265867032198003,
+      "clip_ratio/low_min": 3.5744412343774457e-05,
+      "clip_ratio/region_mean": 0.0014572408763342537,
+      "epoch": 2.317201166180758,
+      "grad_norm": 0.12135004252195358,
+      "learning_rate": 1e-06,
+      "loss": 0.0056,
+      "step": 226
+    },
+    {
+      "clip_ratio/high_max": 0.0016768822570156772,
+      "clip_ratio/high_mean": 0.0008354935489478521,
+      "clip_ratio/low_mean": 0.0005116177226227592,
+      "clip_ratio/low_min": 1.498441633884795e-05,
+      "clip_ratio/region_mean": 0.0013471112761180848,
+      "epoch": 2.326530612244898,
+      "grad_norm": 0.11290162801742554,
+      "learning_rate": 1e-06,
+      "loss": 0.0062,
+      "step": 227
+    },
+    {
+      "clip_ratio/high_max": 0.0017869251059892122,
+      "clip_ratio/high_mean": 0.0008068411007116083,
+      "clip_ratio/low_mean": 0.0006611987846554257,
+      "clip_ratio/low_min": 4.146084211242851e-05,
+      "clip_ratio/region_mean": 0.0014680398926429916,
+      "epoch": 2.335860058309038,
+      "grad_norm": 0.13063408434391022,
+      "learning_rate": 1e-06,
+      "loss": 0.0431,
+      "step": 228
+    },
+    {
+      "clip_ratio/high_max": 0.0022113891027402133,
+      "clip_ratio/high_mean": 0.0008729533656151034,
+      "clip_ratio/low_mean": 0.0006115404485171894,
+      "clip_ratio/low_min": 2.9467566491803154e-05,
+      "clip_ratio/region_mean": 0.0014844938414171338,
+      "epoch": 2.345189504373178,
+      "grad_norm": 0.13261906802654266,
+      "learning_rate": 1e-06,
+      "loss": -0.0004,
+      "step": 229
+    },
+    {
+      "clip_ratio/high_max": 0.0024802484767860733,
+      "clip_ratio/high_mean": 0.0010233752946078312,
+      "clip_ratio/low_mean": 0.0005022480663683382,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015256233964464627,
+      "epoch": 2.354518950437318,
+      "grad_norm": 0.11898131668567657,
+      "learning_rate": 1e-06,
+      "loss": -0.0371,
+      "step": 230
+    },
+    {
+      "clip_ratio/high_max": 0.0023322810884565115,
+      "clip_ratio/high_mean": 0.0008943115790316369,
+      "clip_ratio/low_mean": 0.0007600811477459501,
+      "clip_ratio/low_min": 3.761400239454815e-05,
+      "clip_ratio/region_mean": 0.0016543927558814175,
+      "epoch": 2.363848396501458,
+      "grad_norm": 0.1239127516746521,
+      "learning_rate": 1e-06,
+      "loss": 0.0069,
+      "step": 231
+    },
+    {
+      "clip_ratio/high_max": 0.002162837205105461,
+      "clip_ratio/high_mean": 0.0008315077648148872,
+      "clip_ratio/low_mean": 0.0007176635408541188,
+      "clip_ratio/low_min": 3.620522147684824e-05,
+      "clip_ratio/region_mean": 0.0015491713274968788,
+      "epoch": 2.373177842565598,
+      "grad_norm": 0.1227312833070755,
+      "learning_rate": 1e-06,
+      "loss": 0.02,
+      "step": 232
+    },
+    {
+      "clip_ratio/high_max": 0.00214016082463786,
+      "clip_ratio/high_mean": 0.000906252489585313,
+      "clip_ratio/low_mean": 0.000733572494937107,
+      "clip_ratio/low_min": 3.462603854131885e-05,
+      "clip_ratio/region_mean": 0.0016398249645135365,
+      "epoch": 2.3825072886297374,
+      "grad_norm": 0.11811984330415726,
+      "learning_rate": 1e-06,
+      "loss": 0.0029,
+      "step": 233
+    },
+    {
+      "clip_ratio/high_max": 0.0020093668572371826,
+      "clip_ratio/high_mean": 0.0008168029089574702,
+      "clip_ratio/low_mean": 0.0008965912747953553,
+      "clip_ratio/low_min": 7.428751814586576e-05,
+      "clip_ratio/region_mean": 0.0017133941582869738,
+      "epoch": 2.3918367346938774,
+      "grad_norm": 0.1290924847126007,
+      "learning_rate": 1e-06,
+      "loss": 0.039,
+      "step": 234
+    },
+    {
+      "clip_ratio/high_max": 0.001914288724947255,
+      "clip_ratio/high_mean": 0.0008547640100005083,
+      "clip_ratio/low_mean": 0.0008121297823890927,
+      "clip_ratio/low_min": 4.846118827117607e-05,
+      "clip_ratio/region_mean": 0.0016668937823851593,
+      "epoch": 2.4011661807580174,
+      "grad_norm": 0.12953390181064606,
+      "learning_rate": 1e-06,
+      "loss": -0.0013,
+      "step": 235
+    },
+    {
+      "clip_ratio/high_max": 0.0024019067714107223,
+      "clip_ratio/high_mean": 0.0008915145008359104,
+      "clip_ratio/low_mean": 0.0008578787019359879,
+      "clip_ratio/low_min": 9.866466643870808e-05,
+      "clip_ratio/region_mean": 0.0017493932027718984,
+      "epoch": 2.4104956268221573,
+      "grad_norm": 0.1239464282989502,
+      "learning_rate": 1e-06,
+      "loss": 0.0261,
+      "step": 236
+    },
+    {
+      "clip_ratio/high_max": 0.00219047720747767,
+      "clip_ratio/high_mean": 0.0009286063177569304,
+      "clip_ratio/low_mean": 0.00079981864655565,
+      "clip_ratio/low_min": 4.524950600170996e-05,
+      "clip_ratio/region_mean": 0.0017284249770455062,
+      "epoch": 2.4198250728862973,
+      "grad_norm": 0.12453047186136246,
+      "learning_rate": 1e-06,
+      "loss": 0.018,
+      "step": 237
+    },
+    {
+      "clip_ratio/high_max": 0.002245578245492652,
+      "clip_ratio/high_mean": 0.0010236589005216956,
+      "clip_ratio/low_mean": 0.0010214083795290207,
+      "clip_ratio/low_min": 7.412554805341642e-05,
+      "clip_ratio/region_mean": 0.0020450672527658753,
+      "epoch": 2.4291545189504373,
+      "grad_norm": 0.12429088354110718,
+      "learning_rate": 1e-06,
+      "loss": 0.0183,
+      "step": 238
+    },
+    {
+      "clip_ratio/high_max": 0.0021250942118058447,
+      "clip_ratio/high_mean": 0.0009698486210254487,
+      "clip_ratio/low_mean": 0.0009306999745604116,
+      "clip_ratio/low_min": 5.981852791592246e-05,
+      "clip_ratio/region_mean": 0.0019005486101377755,
+      "epoch": 2.4384839650145773,
+      "grad_norm": 0.12769249081611633,
+      "learning_rate": 1e-06,
+      "loss": 0.0028,
+      "step": 239
+    },
+    {
+      "clip_ratio/high_max": 0.0023410784924635664,
+      "clip_ratio/high_mean": 0.0010047145915450528,
+      "clip_ratio/low_mean": 0.0008066290483839111,
+      "clip_ratio/low_min": 9.084598423214629e-05,
+      "clip_ratio/region_mean": 0.0018113436599378474,
+      "epoch": 2.4478134110787173,
+      "grad_norm": 0.121663399040699,
+      "learning_rate": 1e-06,
+      "loss": -0.0246,
+      "step": 240
+    },
+    {
+      "clip_ratio/high_max": 0.0020401714500621893,
+      "clip_ratio/high_mean": 0.0009104738965106662,
+      "clip_ratio/low_mean": 0.0009578589706507046,
+      "clip_ratio/low_min": 5.7134509916068055e-05,
+      "clip_ratio/region_mean": 0.0018683328744373284,
+      "epoch": 2.4571428571428573,
+      "grad_norm": 0.11453356593847275,
+      "learning_rate": 1e-06,
+      "loss": 0.0201,
+      "step": 241
+    },
+    {
+      "clip_ratio/high_max": 0.00280039736389881,
+      "clip_ratio/high_mean": 0.0011022610815416556,
+      "clip_ratio/low_mean": 0.0010881947564485017,
+      "clip_ratio/low_min": 0.00011510461445141118,
+      "clip_ratio/region_mean": 0.002190455881645903,
+      "epoch": 2.466472303206997,
+      "grad_norm": 0.11957170069217682,
+      "learning_rate": 1e-06,
+      "loss": -0.0131,
+      "step": 242
+    },
+    {
+      "clip_ratio/high_max": 0.002436354843666777,
+      "clip_ratio/high_mean": 0.0010595270650810562,
+      "clip_ratio/low_mean": 0.0010193988691753475,
+      "clip_ratio/low_min": 0.0001645911943342071,
+      "clip_ratio/region_mean": 0.002078926030662842,
+      "epoch": 2.4758017492711373,
+      "grad_norm": 0.12470949441194534,
+      "learning_rate": 1e-06,
+      "loss": -0.0172,
+      "step": 243
+    },
+    {
+      "clip_ratio/high_max": 0.001885615049104672,
+      "clip_ratio/high_mean": 0.0009377739079354797,
+      "clip_ratio/low_mean": 0.0009602939389878884,
+      "clip_ratio/low_min": 8.107810572255403e-05,
+      "clip_ratio/region_mean": 0.0018980678432853892,
+      "epoch": 2.485131195335277,
+      "grad_norm": 0.12089991569519043,
+      "learning_rate": 1e-06,
+      "loss": -0.0086,
+      "step": 244
+    },
+    {
+      "clip_ratio/high_max": 0.002278977062815102,
+      "clip_ratio/high_mean": 0.0009759604818100343,
+      "clip_ratio/low_mean": 0.0009597634143574396,
+      "clip_ratio/low_min": 1.600102405063808e-05,
+      "clip_ratio/region_mean": 0.0019357238852535374,
+      "epoch": 2.494460641399417,
+      "grad_norm": 0.1121646985411644,
+      "learning_rate": 1e-06,
+      "loss": -0.0093,
+      "step": 245
+    },
+    {
+      "clip_ratio/high_max": 0.00190876182023203,
+      "clip_ratio/high_mean": 0.0008337687540915795,
+      "clip_ratio/low_mean": 0.0010464253391546663,
+      "clip_ratio/low_min": 0.00016330648304574424,
+      "clip_ratio/region_mean": 0.001880194067780394,
+      "epoch": 2.503790087463557,
+      "grad_norm": 0.11362891644239426,
+      "learning_rate": 1e-06,
+      "loss": 0.0079,
+      "step": 246
+    },
+    {
+      "clip_ratio/high_max": 0.001980891709536081,
+      "clip_ratio/high_mean": 0.0008288653825729853,
+      "clip_ratio/low_mean": 0.001052274064932135,
+      "clip_ratio/low_min": 0.00010290669888490811,
+      "clip_ratio/region_mean": 0.001881139432953205,
+      "epoch": 2.513119533527697,
+      "grad_norm": 0.1142832487821579,
+      "learning_rate": 1e-06,
+      "loss": 0.0023,
+      "step": 247
+    },
+    {
+      "clip_ratio/high_max": 0.002474515298672486,
+      "clip_ratio/high_mean": 0.0009529462004138622,
+      "clip_ratio/low_mean": 0.0013267577760416316,
+      "clip_ratio/low_min": 0.00015478871137020178,
+      "clip_ratio/region_mean": 0.0022797039346187375,
+      "epoch": 2.522448979591837,
+      "grad_norm": 0.11705332249403,
+      "learning_rate": 1e-06,
+      "loss": 0.0166,
+      "step": 248
+    },
+    {
+      "clip_ratio/high_max": 0.0026798795515787788,
+      "clip_ratio/high_mean": 0.0009887291053018998,
+      "clip_ratio/low_mean": 0.0010259262526233215,
+      "clip_ratio/low_min": 1.391052774124546e-05,
+      "clip_ratio/region_mean": 0.002014655343373306,
+      "epoch": 2.5317784256559768,
+      "grad_norm": 0.12924474477767944,
+      "learning_rate": 1e-06,
+      "loss": 0.0399,
+      "step": 249
+    },
+    {
+      "clip_ratio/high_max": 0.0022953186598897446,
+      "clip_ratio/high_mean": 0.001018163280605222,
+      "clip_ratio/low_mean": 0.0011129522281407844,
+      "clip_ratio/low_min": 2.8344958991510794e-05,
+      "clip_ratio/region_mean": 0.00213111552875489,
+      "epoch": 2.5411078717201168,
+      "grad_norm": 0.10528191179037094,
+      "learning_rate": 1e-06,
+      "loss": 0.0259,
+      "step": 250
+    },
+    {
+      "clip_ratio/high_max": 0.0022079543050494976,
+      "clip_ratio/high_mean": 0.0010806344398588408,
+      "clip_ratio/low_mean": 0.0009955247714970028,
+      "clip_ratio/low_min": 6.985532309045084e-05,
+      "clip_ratio/region_mean": 0.0020761592386406846,
+      "epoch": 2.5504373177842563,
+      "grad_norm": 0.1289607435464859,
+      "learning_rate": 1e-06,
+      "loss": -0.0497,
+      "step": 251
+    },
+    {
+      "clip_ratio/high_max": 0.0028587446286110207,
+      "clip_ratio/high_mean": 0.0011241016181884333,
+      "clip_ratio/low_mean": 0.0009463640653848415,
+      "clip_ratio/low_min": 6.59255401842529e-05,
+      "clip_ratio/region_mean": 0.002070465692668222,
+      "epoch": 2.5597667638483967,
+      "grad_norm": 0.12235015630722046,
+      "learning_rate": 1e-06,
+      "loss": -0.0364,
+      "step": 252
+    },
+    {
+      "clip_ratio/high_max": 0.0025497422539046966,
+      "clip_ratio/high_mean": 0.0010962882515741512,
+      "clip_ratio/low_mean": 0.0012082916218787432,
+      "clip_ratio/low_min": 0.00020721147757285507,
+      "clip_ratio/region_mean": 0.002304579902556725,
+      "epoch": 2.5690962099125363,
+      "grad_norm": 0.13087229430675507,
+      "learning_rate": 1e-06,
+      "loss": -0.0165,
+      "step": 253
+    },
+    {
+      "clip_ratio/high_max": 0.0027529015715117566,
+      "clip_ratio/high_mean": 0.0011465288989711553,
+      "clip_ratio/low_mean": 0.0011475638675619848,
+      "clip_ratio/low_min": 3.068127898586681e-05,
+      "clip_ratio/region_mean": 0.0022940927956369705,
+      "epoch": 2.5784256559766763,
+      "grad_norm": 0.13189953565597534,
+      "learning_rate": 1e-06,
+      "loss": -0.0203,
+      "step": 254
+    },
+    {
+      "clip_ratio/high_max": 0.0021623915672535077,
+      "clip_ratio/high_mean": 0.0009950409239536384,
+      "clip_ratio/low_mean": 0.001243158587385551,
+      "clip_ratio/low_min": 0.0001403538899467094,
+      "clip_ratio/region_mean": 0.002238199580460787,
+      "epoch": 2.5877551020408163,
+      "grad_norm": 0.12326417863368988,
+      "learning_rate": 1e-06,
+      "loss": 0.0205,
+      "step": 255
+    },
+    {
+      "clip_ratio/high_max": 0.002177543705329299,
+      "clip_ratio/high_mean": 0.0009701890376163647,
+      "clip_ratio/low_mean": 0.001118037096603075,
+      "clip_ratio/low_min": 0.0001399228831360233,
+      "clip_ratio/region_mean": 0.0020882261014776304,
+      "epoch": 2.5970845481049563,
+      "grad_norm": 0.1164150983095169,
+      "learning_rate": 1e-06,
+      "loss": 0.0143,
+      "step": 256
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0235072544642857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4093.0,
+      "completions/mean_length": 645.2140502929688,
+      "completions/mean_terminated_length": 562.1427612304688,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 2.6064139941690962,
+      "grad_norm": 0.13600456714630127,
+      "learning_rate": 1e-06,
+      "loss": -0.0001,
+      "num_tokens": 167366571.0,
+      "reward": 0.5756138563156128,
+      "reward_std": 0.19903387129306793,
+      "rewards/simpleverify_reward/mean": 0.5756138563156128,
+      "rewards/simpleverify_reward/std": 0.4942580759525299,
+      "step": 257
+    },
+    {
+      "clip_ratio/high_max": 0.00189734512969153,
+      "clip_ratio/high_mean": 0.0008366046840819763,
+      "clip_ratio/low_mean": 0.0006608965504710795,
+      "clip_ratio/low_min": 1.5144172721193172e-05,
+      "clip_ratio/region_mean": 0.0014975012600189075,
+      "epoch": 2.6157434402332362,
+      "grad_norm": 0.12501174211502075,
+      "learning_rate": 1e-06,
+      "loss": 0.0082,
+      "step": 258
+    },
+    {
+      "clip_ratio/high_max": 0.002123884820321109,
+      "clip_ratio/high_mean": 0.0008968435904534999,
+      "clip_ratio/low_mean": 0.0006447752157328068,
+      "clip_ratio/low_min": 1.1552680007298477e-05,
+      "clip_ratio/region_mean": 0.001541618796181865,
+      "epoch": 2.6250728862973762,
+      "grad_norm": 0.12504319846630096,
+      "learning_rate": 1e-06,
+      "loss": 0.0226,
+      "step": 259
+    },
+    {
+      "clip_ratio/high_max": 0.002071463044558186,
+      "clip_ratio/high_mean": 0.0009713429608382285,
+      "clip_ratio/low_mean": 0.0006334486652121996,
+      "clip_ratio/low_min": 4.235206506564282e-05,
+      "clip_ratio/region_mean": 0.0016047916797106154,
+      "epoch": 2.6344023323615158,
+      "grad_norm": 0.11961305141448975,
+      "learning_rate": 1e-06,
+      "loss": -0.0031,
+      "step": 260
+    },
+    {
+      "clip_ratio/high_max": 0.002227572003903333,
+      "clip_ratio/high_mean": 0.0009052047134900931,
+      "clip_ratio/low_mean": 0.0006609805350308307,
+      "clip_ratio/low_min": 7.137222746678162e-05,
+      "clip_ratio/region_mean": 0.0015661852557968814,
+      "epoch": 2.643731778425656,
+      "grad_norm": 0.12769760191440582,
+      "learning_rate": 1e-06,
+      "loss": 0.0215,
+      "step": 261
+    },
+    {
+      "clip_ratio/high_max": 0.0023088831221684813,
+      "clip_ratio/high_mean": 0.0009140516522165854,
+      "clip_ratio/low_mean": 0.0007019542936177459,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001616005931282416,
+      "epoch": 2.6530612244897958,
+      "grad_norm": 0.1286620795726776,
+      "learning_rate": 1e-06,
+      "loss": -0.022,
+      "step": 262
+    },
+    {
+      "clip_ratio/high_max": 0.0018435072524880525,
+      "clip_ratio/high_mean": 0.0008288394874398364,
+      "clip_ratio/low_mean": 0.0006485298545158003,
+      "clip_ratio/low_min": 2.039151695498731e-05,
+      "clip_ratio/region_mean": 0.0014773693328606896,
+      "epoch": 2.6623906705539357,
+      "grad_norm": 0.12441161274909973,
+      "learning_rate": 1e-06,
+      "loss": -0.008,
+      "step": 263
+    },
+    {
+      "clip_ratio/high_max": 0.0023731261062494013,
+      "clip_ratio/high_mean": 0.0009713482832012232,
+      "clip_ratio/low_mean": 0.0007894066784501774,
+      "clip_ratio/low_min": 1.1546277164597996e-05,
+      "clip_ratio/region_mean": 0.001760754981660284,
+      "epoch": 2.6717201166180757,
+      "grad_norm": 0.1379632204771042,
+      "learning_rate": 1e-06,
+      "loss": 0.0174,
+      "step": 264
+    },
+    {
+      "clip_ratio/high_max": 0.0024549766967538744,
+      "clip_ratio/high_mean": 0.0009812457210500725,
+      "clip_ratio/low_mean": 0.0007618412128067575,
+      "clip_ratio/low_min": 3.2005257708078716e-05,
+      "clip_ratio/region_mean": 0.0017430869120289572,
+      "epoch": 2.6810495626822157,
+      "grad_norm": 0.12083766609430313,
+      "learning_rate": 1e-06,
+      "loss": -0.0256,
+      "step": 265
+    },
+    {
+      "clip_ratio/high_max": 0.0020812658258364536,
+      "clip_ratio/high_mean": 0.0009935599227901548,
+      "clip_ratio/low_mean": 0.0007724324095761403,
+      "clip_ratio/low_min": 3.0486315154121257e-05,
+      "clip_ratio/region_mean": 0.0017659923323662952,
+      "epoch": 2.6903790087463557,
+      "grad_norm": 0.12071080505847931,
+      "learning_rate": 1e-06,
+      "loss": -0.0133,
+      "step": 266
+    },
+    {
+      "clip_ratio/high_max": 0.002290940439706901,
+      "clip_ratio/high_mean": 0.0008696704117028276,
+      "clip_ratio/low_mean": 0.0008006323096196866,
+      "clip_ratio/low_min": 3.3404811802029144e-05,
+      "clip_ratio/region_mean": 0.0016703027649782598,
+      "epoch": 2.6997084548104957,
+      "grad_norm": 0.12483668327331543,
+      "learning_rate": 1e-06,
+      "loss": 0.0078,
+      "step": 267
+    },
+    {
+      "clip_ratio/high_max": 0.0021949295878584962,
+      "clip_ratio/high_mean": 0.0010906782563324668,
+      "clip_ratio/low_mean": 0.0006446836814575363,
+      "clip_ratio/low_min": 2.8325403036433272e-05,
+      "clip_ratio/region_mean": 0.0017353619332425296,
+      "epoch": 2.7090379008746357,
+      "grad_norm": 0.11658875644207001,
+      "learning_rate": 1e-06,
+      "loss": -0.0868,
+      "step": 268
+    },
+    {
+      "clip_ratio/high_max": 0.0025025291033671238,
+      "clip_ratio/high_mean": 0.0010405593529867474,
+      "clip_ratio/low_mean": 0.0009900394070427865,
+      "clip_ratio/low_min": 9.010043140733615e-05,
+      "clip_ratio/region_mean": 0.0020305987345636822,
+      "epoch": 2.7183673469387752,
+      "grad_norm": 0.1365434229373932,
+      "learning_rate": 1e-06,
+      "loss": -0.0118,
+      "step": 269
+    },
+    {
+      "clip_ratio/high_max": 0.0020535689436655957,
+      "clip_ratio/high_mean": 0.0007929516432341188,
+      "clip_ratio/low_mean": 0.0009509364226687467,
+      "clip_ratio/low_min": 0.00014552040920534637,
+      "clip_ratio/region_mean": 0.0017438880313420668,
+      "epoch": 2.7276967930029157,
+      "grad_norm": 0.11683172732591629,
+      "learning_rate": 1e-06,
+      "loss": 0.0009,
+      "step": 270
+    },
+    {
+      "clip_ratio/high_max": 0.0019331066941958852,
+      "clip_ratio/high_mean": 0.000874589912200463,
+      "clip_ratio/low_mean": 0.0009247772431990597,
+      "clip_ratio/low_min": 0.00011506138616823591,
+      "clip_ratio/region_mean": 0.0017993671935983002,
+      "epoch": 2.7370262390670552,
+      "grad_norm": 0.11880014836788177,
+      "learning_rate": 1e-06,
+      "loss": 0.0068,
+      "step": 271
+    },
+    {
+      "clip_ratio/high_max": 0.0023409619607264176,
+      "clip_ratio/high_mean": 0.0009553683612466557,
+      "clip_ratio/low_mean": 0.0010207238974544453,
+      "clip_ratio/low_min": 5.3695403039455414e-05,
+      "clip_ratio/region_mean": 0.001976092222321313,
+      "epoch": 2.746355685131195,
+      "grad_norm": 1.8470609188079834,
+      "learning_rate": 1e-06,
+      "loss": 0.0152,
+      "step": 272
+    },
+    {
+      "clip_ratio/high_max": 0.002387445463682525,
+      "clip_ratio/high_mean": 0.0009731576956255594,
+      "clip_ratio/low_mean": 0.0010243844062642893,
+      "clip_ratio/low_min": 8.229292325268034e-05,
+      "clip_ratio/region_mean": 0.0019975421455455944,
+      "epoch": 2.755685131195335,
+      "grad_norm": 0.11688001453876495,
+      "learning_rate": 1e-06,
+      "loss": 0.015,
+      "step": 273
+    },
+    {
+      "clip_ratio/high_max": 0.0024668668847880326,
+      "clip_ratio/high_mean": 0.0010067965176858706,
+      "clip_ratio/low_mean": 0.001008177121548215,
+      "clip_ratio/low_min": 0.00015318513851525495,
+      "clip_ratio/region_mean": 0.002014973622863181,
+      "epoch": 2.765014577259475,
+      "grad_norm": 0.12078815698623657,
+      "learning_rate": 1e-06,
+      "loss": 0.0151,
+      "step": 274
+    },
+    {
+      "clip_ratio/high_max": 0.002466472207743209,
+      "clip_ratio/high_mean": 0.0010551300292718224,
+      "clip_ratio/low_mean": 0.0010188171290792525,
+      "clip_ratio/low_min": 6.0372036386979744e-05,
+      "clip_ratio/region_mean": 0.002073947194730863,
+      "epoch": 2.774344023323615,
+      "grad_norm": 0.1232159286737442,
+      "learning_rate": 1e-06,
+      "loss": 0.0004,
+      "step": 275
+    },
+    {
+      "clip_ratio/high_max": 0.00244445198768517,
+      "clip_ratio/high_mean": 0.00110145202052081,
+      "clip_ratio/low_mean": 0.0009152304264716804,
+      "clip_ratio/low_min": 2.8493275749497116e-05,
+      "clip_ratio/region_mean": 0.0020166824615444057,
+      "epoch": 2.783673469387755,
+      "grad_norm": 0.12969809770584106,
+      "learning_rate": 1e-06,
+      "loss": -0.0251,
+      "step": 276
+    },
+    {
+      "clip_ratio/high_max": 0.00231015557073988,
+      "clip_ratio/high_mean": 0.0009708871039038058,
+      "clip_ratio/low_mean": 0.0010659317322279094,
+      "clip_ratio/low_min": 2.89754279947374e-05,
+      "clip_ratio/region_mean": 0.0020368187615531497,
+      "epoch": 2.793002915451895,
+      "grad_norm": 0.1141471341252327,
+      "learning_rate": 1e-06,
+      "loss": -0.0276,
+      "step": 277
+    },
+    {
+      "clip_ratio/high_max": 0.0024809273818391375,
+      "clip_ratio/high_mean": 0.0010630047945596743,
+      "clip_ratio/low_mean": 0.001058063775417395,
+      "clip_ratio/low_min": 8.341390639543533e-05,
+      "clip_ratio/region_mean": 0.002121068573615048,
+      "epoch": 2.8023323615160347,
+      "grad_norm": 0.12116222828626633,
+      "learning_rate": 1e-06,
+      "loss": -0.0181,
+      "step": 278
+    },
+    {
+      "clip_ratio/high_max": 0.002853109879652038,
+      "clip_ratio/high_mean": 0.001015771975289681,
+      "clip_ratio/low_mean": 0.001141551463661017,
+      "clip_ratio/low_min": 0.00010783191646623891,
+      "clip_ratio/region_mean": 0.002157323411665857,
+      "epoch": 2.811661807580175,
+      "grad_norm": 0.12891864776611328,
+      "learning_rate": 1e-06,
+      "loss": 0.0098,
+      "step": 279
+    },
+    {
+      "clip_ratio/high_max": 0.0024491413932992145,
+      "clip_ratio/high_mean": 0.0009487917632213794,
+      "clip_ratio/low_mean": 0.001186088913527783,
+      "clip_ratio/low_min": 0.00017845502134150593,
+      "clip_ratio/region_mean": 0.002134880742232781,
+      "epoch": 2.8209912536443147,
+      "grad_norm": 0.12250273674726486,
+      "learning_rate": 1e-06,
+      "loss": 0.0034,
+      "step": 280
+    },
+    {
+      "clip_ratio/high_max": 0.00224468921078369,
+      "clip_ratio/high_mean": 0.0009757881198311225,
+      "clip_ratio/low_mean": 0.0012388760733301751,
+      "clip_ratio/low_min": 0.0001514393552497495,
+      "clip_ratio/region_mean": 0.002214664244093001,
+      "epoch": 2.8303206997084547,
+      "grad_norm": 0.12632285058498383,
+      "learning_rate": 1e-06,
+      "loss": -0.0003,
+      "step": 281
+    },
+    {
+      "clip_ratio/high_max": 0.0022118565830169246,
+      "clip_ratio/high_mean": 0.0009181990608340129,
+      "clip_ratio/low_mean": 0.0011776583960454445,
+      "clip_ratio/low_min": 0.00015783298840688076,
+      "clip_ratio/region_mean": 0.0020958573804819025,
+      "epoch": 2.8396501457725947,
+      "grad_norm": 0.11565697938203812,
+      "learning_rate": 1e-06,
+      "loss": 0.0332,
+      "step": 282
+    },
+    {
+      "clip_ratio/high_max": 0.0024144740746123716,
+      "clip_ratio/high_mean": 0.0010288224366377108,
+      "clip_ratio/low_mean": 0.0012387945535010658,
+      "clip_ratio/low_min": 0.0001487179906689562,
+      "clip_ratio/region_mean": 0.0022676170265185647,
+      "epoch": 2.8489795918367347,
+      "grad_norm": 0.11546675115823746,
+      "learning_rate": 1e-06,
+      "loss": 0.0167,
+      "step": 283
+    },
+    {
+      "clip_ratio/high_max": 0.0025626233327784576,
+      "clip_ratio/high_mean": 0.0010337062776670791,
+      "clip_ratio/low_mean": 0.0012107248803658877,
+      "clip_ratio/low_min": 0.00011483960952318739,
+      "clip_ratio/region_mean": 0.002244431132567115,
+      "epoch": 2.8583090379008746,
+      "grad_norm": 0.12259412556886673,
+      "learning_rate": 1e-06,
+      "loss": 0.02,
+      "step": 284
+    },
+    {
+      "clip_ratio/high_max": 0.0023987821296032052,
+      "clip_ratio/high_mean": 0.0010659938106982736,
+      "clip_ratio/low_mean": 0.00112719381286297,
+      "clip_ratio/low_min": 5.6721270084381104e-05,
+      "clip_ratio/region_mean": 0.0021931875817244872,
+      "epoch": 2.8676384839650146,
+      "grad_norm": 0.14035850763320923,
+      "learning_rate": 1e-06,
+      "loss": -0.0136,
+      "step": 285
+    },
+    {
+      "clip_ratio/high_max": 0.002533859576942632,
+      "clip_ratio/high_mean": 0.0009809613566176267,
+      "clip_ratio/low_mean": 0.0012055512779625133,
+      "clip_ratio/low_min": 0.00013177785331208725,
+      "clip_ratio/region_mean": 0.002186512661864981,
+      "epoch": 2.8769679300291546,
+      "grad_norm": 0.1203429102897644,
+      "learning_rate": 1e-06,
+      "loss": 0.0101,
+      "step": 286
+    },
+    {
+      "clip_ratio/high_max": 0.0023078008016454987,
+      "clip_ratio/high_mean": 0.0009636809063522378,
+      "clip_ratio/low_mean": 0.0012416589415806811,
+      "clip_ratio/low_min": 0.00014422896219912218,
+      "clip_ratio/region_mean": 0.0022053398060961626,
+      "epoch": 2.8862973760932946,
+      "grad_norm": 0.12423661351203918,
+      "learning_rate": 1e-06,
+      "loss": 0.0359,
+      "step": 287
+    },
+    {
+      "clip_ratio/high_max": 0.0025863796399789862,
+      "clip_ratio/high_mean": 0.0010663768753147451,
+      "clip_ratio/low_mean": 0.0009776043771125842,
+      "clip_ratio/low_min": 1.75168170244433e-05,
+      "clip_ratio/region_mean": 0.0020439811996766366,
+      "epoch": 2.8956268221574346,
+      "grad_norm": 0.11673915386199951,
+      "learning_rate": 1e-06,
+      "loss": -0.021,
+      "step": 288
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.025913783482142905,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4047.0,
+      "completions/mean_length": 661.5585327148438,
+      "completions/mean_terminated_length": 570.1915283203125,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 3.00932944606414,
+      "grad_norm": 0.12585154175758362,
+      "learning_rate": 1e-06,
+      "loss": -0.0473,
+      "num_tokens": 186191657.0,
+      "reward": 0.5730329751968384,
+      "reward_std": 0.19180303812026978,
+      "rewards/simpleverify_reward/mean": 0.5730329155921936,
+      "rewards/simpleverify_reward/std": 0.4946460723876953,
+      "step": 289
+    },
+    {
+      "clip_ratio/high_max": 0.002137782401405275,
+      "clip_ratio/high_mean": 0.0008485483122058213,
+      "clip_ratio/low_mean": 0.0006079304330341984,
+      "clip_ratio/low_min": 4.293019082979299e-05,
+      "clip_ratio/region_mean": 0.0014564787707058713,
+      "epoch": 3.01865889212828,
+      "grad_norm": 0.1152670606970787,
+      "learning_rate": 1e-06,
+      "loss": 0.0319,
+      "step": 290
+    },
+    {
+      "clip_ratio/high_max": 0.0022789640570408665,
+      "clip_ratio/high_mean": 0.0009204337584378663,
+      "clip_ratio/low_mean": 0.0006990835590841016,
+      "clip_ratio/low_min": 4.5441803194989916e-05,
+      "clip_ratio/region_mean": 0.0016195172938751057,
+      "epoch": 3.02798833819242,
+      "grad_norm": 0.1447540521621704,
+      "learning_rate": 1e-06,
+      "loss": 0.0363,
+      "step": 291
+    },
+    {
+      "clip_ratio/high_max": 0.0018935483803943498,
+      "clip_ratio/high_mean": 0.0008393884145334596,
+      "clip_ratio/low_mean": 0.0005619536314043216,
+      "clip_ratio/low_min": 1.2286220226087607e-05,
+      "clip_ratio/region_mean": 0.001401342022290919,
+      "epoch": 3.03731778425656,
+      "grad_norm": 0.1221446692943573,
+      "learning_rate": 1e-06,
+      "loss": -0.0261,
+      "step": 292
+    },
+    {
+      "clip_ratio/high_max": 0.0018411432683933526,
+      "clip_ratio/high_mean": 0.0008469480471831048,
+      "clip_ratio/low_mean": 0.0006023529585945653,
+      "clip_ratio/low_min": 2.819645669660531e-05,
+      "clip_ratio/region_mean": 0.0014493010166916065,
+      "epoch": 3.0466472303206995,
+      "grad_norm": 0.12960156798362732,
+      "learning_rate": 1e-06,
+      "loss": -0.0301,
+      "step": 293
+    },
+    {
+      "clip_ratio/high_max": 0.0019150481602991931,
+      "clip_ratio/high_mean": 0.0007621312506671529,
+      "clip_ratio/low_mean": 0.000683142608977505,
+      "clip_ratio/low_min": 6.466388458647998e-05,
+      "clip_ratio/region_mean": 0.0014452738832915202,
+      "epoch": 3.0559766763848395,
+      "grad_norm": 0.12239605188369751,
+      "learning_rate": 1e-06,
+      "loss": 0.0301,
+      "step": 294
+    },
+    {
+      "clip_ratio/high_max": 0.002183535609219689,
+      "clip_ratio/high_mean": 0.0009533421907690354,
+      "clip_ratio/low_mean": 0.0007072594580677105,
+      "clip_ratio/low_min": 4.035053825646173e-05,
+      "clip_ratio/region_mean": 0.0016606016215519048,
+      "epoch": 3.0653061224489795,
+      "grad_norm": 0.13197290897369385,
+      "learning_rate": 1e-06,
+      "loss": -0.0128,
+      "step": 295
+    },
+    {
+      "clip_ratio/high_max": 0.001975508115720004,
+      "clip_ratio/high_mean": 0.0008964744829427218,
+      "clip_ratio/low_mean": 0.0007166163568399497,
+      "clip_ratio/low_min": 5.755432994192233e-05,
+      "clip_ratio/region_mean": 0.001613090844330145,
+      "epoch": 3.0746355685131195,
+      "grad_norm": 0.1390487551689148,
+      "learning_rate": 1e-06,
+      "loss": -0.017,
+      "step": 296
+    },
+    {
+      "clip_ratio/high_max": 0.0018032961306744255,
+      "clip_ratio/high_mean": 0.000826057068479713,
+      "clip_ratio/low_mean": 0.0006996557622187538,
+      "clip_ratio/low_min": 5.9933614465990104e-05,
+      "clip_ratio/region_mean": 0.0015257128106895834,
+      "epoch": 3.0839650145772595,
+      "grad_norm": 0.12430579960346222,
+      "learning_rate": 1e-06,
+      "loss": 0.0099,
+      "step": 297
+    },
+    {
+      "clip_ratio/high_max": 0.002024217421421781,
+      "clip_ratio/high_mean": 0.0008760691252973629,
+      "clip_ratio/low_mean": 0.0008207828032027464,
+      "clip_ratio/low_min": 9.32840766836307e-05,
+      "clip_ratio/region_mean": 0.00169685192668112,
+      "epoch": 3.0932944606413995,
+      "grad_norm": 0.12031778693199158,
+      "learning_rate": 1e-06,
+      "loss": 0.0288,
+      "step": 298
+    },
+    {
+      "clip_ratio/high_max": 0.0019508445147948805,
+      "clip_ratio/high_mean": 0.0008287017099064542,
+      "clip_ratio/low_mean": 0.0008302461119455984,
+      "clip_ratio/low_min": 0.00011111687399534276,
+      "clip_ratio/region_mean": 0.0016589478109381162,
+      "epoch": 3.1026239067055394,
+      "grad_norm": 0.12004540860652924,
+      "learning_rate": 1e-06,
+      "loss": 0.0319,
+      "step": 299
+    },
+    {
+      "clip_ratio/high_max": 0.0020195316392346285,
+      "clip_ratio/high_mean": 0.0008908537165552843,
+      "clip_ratio/low_mean": 0.0008761511599004734,
+      "clip_ratio/low_min": 7.549395741079934e-05,
+      "clip_ratio/region_mean": 0.0017670048837317154,
+      "epoch": 3.1119533527696794,
+      "grad_norm": 0.12134338170289993,
+      "learning_rate": 1e-06,
+      "loss": 0.0148,
+      "step": 300
+    },
+    {
+      "clip_ratio/high_max": 0.0022909149483894,
+      "clip_ratio/high_mean": 0.0010250048399029765,
+      "clip_ratio/low_mean": 0.0008124986998154782,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018375034997006878,
+      "epoch": 3.1212827988338194,
+      "grad_norm": 0.11783631891012192,
+      "learning_rate": 1e-06,
+      "loss": -0.0093,
+      "step": 301
+    },
+    {
+      "clip_ratio/high_max": 0.0020253909897292033,
+      "clip_ratio/high_mean": 0.0008927894159569405,
+      "clip_ratio/low_mean": 0.000938629043957917,
+      "clip_ratio/low_min": 0.0001440833921151352,
+      "clip_ratio/region_mean": 0.0018314184417249635,
+      "epoch": 3.130612244897959,
+      "grad_norm": 0.13193082809448242,
+      "learning_rate": 1e-06,
+      "loss": 0.03,
+      "step": 302
+    },
+    {
+      "clip_ratio/high_max": 0.0021273178717819974,
+      "clip_ratio/high_mean": 0.0009635814203647897,
+      "clip_ratio/low_mean": 0.000997285282210214,
+      "clip_ratio/low_min": 0.00011542265019670594,
+      "clip_ratio/region_mean": 0.0019608666843851097,
+      "epoch": 3.139941690962099,
+      "grad_norm": 0.11437549442052841,
+      "learning_rate": 1e-06,
+      "loss": 0.0228,
+      "step": 303
+    },
+    {
+      "clip_ratio/high_max": 0.0022696309024468064,
+      "clip_ratio/high_mean": 0.0008982518356788205,
+      "clip_ratio/low_mean": 0.0009465095863561146,
+      "clip_ratio/low_min": 8.449858705716906e-05,
+      "clip_ratio/region_mean": 0.001844761412939988,
+      "epoch": 3.149271137026239,
+      "grad_norm": 0.12392532080411911,
+      "learning_rate": 1e-06,
+      "loss": 0.0046,
+      "step": 304
+    },
+    {
+      "clip_ratio/high_max": 0.002378022516495548,
+      "clip_ratio/high_mean": 0.0011258928370807553,
+      "clip_ratio/low_mean": 0.0008753391516620468,
+      "clip_ratio/low_min": 4.3759836444223765e-05,
+      "clip_ratio/region_mean": 0.0020012320201203693,
+      "epoch": 3.158600583090379,
+      "grad_norm": 0.13241757452487946,
+      "learning_rate": 1e-06,
+      "loss": -0.0413,
+      "step": 305
+    },
+    {
+      "clip_ratio/high_max": 0.0021378254059527535,
+      "clip_ratio/high_mean": 0.0008881575868144864,
+      "clip_ratio/low_mean": 0.0010044291338999756,
+      "clip_ratio/low_min": 4.371010982140433e-05,
+      "clip_ratio/region_mean": 0.0018925867625512183,
+      "epoch": 3.167930029154519,
+      "grad_norm": 0.12934567034244537,
+      "learning_rate": 1e-06,
+      "loss": -0.0051,
+      "step": 306
+    },
+    {
+      "clip_ratio/high_max": 0.002058826637949096,
+      "clip_ratio/high_mean": 0.0009078570565179689,
+      "clip_ratio/low_mean": 0.001123340851336252,
+      "clip_ratio/low_min": 9.600896737538278e-05,
+      "clip_ratio/region_mean": 0.0020311979242251255,
+      "epoch": 3.177259475218659,
+      "grad_norm": 0.11368083208799362,
+      "learning_rate": 1e-06,
+      "loss": -0.0034,
+      "step": 307
+    },
+    {
+      "clip_ratio/high_max": 0.002240779111161828,
+      "clip_ratio/high_mean": 0.0008956037800089689,
+      "clip_ratio/low_mean": 0.0010629086536937393,
+      "clip_ratio/low_min": 4.8934634833130985e-05,
+      "clip_ratio/region_mean": 0.0019585124027798884,
+      "epoch": 3.186588921282799,
+      "grad_norm": 0.11548085510730743,
+      "learning_rate": 1e-06,
+      "loss": 0.0053,
+      "step": 308
+    },
+    {
+      "clip_ratio/high_max": 0.0019406232968322001,
+      "clip_ratio/high_mean": 0.0008760828750382643,
+      "clip_ratio/low_mean": 0.0012027405537082814,
+      "clip_ratio/low_min": 0.00011959243602177594,
+      "clip_ratio/region_mean": 0.0020788234687643126,
+      "epoch": 3.195918367346939,
+      "grad_norm": 0.12193051725625992,
+      "learning_rate": 1e-06,
+      "loss": 0.0304,
+      "step": 309
+    },
+    {
+      "clip_ratio/high_max": 0.0019967893094872124,
+      "clip_ratio/high_mean": 0.000845413793285843,
+      "clip_ratio/low_mean": 0.0011639131953415927,
+      "clip_ratio/low_min": 0.00018040690065390663,
+      "clip_ratio/region_mean": 0.002009326984989457,
+      "epoch": 3.205247813411079,
+      "grad_norm": 0.13364604115486145,
+      "learning_rate": 1e-06,
+      "loss": 0.0107,
+      "step": 310
+    },
+    {
+      "clip_ratio/high_max": 0.00245230059226742,
+      "clip_ratio/high_mean": 0.0010746660082077142,
+      "clip_ratio/low_mean": 0.0010397617770649958,
+      "clip_ratio/low_min": 4.257760610926198e-05,
+      "clip_ratio/region_mean": 0.0021144277998246253,
+      "epoch": 3.2145772594752184,
+      "grad_norm": 0.11715587973594666,
+      "learning_rate": 1e-06,
+      "loss": -0.0047,
+      "step": 311
+    },
+    {
+      "clip_ratio/high_max": 0.00253189260547515,
+      "clip_ratio/high_mean": 0.0010551489103818312,
+      "clip_ratio/low_mean": 0.001098928429200896,
+      "clip_ratio/low_min": 0.0001630687866054359,
+      "clip_ratio/region_mean": 0.002154077374143526,
+      "epoch": 3.2239067055393584,
+      "grad_norm": 0.12289279699325562,
+      "learning_rate": 1e-06,
+      "loss": -0.025,
+      "step": 312
+    },
+    {
+      "clip_ratio/high_max": 0.002433523208310362,
+      "clip_ratio/high_mean": 0.0009889512821246171,
+      "clip_ratio/low_mean": 0.0011194155713383225,
+      "clip_ratio/low_min": 7.554292278655339e-05,
+      "clip_ratio/region_mean": 0.0021083668834762648,
+      "epoch": 3.2332361516034984,
+      "grad_norm": 0.11821790039539337,
+      "learning_rate": 1e-06,
+      "loss": -0.0127,
+      "step": 313
+    },
+    {
+      "clip_ratio/high_max": 0.0023131492198444903,
+      "clip_ratio/high_mean": 0.0009166020208795089,
+      "clip_ratio/low_mean": 0.0010815071509568952,
+      "clip_ratio/low_min": 8.322376197611447e-05,
+      "clip_ratio/region_mean": 0.001998109211854171,
+      "epoch": 3.2425655976676384,
+      "grad_norm": 0.11479413509368896,
+      "learning_rate": 1e-06,
+      "loss": 0.0202,
+      "step": 314
+    },
+    {
+      "clip_ratio/high_max": 0.002730425709160045,
+      "clip_ratio/high_mean": 0.0011020803649444133,
+      "clip_ratio/low_mean": 0.001084739196812734,
+      "clip_ratio/low_min": 5.358674025046639e-05,
+      "clip_ratio/region_mean": 0.0021868195617571473,
+      "epoch": 3.2518950437317784,
+      "grad_norm": 0.12977640330791473,
+      "learning_rate": 1e-06,
+      "loss": -0.0089,
+      "step": 315
+    },
+    {
+      "clip_ratio/high_max": 0.0025748576008481905,
+      "clip_ratio/high_mean": 0.0009752476362336893,
+      "clip_ratio/low_mean": 0.0011016235148417763,
+      "clip_ratio/low_min": 6.810174363636179e-05,
+      "clip_ratio/region_mean": 0.002076871147437487,
+      "epoch": 3.2612244897959184,
+      "grad_norm": 0.12590360641479492,
+      "learning_rate": 1e-06,
+      "loss": -0.0117,
+      "step": 316
+    },
+    {
+      "clip_ratio/high_max": 0.0022781192892580293,
+      "clip_ratio/high_mean": 0.0010031983838416636,
+      "clip_ratio/low_mean": 0.0011138434056192636,
+      "clip_ratio/low_min": 4.892947072221432e-05,
+      "clip_ratio/region_mean": 0.0021170417821849696,
+      "epoch": 3.2705539358600584,
+      "grad_norm": 0.1301450878381729,
+      "learning_rate": 1e-06,
+      "loss": 0.0104,
+      "step": 317
+    },
+    {
+      "clip_ratio/high_max": 0.002370061061810702,
+      "clip_ratio/high_mean": 0.00111892673157854,
+      "clip_ratio/low_mean": 0.0008830380111248815,
+      "clip_ratio/low_min": 4.488076410780195e-05,
+      "clip_ratio/region_mean": 0.002001964741793927,
+      "epoch": 3.2798833819241984,
+      "grad_norm": 0.12157122045755386,
+      "learning_rate": 1e-06,
+      "loss": -0.0382,
+      "step": 318
+    },
+    {
+      "clip_ratio/high_max": 0.0024365238241443876,
+      "clip_ratio/high_mean": 0.0010116751200257568,
+      "clip_ratio/low_mean": 0.0009448562050238252,
+      "clip_ratio/low_min": 5.7639103033579886e-05,
+      "clip_ratio/region_mean": 0.00195653139235219,
+      "epoch": 3.2892128279883384,
+      "grad_norm": 0.1190371960401535,
+      "learning_rate": 1e-06,
+      "loss": -0.0196,
+      "step": 319
+    },
+    {
+      "clip_ratio/high_max": 0.0023169510532170534,
+      "clip_ratio/high_mean": 0.0010022043934441172,
+      "clip_ratio/low_mean": 0.0008849693440424744,
+      "clip_ratio/low_min": 6.96743718435755e-05,
+      "clip_ratio/region_mean": 0.0018871736974688247,
+      "epoch": 3.298542274052478,
+      "grad_norm": 0.11660901457071304,
+      "learning_rate": 1e-06,
+      "loss": -0.0,
+      "step": 320
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.024762834821428603,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4043.0,
+      "completions/mean_length": 641.1109619140625,
+      "completions/mean_terminated_length": 553.3856811523438,
+      "completions/min_length": 5.0,
+      "completions/min_terminated_length": 5.0,
+      "epoch": 3.307871720116618,
+      "grad_norm": 0.13143830001354218,
+      "learning_rate": 1e-06,
+      "loss": -0.0357,
+      "num_tokens": 204588540.0,
+      "reward": 0.5855538845062256,
+      "reward_std": 0.18952247500419617,
+      "rewards/simpleverify_reward/mean": 0.5855538249015808,
+      "rewards/simpleverify_reward/std": 0.49263474345207214,
+      "step": 321
+    },
+    {
+      "clip_ratio/high_max": 0.002081824990455061,
+      "clip_ratio/high_mean": 0.0008205740396078909,
+      "clip_ratio/low_mean": 0.0006175091480145056,
+      "clip_ratio/low_min": 9.681323626864469e-05,
+      "clip_ratio/region_mean": 0.0014380832144524902,
+      "epoch": 3.317201166180758,
+      "grad_norm": 0.12832291424274445,
+      "learning_rate": 1e-06,
+      "loss": 0.0216,
+      "step": 322
+    },
+    {
+      "clip_ratio/high_max": 0.002198897178459447,
+      "clip_ratio/high_mean": 0.0008805768675301806,
+      "clip_ratio/low_mean": 0.0005757844255640521,
+      "clip_ratio/low_min": 7.385367462120485e-05,
+      "clip_ratio/region_mean": 0.0014563612530764658,
+      "epoch": 3.326530612244898,
+      "grad_norm": 0.12534086406230927,
+      "learning_rate": 1e-06,
+      "loss": 0.018,
+      "step": 323
+    },
+    {
+      "clip_ratio/high_max": 0.0015997640330169816,
+      "clip_ratio/high_mean": 0.0006852754158899188,
+      "clip_ratio/low_mean": 0.0005495695841091219,
+      "clip_ratio/low_min": 4.164241727266926e-05,
+      "clip_ratio/region_mean": 0.0012348450109129772,
+      "epoch": 3.335860058309038,
+      "grad_norm": 0.12279549241065979,
+      "learning_rate": 1e-06,
+      "loss": -0.0018,
+      "step": 324
+    },
+    {
+      "clip_ratio/high_max": 0.001829999739129562,
+      "clip_ratio/high_mean": 0.0007402574010484386,
+      "clip_ratio/low_mean": 0.0007085395145622897,
+      "clip_ratio/low_min": 5.3915501666779164e-05,
+      "clip_ratio/region_mean": 0.0014487969128822442,
+      "epoch": 3.345189504373178,
+      "grad_norm": 0.13169774413108826,
+      "learning_rate": 1e-06,
+      "loss": 0.0094,
+      "step": 325
+    },
+    {
+      "clip_ratio/high_max": 0.002394345690845512,
+      "clip_ratio/high_mean": 0.000877591624885099,
+      "clip_ratio/low_mean": 0.0007403057497867849,
+      "clip_ratio/low_min": 6.64371509628836e-05,
+      "clip_ratio/region_mean": 0.001617897352844011,
+      "epoch": 3.354518950437318,
+      "grad_norm": 0.2514910399913788,
+      "learning_rate": 1e-06,
+      "loss": -0.0004,
+      "step": 326
+    },
+    {
+      "clip_ratio/high_max": 0.0021359779057092965,
+      "clip_ratio/high_mean": 0.000823063757707132,
+      "clip_ratio/low_mean": 0.000759678914619144,
+      "clip_ratio/low_min": 2.57315850831219e-05,
+      "clip_ratio/region_mean": 0.0015827426395844668,
+      "epoch": 3.363848396501458,
+      "grad_norm": 0.13429471850395203,
+      "learning_rate": 1e-06,
+      "loss": -0.0016,
+      "step": 327
+    },
+    {
+      "clip_ratio/high_max": 0.0020860579461441375,
+      "clip_ratio/high_mean": 0.0009367309721710626,
+      "clip_ratio/low_mean": 0.0006439800363295944,
+      "clip_ratio/low_min": 1.3525211215892341e-05,
+      "clip_ratio/region_mean": 0.0015807109957677312,
+      "epoch": 3.373177842565598,
+      "grad_norm": 0.13580140471458435,
+      "learning_rate": 1e-06,
+      "loss": -0.028,
+      "step": 328
+    },
+    {
+      "clip_ratio/high_max": 0.002187342368415557,
+      "clip_ratio/high_mean": 0.0009016236690513324,
+      "clip_ratio/low_mean": 0.0007877179668867029,
+      "clip_ratio/low_min": 2.0298797608120367e-05,
+      "clip_ratio/region_mean": 0.001689341625024099,
+      "epoch": 3.3825072886297374,
+      "grad_norm": 0.12560588121414185,
+      "learning_rate": 1e-06,
+      "loss": -0.0256,
+      "step": 329
+    },
+    {
+      "clip_ratio/high_max": 0.0021584258756774943,
+      "clip_ratio/high_mean": 0.0008691073508089175,
+      "clip_ratio/low_mean": 0.0008270080215879716,
+      "clip_ratio/low_min": 7.240215199999511e-05,
+      "clip_ratio/region_mean": 0.0016961153742158785,
+      "epoch": 3.3918367346938774,
+      "grad_norm": 0.12447452545166016,
+      "learning_rate": 1e-06,
+      "loss": 0.0266,
+      "step": 330
+    },
+    {
+      "clip_ratio/high_max": 0.0020215578115312383,
+      "clip_ratio/high_mean": 0.0008933574263210176,
+      "clip_ratio/low_mean": 0.0009193233445330407,
+      "clip_ratio/low_min": 0.00010749709053925471,
+      "clip_ratio/region_mean": 0.0018126807772205211,
+      "epoch": 3.4011661807580174,
+      "grad_norm": 0.12126223742961884,
+      "learning_rate": 1e-06,
+      "loss": -0.0146,
+      "step": 331
+    },
+    {
+      "clip_ratio/high_max": 0.00199752864864422,
+      "clip_ratio/high_mean": 0.0008398552072321763,
+      "clip_ratio/low_mean": 0.000931416725507006,
+      "clip_ratio/low_min": 5.21423698955914e-05,
+      "clip_ratio/region_mean": 0.001771271912730299,
+      "epoch": 3.4104956268221573,
+      "grad_norm": 0.12881004810333252,
+      "learning_rate": 1e-06,
+      "loss": 0.0245,
+      "step": 332
+    },
+    {
+      "clip_ratio/high_max": 0.0021911727671977133,
+      "clip_ratio/high_mean": 0.0009148283461399842,
+      "clip_ratio/low_mean": 0.0009713266081234906,
+      "clip_ratio/low_min": 4.489740058488678e-05,
+      "clip_ratio/region_mean": 0.0018861549397115596,
+      "epoch": 3.4198250728862973,
+      "grad_norm": 0.12141360342502594,
+      "learning_rate": 1e-06,
+      "loss": -0.0069,
+      "step": 333
+    },
+    {
+      "clip_ratio/high_max": 0.002078093330055708,
+      "clip_ratio/high_mean": 0.0009021739751915447,
+      "clip_ratio/low_mean": 0.0008745343566260999,
+      "clip_ratio/low_min": 1.9218941815779544e-05,
+      "clip_ratio/region_mean": 0.001776708279066952,
+      "epoch": 3.4291545189504373,
+      "grad_norm": 0.12522944808006287,
+      "learning_rate": 1e-06,
+      "loss": -0.0024,
+      "step": 334
+    },
+    {
+      "clip_ratio/high_max": 0.0020181925283395685,
+      "clip_ratio/high_mean": 0.0008280649199150503,
+      "clip_ratio/low_mean": 0.0008462995047011646,
+      "clip_ratio/low_min": 3.677912900457159e-05,
+      "clip_ratio/region_mean": 0.0016743644227972254,
+      "epoch": 3.4384839650145773,
+      "grad_norm": 0.11012354493141174,
+      "learning_rate": 1e-06,
+      "loss": 0.0217,
+      "step": 335
+    },
+    {
+      "clip_ratio/high_max": 0.0025067896640393883,
+      "clip_ratio/high_mean": 0.0010184146703977603,
+      "clip_ratio/low_mean": 0.000887044805494952,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019054595031775534,
+      "epoch": 3.4478134110787173,
+      "grad_norm": 0.1256856620311737,
+      "learning_rate": 1e-06,
+      "loss": -0.0193,
+      "step": 336
+    },
+    {
+      "clip_ratio/high_max": 0.002139975727914134,
+      "clip_ratio/high_mean": 0.000913791202037828,
+      "clip_ratio/low_mean": 0.0009576292795827612,
+      "clip_ratio/low_min": 5.092341234558262e-05,
+      "clip_ratio/region_mean": 0.0018714204561547376,
+      "epoch": 3.4571428571428573,
+      "grad_norm": 0.11477341502904892,
+      "learning_rate": 1e-06,
+      "loss": -0.028,
+      "step": 337
+    },
+    {
+      "clip_ratio/high_max": 0.0022000954413670115,
+      "clip_ratio/high_mean": 0.0008663976386742434,
+      "clip_ratio/low_mean": 0.0010844683711184189,
+      "clip_ratio/low_min": 0.00013487641263054684,
+      "clip_ratio/region_mean": 0.0019508659897837788,
+      "epoch": 3.466472303206997,
+      "grad_norm": 0.11449246108531952,
+      "learning_rate": 1e-06,
+      "loss": 0.0444,
+      "step": 338
+    },
+    {
+      "clip_ratio/high_max": 0.0023874429971328937,
+      "clip_ratio/high_mean": 0.0010054052509076428,
+      "clip_ratio/low_mean": 0.001076638613085379,
+      "clip_ratio/low_min": 6.585190112673445e-05,
+      "clip_ratio/region_mean": 0.002082043924019672,
+      "epoch": 3.4758017492711373,
+      "grad_norm": 0.12477981299161911,
+      "learning_rate": 1e-06,
+      "loss": -0.0163,
+      "step": 339
+    },
+    {
+      "clip_ratio/high_max": 0.0026525383509579115,
+      "clip_ratio/high_mean": 0.0010539999675529543,
+      "clip_ratio/low_mean": 0.0009186538081848994,
+      "clip_ratio/low_min": 7.1315276727546e-05,
+      "clip_ratio/region_mean": 0.0019726537939277478,
+      "epoch": 3.485131195335277,
+      "grad_norm": 0.12025745213031769,
+      "learning_rate": 1e-06,
+      "loss": -0.0332,
+      "step": 340
+    },
+    {
+      "clip_ratio/high_max": 0.0023425274484907277,
+      "clip_ratio/high_mean": 0.0009884111386782024,
+      "clip_ratio/low_mean": 0.0010804726916830987,
+      "clip_ratio/low_min": 0.00011691356212395476,
+      "clip_ratio/region_mean": 0.002068883790343534,
+      "epoch": 3.494460641399417,
+      "grad_norm": 0.12572072446346283,
+      "learning_rate": 1e-06,
+      "loss": 0.0239,
+      "step": 341
+    },
+    {
+      "clip_ratio/high_max": 0.0024941686606325675,
+      "clip_ratio/high_mean": 0.0009909349610097706,
+      "clip_ratio/low_mean": 0.0008940499683376402,
+      "clip_ratio/low_min": 2.4437509637209587e-05,
+      "clip_ratio/region_mean": 0.0018849849584512413,
+      "epoch": 3.503790087463557,
+      "grad_norm": 0.11796656250953674,
+      "learning_rate": 1e-06,
+      "loss": -0.0111,
+      "step": 342
+    },
+    {
+      "clip_ratio/high_max": 0.0023195863541332074,
+      "clip_ratio/high_mean": 0.000934428697291878,
+      "clip_ratio/low_mean": 0.0009639933014113922,
+      "clip_ratio/low_min": 2.5450475732213818e-05,
+      "clip_ratio/region_mean": 0.0018984220077982172,
+      "epoch": 3.513119533527697,
+      "grad_norm": 0.1288626492023468,
+      "learning_rate": 1e-06,
+      "loss": 0.0287,
+      "step": 343
+    },
+    {
+      "clip_ratio/high_max": 0.002286431972606806,
+      "clip_ratio/high_mean": 0.0008174578824764467,
+      "clip_ratio/low_mean": 0.0010058387506433064,
+      "clip_ratio/low_min": 0.0001309450399276102,
+      "clip_ratio/region_mean": 0.0018232966467621736,
+      "epoch": 3.522448979591837,
+      "grad_norm": 0.13154514133930206,
+      "learning_rate": 1e-06,
+      "loss": 0.0587,
+      "step": 344
+    },
+    {
+      "clip_ratio/high_max": 0.002750756233581342,
+      "clip_ratio/high_mean": 0.0010733838680607732,
+      "clip_ratio/low_mean": 0.0010943487650365569,
+      "clip_ratio/low_min": 0.00012019909354421543,
+      "clip_ratio/region_mean": 0.002167732636735309,
+      "epoch": 3.5317784256559768,
+      "grad_norm": 0.12297602742910385,
+      "learning_rate": 1e-06,
+      "loss": -0.01,
+      "step": 345
+    },
+    {
+      "clip_ratio/high_max": 0.0024775554265943356,
+      "clip_ratio/high_mean": 0.0010784051792143146,
+      "clip_ratio/low_mean": 0.0009597678636055207,
+      "clip_ratio/low_min": 2.7895559469470754e-05,
+      "clip_ratio/region_mean": 0.00203817302826792,
+      "epoch": 3.5411078717201168,
+      "grad_norm": 0.12645237147808075,
+      "learning_rate": 1e-06,
+      "loss": -0.0024,
+      "step": 346
+    },
+    {
+      "clip_ratio/high_max": 0.0021589628449874,
+      "clip_ratio/high_mean": 0.0009194501835736446,
+      "clip_ratio/low_mean": 0.0010389943017798942,
+      "clip_ratio/low_min": 5.7541479691280983e-05,
+      "clip_ratio/region_mean": 0.001958444496267475,
+      "epoch": 3.5504373177842563,
+      "grad_norm": 0.12677398324012756,
+      "learning_rate": 1e-06,
+      "loss": 0.0225,
+      "step": 347
+    },
+    {
+      "clip_ratio/high_max": 0.0023575429804623127,
+      "clip_ratio/high_mean": 0.000881888310686918,
+      "clip_ratio/low_mean": 0.001199412508867681,
+      "clip_ratio/low_min": 8.382833038922399e-05,
+      "clip_ratio/region_mean": 0.002081300764984917,
+      "epoch": 3.5597667638483967,
+      "grad_norm": 0.12611868977546692,
+      "learning_rate": 1e-06,
+      "loss": 0.0482,
+      "step": 348
+    },
+    {
+      "clip_ratio/high_max": 0.002348348185478244,
+      "clip_ratio/high_mean": 0.0009638598930905573,
+      "clip_ratio/low_mean": 0.0009683600837888662,
+      "clip_ratio/low_min": 7.389452457573498e-05,
+      "clip_ratio/region_mean": 0.0019322199950693175,
+      "epoch": 3.5690962099125363,
+      "grad_norm": 0.11160127073526382,
+      "learning_rate": 1e-06,
+      "loss": -0.0218,
+      "step": 349
+    },
+    {
+      "clip_ratio/high_max": 0.0027372134500183165,
+      "clip_ratio/high_mean": 0.0011217370083613787,
+      "clip_ratio/low_mean": 0.001023474003886804,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0021452109576785006,
+      "epoch": 3.5784256559766763,
+      "grad_norm": 0.13964039087295532,
+      "learning_rate": 1e-06,
+      "loss": -0.0759,
+      "step": 350
+    },
+    {
+      "clip_ratio/high_max": 0.0023659457729081623,
+      "clip_ratio/high_mean": 0.001062861530954251,
+      "clip_ratio/low_mean": 0.0009581357580827898,
+      "clip_ratio/low_min": 5.1994421482959297e-05,
+      "clip_ratio/region_mean": 0.002020997279032599,
+      "epoch": 3.5877551020408163,
+      "grad_norm": 0.13347944617271423,
+      "learning_rate": 1e-06,
+      "loss": -0.0418,
+      "step": 351
+    },
+    {
+      "clip_ratio/high_max": 0.0024200145271606743,
+      "clip_ratio/high_mean": 0.0010322356520191533,
+      "clip_ratio/low_mean": 0.0011306301748845726,
+      "clip_ratio/low_min": 8.044620335567743e-05,
+      "clip_ratio/region_mean": 0.0021628658214467578,
+      "epoch": 3.5970845481049563,
+      "grad_norm": 0.12605777382850647,
+      "learning_rate": 1e-06,
+      "loss": 0.0252,
+      "step": 352
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0276576450892857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4082.0,
+      "completions/mean_length": 654.5404663085938,
+      "completions/mean_terminated_length": 556.650390625,
+      "completions/min_length": 4.0,
+      "completions/min_terminated_length": 4.0,
+      "epoch": 3.6064139941690962,
+      "grad_norm": 0.13812322914600372,
+      "learning_rate": 1e-06,
+      "loss": 0.0277,
+      "num_tokens": 222893389.0,
+      "reward": 0.6016671657562256,
+      "reward_std": 0.1861223429441452,
+      "rewards/simpleverify_reward/mean": 0.6016671061515808,
+      "rewards/simpleverify_reward/std": 0.4895632266998291,
+      "step": 353
+    },
+    {
+      "clip_ratio/high_max": 0.0018422315115458332,
+      "clip_ratio/high_mean": 0.0008274892297777114,
+      "clip_ratio/low_mean": 0.0005405372648965567,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013680264855793212,
+      "epoch": 3.6157434402332362,
+      "grad_norm": 0.1303921788930893,
+      "learning_rate": 1e-06,
+      "loss": 0.0182,
+      "step": 354
+    },
+    {
+      "clip_ratio/high_max": 0.002004551875870675,
+      "clip_ratio/high_mean": 0.000853116056532599,
+      "clip_ratio/low_mean": 0.0005362414376577362,
+      "clip_ratio/low_min": 5.404989406088134e-05,
+      "clip_ratio/region_mean": 0.001389357497828314,
+      "epoch": 3.6250728862973762,
+      "grad_norm": 0.12853002548217773,
+      "learning_rate": 1e-06,
+      "loss": -0.0176,
+      "step": 355
+    },
+    {
+      "clip_ratio/high_max": 0.001959608773177024,
+      "clip_ratio/high_mean": 0.0008215340203605592,
+      "clip_ratio/low_mean": 0.0006006097537465394,
+      "clip_ratio/low_min": 1.2077294741175137e-05,
+      "clip_ratio/region_mean": 0.0014221437741070986,
+      "epoch": 3.6344023323615158,
+      "grad_norm": 0.124964639544487,
+      "learning_rate": 1e-06,
+      "loss": 0.0205,
+      "step": 356
+    },
+    {
+      "clip_ratio/high_max": 0.002032963962847134,
+      "clip_ratio/high_mean": 0.0007587872114527272,
+      "clip_ratio/low_mean": 0.0006263037448661635,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013850909745087847,
+      "epoch": 3.643731778425656,
+      "grad_norm": 0.12105780839920044,
+      "learning_rate": 1e-06,
+      "loss": -0.0116,
+      "step": 357
+    },
+    {
+      "clip_ratio/high_max": 0.00184320433618268,
+      "clip_ratio/high_mean": 0.000894339007572853,
+      "clip_ratio/low_mean": 0.0006461292405219865,
+      "clip_ratio/low_min": 5.668202902597841e-05,
+      "clip_ratio/region_mean": 0.0015404682562802918,
+      "epoch": 3.6530612244897958,
+      "grad_norm": 0.1163407415151596,
+      "learning_rate": 1e-06,
+      "loss": -0.0305,
+      "step": 358
+    },
+    {
+      "clip_ratio/high_max": 0.0020696637948276475,
+      "clip_ratio/high_mean": 0.0009150950991170248,
+      "clip_ratio/low_mean": 0.000654317571388674,
+      "clip_ratio/low_min": 8.144675121002365e-05,
+      "clip_ratio/region_mean": 0.0015694126777816564,
+      "epoch": 3.6623906705539357,
+      "grad_norm": 0.1236238107085228,
+      "learning_rate": 1e-06,
+      "loss": -0.0165,
+      "step": 359
+    },
+    {
+      "clip_ratio/high_max": 0.0020081195543752983,
+      "clip_ratio/high_mean": 0.0007979373476700857,
+      "clip_ratio/low_mean": 0.0007721411038801307,
+      "clip_ratio/low_min": 1.3796909115626477e-05,
+      "clip_ratio/region_mean": 0.001570078449731227,
+      "epoch": 3.6717201166180757,
+      "grad_norm": 0.13558252155780792,
+      "learning_rate": 1e-06,
+      "loss": 0.0429,
+      "step": 360
+    },
+    {
+      "clip_ratio/high_max": 0.0016852559238031972,
+      "clip_ratio/high_mean": 0.0007392789998448279,
+      "clip_ratio/low_mean": 0.0007591812009195564,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014984601948526688,
+      "epoch": 3.6810495626822157,
+      "grad_norm": 0.1261928379535675,
+      "learning_rate": 1e-06,
+      "loss": 0.012,
+      "step": 361
+    },
+    {
+      "clip_ratio/high_max": 0.0020385598800203297,
+      "clip_ratio/high_mean": 0.0009380048486491432,
+      "clip_ratio/low_mean": 0.0008080076513579115,
+      "clip_ratio/low_min": 3.826955253316555e-05,
+      "clip_ratio/region_mean": 0.0017460125018260442,
+      "epoch": 3.6903790087463557,
+      "grad_norm": 0.1256881207227707,
+      "learning_rate": 1e-06,
+      "loss": -0.0075,
+      "step": 362
+    },
+    {
+      "clip_ratio/high_max": 0.0019200395836378448,
+      "clip_ratio/high_mean": 0.0009273255309381057,
+      "clip_ratio/low_mean": 0.0007339983148995088,
+      "clip_ratio/low_min": 1.0578875844657887e-05,
+      "clip_ratio/region_mean": 0.0016613238731224556,
+      "epoch": 3.6997084548104957,
+      "grad_norm": 0.1309698224067688,
+      "learning_rate": 1e-06,
+      "loss": -0.014,
+      "step": 363
+    },
+    {
+      "clip_ratio/high_max": 0.002184991230024025,
+      "clip_ratio/high_mean": 0.0010025849041994661,
+      "clip_ratio/low_mean": 0.0007874047514633276,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001789989612007048,
+      "epoch": 3.7090379008746357,
+      "grad_norm": 0.12068425863981247,
+      "learning_rate": 1e-06,
+      "loss": 0.0022,
+      "step": 364
+    },
+    {
+      "clip_ratio/high_max": 0.002338721649721265,
+      "clip_ratio/high_mean": 0.0009570841084496351,
+      "clip_ratio/low_mean": 0.0008360726988030365,
+      "clip_ratio/low_min": 7.409069257846568e-05,
+      "clip_ratio/region_mean": 0.00179315678178682,
+      "epoch": 3.7183673469387752,
+      "grad_norm": 0.13551045954227448,
+      "learning_rate": 1e-06,
+      "loss": 0.019,
+      "step": 365
+    },
+    {
+      "clip_ratio/high_max": 0.002144247839169111,
+      "clip_ratio/high_mean": 0.0010146544918825384,
+      "clip_ratio/low_mean": 0.0008285586518468335,
+      "clip_ratio/low_min": 8.70294970809482e-05,
+      "clip_ratio/region_mean": 0.0018432131328154355,
+      "epoch": 3.7276967930029157,
+      "grad_norm": 0.13341477513313293,
+      "learning_rate": 1e-06,
+      "loss": -0.0283,
+      "step": 366
+    },
+    {
+      "clip_ratio/high_max": 0.0022179721199790947,
+      "clip_ratio/high_mean": 0.0009786770642676856,
+      "clip_ratio/low_mean": 0.0007432454131048871,
+      "clip_ratio/low_min": 8.223997792811133e-05,
+      "clip_ratio/region_mean": 0.0017219224901054986,
+      "epoch": 3.7370262390670552,
+      "grad_norm": 0.12940876185894012,
+      "learning_rate": 1e-06,
+      "loss": -0.0302,
+      "step": 367
+    },
+    {
+      "clip_ratio/high_max": 0.002196530156652443,
+      "clip_ratio/high_mean": 0.0010002526360040065,
+      "clip_ratio/low_mean": 0.0008059659430728061,
+      "clip_ratio/low_min": 3.0508474083035253e-05,
+      "clip_ratio/region_mean": 0.0018062185772578232,
+      "epoch": 3.746355685131195,
+      "grad_norm": 0.11200443655252457,
+      "learning_rate": 1e-06,
+      "loss": -0.0069,
+      "step": 368
+    },
+    {
+      "clip_ratio/high_max": 0.002153107438061852,
+      "clip_ratio/high_mean": 0.0009139909452642314,
+      "clip_ratio/low_mean": 0.0008407793829974253,
+      "clip_ratio/low_min": 6.661612587777199e-05,
+      "clip_ratio/region_mean": 0.001754770339175593,
+      "epoch": 3.755685131195335,
+      "grad_norm": 0.12188612669706345,
+      "learning_rate": 1e-06,
+      "loss": -0.0438,
+      "step": 369
+    },
+    {
+      "clip_ratio/high_max": 0.00217586851795204,
+      "clip_ratio/high_mean": 0.0009423261890333379,
+      "clip_ratio/low_mean": 0.000926724766031839,
+      "clip_ratio/low_min": 4.10402362831519e-05,
+      "clip_ratio/region_mean": 0.0018690509314183146,
+      "epoch": 3.765014577259475,
+      "grad_norm": 0.11991658806800842,
+      "learning_rate": 1e-06,
+      "loss": -0.017,
+      "step": 370
+    },
+    {
+      "clip_ratio/high_max": 0.002388638698903378,
+      "clip_ratio/high_mean": 0.001035690598655492,
+      "clip_ratio/low_mean": 0.0008325787930516526,
+      "clip_ratio/low_min": 2.7808289814856835e-05,
+      "clip_ratio/region_mean": 0.0018682693917071447,
+      "epoch": 3.774344023323615,
+      "grad_norm": 0.13602958619594574,
+      "learning_rate": 1e-06,
+      "loss": -0.0342,
+      "step": 371
+    },
+    {
+      "clip_ratio/high_max": 0.00253827672713669,
+      "clip_ratio/high_mean": 0.0009370409970870242,
+      "clip_ratio/low_mean": 0.0008679752863827161,
+      "clip_ratio/low_min": 1.0249262231809553e-05,
+      "clip_ratio/region_mean": 0.001805016290745698,
+      "epoch": 3.783673469387755,
+      "grad_norm": 0.12156805396080017,
+      "learning_rate": 1e-06,
+      "loss": 0.0102,
+      "step": 372
+    },
+    {
+      "clip_ratio/high_max": 0.002167011196434032,
+      "clip_ratio/high_mean": 0.0008962552565208171,
+      "clip_ratio/low_mean": 0.0009696360484667821,
+      "clip_ratio/low_min": 5.224735650699586e-05,
+      "clip_ratio/region_mean": 0.0018658912958926521,
+      "epoch": 3.793002915451895,
+      "grad_norm": 0.1260608732700348,
+      "learning_rate": 1e-06,
+      "loss": 0.0242,
+      "step": 373
+    },
+    {
+      "clip_ratio/high_max": 0.002384423372859601,
+      "clip_ratio/high_mean": 0.0010103317763423547,
+      "clip_ratio/low_mean": 0.0009639660765969893,
+      "clip_ratio/low_min": 2.18150089494884e-05,
+      "clip_ratio/region_mean": 0.0019742978547583334,
+      "epoch": 3.8023323615160347,
+      "grad_norm": 0.12216995656490326,
+      "learning_rate": 1e-06,
+      "loss": -0.0221,
+      "step": 374
+    },
+    {
+      "clip_ratio/high_max": 0.0023978524113772437,
+      "clip_ratio/high_mean": 0.0010300275062036235,
+      "clip_ratio/low_mean": 0.001245647443283815,
+      "clip_ratio/low_min": 0.00014081780500418972,
+      "clip_ratio/region_mean": 0.002275674960401375,
+      "epoch": 3.811661807580175,
+      "grad_norm": 0.12235193699598312,
+      "learning_rate": 1e-06,
+      "loss": 0.0046,
+      "step": 375
+    },
+    {
+      "clip_ratio/high_max": 0.002689565109903924,
+      "clip_ratio/high_mean": 0.0010174690614803694,
+      "clip_ratio/low_mean": 0.000994558442471316,
+      "clip_ratio/low_min": 2.5107806322921533e-05,
+      "clip_ratio/region_mean": 0.002012027493037749,
+      "epoch": 3.8209912536443147,
+      "grad_norm": 0.13323216140270233,
+      "learning_rate": 1e-06,
+      "loss": 0.0106,
+      "step": 376
+    },
+    {
+      "clip_ratio/high_max": 0.002252488280646503,
+      "clip_ratio/high_mean": 0.0009471086705161724,
+      "clip_ratio/low_mean": 0.0010546732955845073,
+      "clip_ratio/low_min": 9.687905276223319e-05,
+      "clip_ratio/region_mean": 0.002001781962462701,
+      "epoch": 3.8303206997084547,
+      "grad_norm": 0.13442502915859222,
+      "learning_rate": 1e-06,
+      "loss": -0.0271,
+      "step": 377
+    },
+    {
+      "clip_ratio/high_max": 0.0022082783034420572,
+      "clip_ratio/high_mean": 0.0009712189985293662,
+      "clip_ratio/low_mean": 0.0009713226254461915,
+      "clip_ratio/low_min": 1.72604250110453e-05,
+      "clip_ratio/region_mean": 0.0019425416066951584,
+      "epoch": 3.8396501457725947,
+      "grad_norm": 0.11734303086996078,
+      "learning_rate": 1e-06,
+      "loss": 0.0118,
+      "step": 378
+    },
+    {
+      "clip_ratio/high_max": 0.0023211157858895604,
+      "clip_ratio/high_mean": 0.000983304671535734,
+      "clip_ratio/low_mean": 0.0010854787615244277,
+      "clip_ratio/low_min": 4.8139254431589507e-05,
+      "clip_ratio/region_mean": 0.0020687834330601618,
+      "epoch": 3.8489795918367347,
+      "grad_norm": 0.1344214528799057,
+      "learning_rate": 1e-06,
+      "loss": -0.0048,
+      "step": 379
+    },
+    {
+      "clip_ratio/high_max": 0.002492693529347889,
+      "clip_ratio/high_mean": 0.0009986599870899227,
+      "clip_ratio/low_mean": 0.0012119804705434944,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.002210640443081502,
+      "epoch": 3.8583090379008746,
+      "grad_norm": 0.13017284870147705,
+      "learning_rate": 1e-06,
+      "loss": 0.0267,
+      "step": 380
+    },
+    {
+      "clip_ratio/high_max": 0.0023947299996507354,
+      "clip_ratio/high_mean": 0.0009566030093992595,
+      "clip_ratio/low_mean": 0.0012776065541402204,
+      "clip_ratio/low_min": 0.00023669379515922628,
+      "clip_ratio/region_mean": 0.0022342095981002785,
+      "epoch": 3.8676384839650146,
+      "grad_norm": 0.1225247010588646,
+      "learning_rate": 1e-06,
+      "loss": 0.0279,
+      "step": 381
+    },
+    {
+      "clip_ratio/high_max": 0.002490529881470138,
+      "clip_ratio/high_mean": 0.0009996802818932338,
+      "clip_ratio/low_mean": 0.001070929425623035,
+      "clip_ratio/low_min": 0.00013358558408071985,
+      "clip_ratio/region_mean": 0.002070609712973237,
+      "epoch": 3.8769679300291546,
+      "grad_norm": 0.13048617541790009,
+      "learning_rate": 1e-06,
+      "loss": -0.014,
+      "step": 382
+    },
+    {
+      "clip_ratio/high_max": 0.0021763901750091463,
+      "clip_ratio/high_mean": 0.0009103734992095269,
+      "clip_ratio/low_mean": 0.0008946762300183764,
+      "clip_ratio/low_min": 3.4789731216733344e-05,
+      "clip_ratio/region_mean": 0.0018050497528747655,
+      "epoch": 3.8862973760932946,
+      "grad_norm": 0.11429255455732346,
+      "learning_rate": 1e-06,
+      "loss": -0.0015,
+      "step": 383
+    },
+    {
+      "clip_ratio/high_max": 0.0023308118325076066,
+      "clip_ratio/high_mean": 0.0008925222246034537,
+      "clip_ratio/low_mean": 0.0011587132430577185,
+      "clip_ratio/low_min": 0.00010190919147134991,
+      "clip_ratio/region_mean": 0.002051235467661172,
+      "epoch": 3.8956268221574346,
+      "grad_norm": 0.11595192551612854,
+      "learning_rate": 1e-06,
+      "loss": 0.0287,
+      "step": 384
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.027692522321428603,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4032.0,
+      "completions/mean_length": 663.2836303710938,
+      "completions/mean_terminated_length": 565.5156860351562,
+      "completions/min_length": 92.0,
+      "completions/min_terminated_length": 92.0,
+      "epoch": 4.0093294460641395,
+      "grad_norm": 0.1376991719007492,
+      "learning_rate": 1e-06,
+      "loss": 0.0001,
+      "num_tokens": 241514562.0,
+      "reward": 0.5964006781578064,
+      "reward_std": 0.1858140379190445,
+      "rewards/simpleverify_reward/mean": 0.5964006781578064,
+      "rewards/simpleverify_reward/std": 0.4906274676322937,
+      "step": 385
+    },
+    {
+      "clip_ratio/high_max": 0.0021251889993436635,
+      "clip_ratio/high_mean": 0.0008307242787850555,
+      "clip_ratio/low_mean": 0.000551025665117777,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013817499675496947,
+      "epoch": 4.01865889212828,
+      "grad_norm": 0.16409343481063843,
+      "learning_rate": 1e-06,
+      "loss": -0.0385,
+      "step": 386
+    },
+    {
+      "clip_ratio/high_max": 0.0018537556170485914,
+      "clip_ratio/high_mean": 0.0007750992826913716,
+      "clip_ratio/low_mean": 0.0004142519519518828,
+      "clip_ratio/low_min": 1.605445686436724e-05,
+      "clip_ratio/region_mean": 0.0011893512000824558,
+      "epoch": 4.0279883381924195,
+      "grad_norm": 0.126278355717659,
+      "learning_rate": 1e-06,
+      "loss": -0.0305,
+      "step": 387
+    },
+    {
+      "clip_ratio/high_max": 0.0021483809978235513,
+      "clip_ratio/high_mean": 0.0008904324367904337,
+      "clip_ratio/low_mean": 0.0005418080754679977,
+      "clip_ratio/low_min": 2.6358845389040653e-05,
+      "clip_ratio/region_mean": 0.0014322405186248943,
+      "epoch": 4.03731778425656,
+      "grad_norm": 0.13643008470535278,
+      "learning_rate": 1e-06,
+      "loss": -0.0375,
+      "step": 388
+    },
+    {
+      "clip_ratio/high_max": 0.001825592560635414,
+      "clip_ratio/high_mean": 0.0008252827028627507,
+      "clip_ratio/low_mean": 0.0005161194167158101,
+      "clip_ratio/low_min": 4.147367417317582e-05,
+      "clip_ratio/region_mean": 0.0013414021450444125,
+      "epoch": 4.0466472303206995,
+      "grad_norm": 0.11691092699766159,
+      "learning_rate": 1e-06,
+      "loss": -0.0049,
+      "step": 389
+    },
+    {
+      "clip_ratio/high_max": 0.0018002502911258489,
+      "clip_ratio/high_mean": 0.0007384081254713237,
+      "clip_ratio/low_mean": 0.0006537441067848704,
+      "clip_ratio/low_min": 6.329604002530687e-05,
+      "clip_ratio/region_mean": 0.0013921522513555828,
+      "epoch": 4.05597667638484,
+      "grad_norm": 0.11648301035165787,
+      "learning_rate": 1e-06,
+      "loss": 0.038,
+      "step": 390
+    },
+    {
+      "clip_ratio/high_max": 0.0021319622101145796,
+      "clip_ratio/high_mean": 0.0008681447943672538,
+      "clip_ratio/low_mean": 0.0006683531482849503,
+      "clip_ratio/low_min": 1.7356289390590973e-05,
+      "clip_ratio/region_mean": 0.001536497933557257,
+      "epoch": 4.0653061224489795,
+      "grad_norm": 0.12540303170681,
+      "learning_rate": 1e-06,
+      "loss": 0.0174,
+      "step": 391
+    },
+    {
+      "clip_ratio/high_max": 0.001953968392626848,
+      "clip_ratio/high_mean": 0.0007701072099735029,
+      "clip_ratio/low_mean": 0.0006839515426690923,
+      "clip_ratio/low_min": 3.984321574534988e-05,
+      "clip_ratio/region_mean": 0.001454058728995733,
+      "epoch": 4.07463556851312,
+      "grad_norm": 0.11700475215911865,
+      "learning_rate": 1e-06,
+      "loss": -0.0003,
+      "step": 392
+    },
+    {
+      "clip_ratio/high_max": 0.0021675254974979907,
+      "clip_ratio/high_mean": 0.0009661978147050831,
+      "clip_ratio/low_mean": 0.0007228480244521052,
+      "clip_ratio/low_min": 2.3316544684348628e-05,
+      "clip_ratio/region_mean": 0.0016890458100533579,
+      "epoch": 4.0839650145772595,
+      "grad_norm": 0.12342169880867004,
+      "learning_rate": 1e-06,
+      "loss": 0.0181,
+      "step": 393
+    },
+    {
+      "clip_ratio/high_max": 0.0021691713627660647,
+      "clip_ratio/high_mean": 0.0008687900899531087,
+      "clip_ratio/low_mean": 0.0008125464610202471,
+      "clip_ratio/low_min": 5.02940674778074e-05,
+      "clip_ratio/region_mean": 0.001681336565525271,
+      "epoch": 4.093294460641399,
+      "grad_norm": 0.121994249522686,
+      "learning_rate": 1e-06,
+      "loss": 0.0254,
+      "step": 394
+    },
+    {
+      "clip_ratio/high_max": 0.001994244616071228,
+      "clip_ratio/high_mean": 0.000891088569915155,
+      "clip_ratio/low_mean": 0.0007734938753856113,
+      "clip_ratio/low_min": 9.669539304013597e-05,
+      "clip_ratio/region_mean": 0.0016645824434817769,
+      "epoch": 4.1026239067055394,
+      "grad_norm": 0.1204499751329422,
+      "learning_rate": 1e-06,
+      "loss": -0.0036,
+      "step": 395
+    },
+    {
+      "clip_ratio/high_max": 0.0020316431546234526,
+      "clip_ratio/high_mean": 0.0008603325077274349,
+      "clip_ratio/low_mean": 0.0007531925039074849,
+      "clip_ratio/low_min": 6.459650103352033e-05,
+      "clip_ratio/region_mean": 0.001613525007996941,
+      "epoch": 4.111953352769679,
+      "grad_norm": 0.1130305752158165,
+      "learning_rate": 1e-06,
+      "loss": -0.0257,
+      "step": 396
+    },
+    {
+      "clip_ratio/high_max": 0.002003321271331515,
+      "clip_ratio/high_mean": 0.0008780704683886142,
+      "clip_ratio/low_mean": 0.0008695145625097211,
+      "clip_ratio/low_min": 4.928806265525054e-05,
+      "clip_ratio/region_mean": 0.0017475850399932824,
+      "epoch": 4.121282798833819,
+      "grad_norm": 0.13229753077030182,
+      "learning_rate": 1e-06,
+      "loss": 0.022,
+      "step": 397
+    },
+    {
+      "clip_ratio/high_max": 0.001952392078237608,
+      "clip_ratio/high_mean": 0.0008767041690589394,
+      "clip_ratio/low_mean": 0.0009058000978257041,
+      "clip_ratio/low_min": 2.510012564016506e-05,
+      "clip_ratio/region_mean": 0.0017825042305048555,
+      "epoch": 4.130612244897959,
+      "grad_norm": 0.13723908364772797,
+      "learning_rate": 1e-06,
+      "loss": -0.021,
+      "step": 398
+    },
+    {
+      "clip_ratio/high_max": 0.0022944982047192752,
+      "clip_ratio/high_mean": 0.0008759930078667821,
+      "clip_ratio/low_mean": 0.0008892982114048209,
+      "clip_ratio/low_min": 4.741707016364671e-05,
+      "clip_ratio/region_mean": 0.0017652911992627196,
+      "epoch": 4.139941690962099,
+      "grad_norm": 0.1374003142118454,
+      "learning_rate": 1e-06,
+      "loss": -0.0061,
+      "step": 399
+    },
+    {
+      "clip_ratio/high_max": 0.0021894147284911014,
+      "clip_ratio/high_mean": 0.0009306346437369939,
+      "clip_ratio/low_mean": 0.0010372229226049967,
+      "clip_ratio/low_min": 9.53391172515694e-05,
+      "clip_ratio/region_mean": 0.0019678575481520966,
+      "epoch": 4.149271137026239,
+      "grad_norm": 0.1257951706647873,
+      "learning_rate": 1e-06,
+      "loss": 0.0418,
+      "step": 400
+    },
+    {
+      "clip_ratio/high_max": 0.002013164856180083,
+      "clip_ratio/high_mean": 0.0008204853465940687,
+      "clip_ratio/low_mean": 0.0009291273418057244,
+      "clip_ratio/low_min": 6.969223613850772e-05,
+      "clip_ratio/region_mean": 0.0017496126747573726,
+      "epoch": 4.158600583090379,
+      "grad_norm": 0.1225176602602005,
+      "learning_rate": 1e-06,
+      "loss": 0.0114,
+      "step": 401
+    },
+    {
+      "clip_ratio/high_max": 0.001962927562999539,
+      "clip_ratio/high_mean": 0.0008756719180382788,
+      "clip_ratio/low_mean": 0.001089063793187961,
+      "clip_ratio/low_min": 0.00016013995991670527,
+      "clip_ratio/region_mean": 0.0019647356966743246,
+      "epoch": 4.167930029154519,
+      "grad_norm": 0.13065384328365326,
+      "learning_rate": 1e-06,
+      "loss": 0.0411,
+      "step": 402
+    },
+    {
+      "clip_ratio/high_max": 0.00213753947900841,
+      "clip_ratio/high_mean": 0.0009142789604084101,
+      "clip_ratio/low_mean": 0.0009486467824899592,
+      "clip_ratio/low_min": 0.00010948617818939965,
+      "clip_ratio/region_mean": 0.00186292571015656,
+      "epoch": 4.1772594752186585,
+      "grad_norm": 0.12291523069143295,
+      "learning_rate": 1e-06,
+      "loss": 0.0106,
+      "step": 403
+    },
+    {
+      "clip_ratio/high_max": 0.0022747897091903724,
+      "clip_ratio/high_mean": 0.0010808745573740453,
+      "clip_ratio/low_mean": 0.0008359467228729045,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019168212747899815,
+      "epoch": 4.186588921282799,
+      "grad_norm": 0.11953331530094147,
+      "learning_rate": 1e-06,
+      "loss": -0.0291,
+      "step": 404
+    },
+    {
+      "clip_ratio/high_max": 0.0023405523170367815,
+      "clip_ratio/high_mean": 0.0010126458400918636,
+      "clip_ratio/low_mean": 0.0008475766117044259,
+      "clip_ratio/low_min": 7.245269898703555e-05,
+      "clip_ratio/region_mean": 0.00186022248090012,
+      "epoch": 4.1959183673469385,
+      "grad_norm": 0.1285480558872223,
+      "learning_rate": 1e-06,
+      "loss": -0.0412,
+      "step": 405
+    },
+    {
+      "clip_ratio/high_max": 0.002241056543425657,
+      "clip_ratio/high_mean": 0.0008542670420865761,
+      "clip_ratio/low_mean": 0.0009642466357036028,
+      "clip_ratio/low_min": 4.7311974412878044e-05,
+      "clip_ratio/region_mean": 0.0018185136505053379,
+      "epoch": 4.205247813411079,
+      "grad_norm": 0.12093605846166611,
+      "learning_rate": 1e-06,
+      "loss": 0.0293,
+      "step": 406
+    },
+    {
+      "clip_ratio/high_max": 0.0026399049020255916,
+      "clip_ratio/high_mean": 0.001024092507577734,
+      "clip_ratio/low_mean": 0.0010683339824026916,
+      "clip_ratio/low_min": 5.9075033277622424e-05,
+      "clip_ratio/region_mean": 0.0020924264899804257,
+      "epoch": 4.214577259475218,
+      "grad_norm": 0.13123777508735657,
+      "learning_rate": 1e-06,
+      "loss": -0.0315,
+      "step": 407
+    },
+    {
+      "clip_ratio/high_max": 0.0019685833176481538,
+      "clip_ratio/high_mean": 0.0008986030152300373,
+      "clip_ratio/low_mean": 0.0010061929096991662,
+      "clip_ratio/low_min": 2.45643268499407e-05,
+      "clip_ratio/region_mean": 0.001904795914015267,
+      "epoch": 4.223906705539359,
+      "grad_norm": 0.10999801754951477,
+      "learning_rate": 1e-06,
+      "loss": -0.0016,
+      "step": 408
+    },
+    {
+      "clip_ratio/high_max": 0.002010038551816251,
+      "clip_ratio/high_mean": 0.0008622361056040972,
+      "clip_ratio/low_mean": 0.0010035943432740169,
+      "clip_ratio/low_min": 1.2320126188569702e-05,
+      "clip_ratio/region_mean": 0.0018658304761629552,
+      "epoch": 4.233236151603498,
+      "grad_norm": 0.13467088341712952,
+      "learning_rate": 1e-06,
+      "loss": -0.0136,
+      "step": 409
+    },
+    {
+      "clip_ratio/high_max": 0.002251534075185191,
+      "clip_ratio/high_mean": 0.0010714400887081865,
+      "clip_ratio/low_mean": 0.0010257547110086307,
+      "clip_ratio/low_min": 5.750133004767122e-05,
+      "clip_ratio/region_mean": 0.002097194803354796,
+      "epoch": 4.242565597667639,
+      "grad_norm": 0.12701420485973358,
+      "learning_rate": 1e-06,
+      "loss": -0.0327,
+      "step": 410
+    },
+    {
+      "clip_ratio/high_max": 0.002209433303505648,
+      "clip_ratio/high_mean": 0.0008623511330370093,
+      "clip_ratio/low_mean": 0.0009772533121576998,
+      "clip_ratio/low_min": 2.49351687671151e-05,
+      "clip_ratio/region_mean": 0.0018396044324617833,
+      "epoch": 4.251895043731778,
+      "grad_norm": 0.12217708677053452,
+      "learning_rate": 1e-06,
+      "loss": 0.0025,
+      "step": 411
+    },
+    {
+      "clip_ratio/high_max": 0.002287858478666749,
+      "clip_ratio/high_mean": 0.0009879158860712778,
+      "clip_ratio/low_mean": 0.001147979961388046,
+      "clip_ratio/low_min": 0.00010002726048696786,
+      "clip_ratio/region_mean": 0.002135895876563154,
+      "epoch": 4.261224489795918,
+      "grad_norm": 0.13534663617610931,
+      "learning_rate": 1e-06,
+      "loss": -0.031,
+      "step": 412
+    },
+    {
+      "clip_ratio/high_max": 0.0024083175958367065,
+      "clip_ratio/high_mean": 0.0010091956610267516,
+      "clip_ratio/low_mean": 0.00103961595959845,
+      "clip_ratio/low_min": 5.982574657537043e-05,
+      "clip_ratio/region_mean": 0.0020488116642809473,
+      "epoch": 4.270553935860058,
+      "grad_norm": 0.12116041034460068,
+      "learning_rate": 1e-06,
+      "loss": -0.0161,
+      "step": 413
+    },
+    {
+      "clip_ratio/high_max": 0.002476684378052596,
+      "clip_ratio/high_mean": 0.001020087773213163,
+      "clip_ratio/low_mean": 0.0011294773466943298,
+      "clip_ratio/low_min": 0.0001158672375822789,
+      "clip_ratio/region_mean": 0.0021495650726137683,
+      "epoch": 4.279883381924198,
+      "grad_norm": 0.13888022303581238,
+      "learning_rate": 1e-06,
+      "loss": 0.0413,
+      "step": 414
+    },
+    {
+      "clip_ratio/high_max": 0.0025053853896679357,
+      "clip_ratio/high_mean": 0.001063837975380011,
+      "clip_ratio/low_mean": 0.001142918521509273,
+      "clip_ratio/low_min": 6.753106572432443e-05,
+      "clip_ratio/region_mean": 0.002206756515079178,
+      "epoch": 4.289212827988338,
+      "grad_norm": 0.12599118053913116,
+      "learning_rate": 1e-06,
+      "loss": 0.0153,
+      "step": 415
+    },
+    {
+      "clip_ratio/high_max": 0.002442953544232296,
+      "clip_ratio/high_mean": 0.0008535168371963664,
+      "clip_ratio/low_mean": 0.0011593517228902783,
+      "clip_ratio/low_min": 0.00012190126744826557,
+      "clip_ratio/region_mean": 0.0020128685573581606,
+      "epoch": 4.298542274052478,
+      "grad_norm": 0.12028119713068008,
+      "learning_rate": 1e-06,
+      "loss": 0.0342,
+      "step": 416
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0281459263392857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4047.0,
+      "completions/mean_length": 648.642578125,
+      "completions/mean_terminated_length": 548.803466796875,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 4.307871720116618,
+      "grad_norm": 0.14324797689914703,
+      "learning_rate": 1e-06,
+      "loss": 0.0201,
+      "num_tokens": 259663763.0,
+      "reward": 0.6117815375328064,
+      "reward_std": 0.18697793781757355,
+      "rewards/simpleverify_reward/mean": 0.6117815375328064,
+      "rewards/simpleverify_reward/std": 0.48735329508781433,
+      "step": 417
+    },
+    {
+      "clip_ratio/high_max": 0.0025448784217587672,
+      "clip_ratio/high_mean": 0.0009797757647902472,
+      "clip_ratio/low_mean": 0.0005918473639212607,
+      "clip_ratio/low_min": 3.9396413740178104e-05,
+      "clip_ratio/region_mean": 0.001571623120980803,
+      "epoch": 4.317201166180758,
+      "grad_norm": 0.15288770198822021,
+      "learning_rate": 1e-06,
+      "loss": 0.0297,
+      "step": 418
+    },
+    {
+      "clip_ratio/high_max": 0.0019862193912558723,
+      "clip_ratio/high_mean": 0.0008217500235332409,
+      "clip_ratio/low_mean": 0.0005634796780213946,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013852296797267627,
+      "epoch": 4.326530612244898,
+      "grad_norm": 0.11524496227502823,
+      "learning_rate": 1e-06,
+      "loss": 0.0023,
+      "step": 419
+    },
+    {
+      "clip_ratio/high_max": 0.0025638552178861573,
+      "clip_ratio/high_mean": 0.0009641898523113923,
+      "clip_ratio/low_mean": 0.000555866122340376,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00152005601557903,
+      "epoch": 4.335860058309038,
+      "grad_norm": 0.2770618498325348,
+      "learning_rate": 1e-06,
+      "loss": -0.0141,
+      "step": 420
+    },
+    {
+      "clip_ratio/high_max": 0.0019508068799041212,
+      "clip_ratio/high_mean": 0.0007618555764565826,
+      "clip_ratio/low_mean": 0.0005801574561701273,
+      "clip_ratio/low_min": 1.6622339899186045e-05,
+      "clip_ratio/region_mean": 0.001342013041721657,
+      "epoch": 4.345189504373177,
+      "grad_norm": 0.12275660783052444,
+      "learning_rate": 1e-06,
+      "loss": -0.0113,
+      "step": 421
+    },
+    {
+      "clip_ratio/high_max": 0.0023303804409806617,
+      "clip_ratio/high_mean": 0.0009664339995651972,
+      "clip_ratio/low_mean": 0.0005654622100337292,
+      "clip_ratio/low_min": 4.6924190428399015e-05,
+      "clip_ratio/region_mean": 0.0015318961814045906,
+      "epoch": 4.354518950437318,
+      "grad_norm": 0.11312326043844223,
+      "learning_rate": 1e-06,
+      "loss": -0.0149,
+      "step": 422
+    },
+    {
+      "clip_ratio/high_max": 0.001929619895236101,
+      "clip_ratio/high_mean": 0.0007667033469260787,
+      "clip_ratio/low_mean": 0.0006466154791269219,
+      "clip_ratio/low_min": 1.1927480954909697e-05,
+      "clip_ratio/region_mean": 0.0014133188560663257,
+      "epoch": 4.363848396501457,
+      "grad_norm": 0.11987412720918655,
+      "learning_rate": 1e-06,
+      "loss": -0.0113,
+      "step": 423
+    },
+    {
+      "clip_ratio/high_max": 0.001966964566236129,
+      "clip_ratio/high_mean": 0.0008856988042680314,
+      "clip_ratio/low_mean": 0.0007417448632622836,
+      "clip_ratio/low_min": 3.838701741187833e-05,
+      "clip_ratio/region_mean": 0.0016274436893581878,
+      "epoch": 4.373177842565598,
+      "grad_norm": 0.14740286767482758,
+      "learning_rate": 1e-06,
+      "loss": -0.016,
+      "step": 424
+    },
+    {
+      "clip_ratio/high_max": 0.00222642799053574,
+      "clip_ratio/high_mean": 0.0009224833866028348,
+      "clip_ratio/low_mean": 0.0006897337052578223,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016122171000461094,
+      "epoch": 4.382507288629737,
+      "grad_norm": 0.12216822803020477,
+      "learning_rate": 1e-06,
+      "loss": -0.0019,
+      "step": 425
+    },
+    {
+      "clip_ratio/high_max": 0.002756641391897574,
+      "clip_ratio/high_mean": 0.001089139046598575,
+      "clip_ratio/low_mean": 0.0007599700657010544,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018491091395844705,
+      "epoch": 4.391836734693878,
+      "grad_norm": 0.12467877566814423,
+      "learning_rate": 1e-06,
+      "loss": -0.0565,
+      "step": 426
+    },
+    {
+      "clip_ratio/high_max": 0.0023627905029570684,
+      "clip_ratio/high_mean": 0.000987551855359925,
+      "clip_ratio/low_mean": 0.0008300238441734109,
+      "clip_ratio/low_min": 6.853339073131792e-05,
+      "clip_ratio/region_mean": 0.0018175756849814206,
+      "epoch": 4.401166180758017,
+      "grad_norm": 0.13413630425930023,
+      "learning_rate": 1e-06,
+      "loss": -0.0134,
+      "step": 427
+    },
+    {
+      "clip_ratio/high_max": 0.002305820496985689,
+      "clip_ratio/high_mean": 0.0009572233175276779,
+      "clip_ratio/low_mean": 0.0007406642334899516,
+      "clip_ratio/low_min": 5.1544185225793626e-05,
+      "clip_ratio/region_mean": 0.00169788756466005,
+      "epoch": 4.410495626822158,
+      "grad_norm": 0.12544167041778564,
+      "learning_rate": 1e-06,
+      "loss": 0.0108,
+      "step": 428
+    },
+    {
+      "clip_ratio/high_max": 0.0024296663541463204,
+      "clip_ratio/high_mean": 0.0008782034237810876,
+      "clip_ratio/low_mean": 0.0007698386643824051,
+      "clip_ratio/low_min": 2.9733916562690865e-05,
+      "clip_ratio/region_mean": 0.0016480420745210722,
+      "epoch": 4.419825072886297,
+      "grad_norm": 0.11271168291568756,
+      "learning_rate": 1e-06,
+      "loss": -0.0123,
+      "step": 429
+    },
+    {
+      "clip_ratio/high_max": 0.001878272378235124,
+      "clip_ratio/high_mean": 0.0007897759987827158,
+      "clip_ratio/low_mean": 0.0008921427961467998,
+      "clip_ratio/low_min": 3.39891357725719e-05,
+      "clip_ratio/region_mean": 0.0016819188313093036,
+      "epoch": 4.429154518950437,
+      "grad_norm": 0.14495782554149628,
+      "learning_rate": 1e-06,
+      "loss": 0.0303,
+      "step": 430
+    },
+    {
+      "clip_ratio/high_max": 0.0023858800195739605,
+      "clip_ratio/high_mean": 0.001056421398970997,
+      "clip_ratio/low_mean": 0.0011328381842758972,
+      "clip_ratio/low_min": 7.150277724576881e-05,
+      "clip_ratio/region_mean": 0.0021892595977988094,
+      "epoch": 4.438483965014577,
+      "grad_norm": 0.18605196475982666,
+      "learning_rate": 1e-06,
+      "loss": 0.0194,
+      "step": 431
+    },
+    {
+      "clip_ratio/high_max": 0.0026346297236159444,
+      "clip_ratio/high_mean": 0.0010570780796115287,
+      "clip_ratio/low_mean": 0.0009099434473682777,
+      "clip_ratio/low_min": 1.806358341127634e-05,
+      "clip_ratio/region_mean": 0.001967021598829888,
+      "epoch": 4.447813411078717,
+      "grad_norm": 0.1287396252155304,
+      "learning_rate": 1e-06,
+      "loss": -0.0099,
+      "step": 432
+    },
+    {
+      "clip_ratio/high_max": 0.0025009670280269347,
+      "clip_ratio/high_mean": 0.0009114547774515813,
+      "clip_ratio/low_mean": 0.0008759734300838318,
+      "clip_ratio/low_min": 5.6137723731808364e-05,
+      "clip_ratio/region_mean": 0.0017874281838885508,
+      "epoch": 4.457142857142857,
+      "grad_norm": 0.12383487075567245,
+      "learning_rate": 1e-06,
+      "loss": 0.0088,
+      "step": 433
+    },
+    {
+      "clip_ratio/high_max": 0.0027183496567886323,
+      "clip_ratio/high_mean": 0.0012200498313177377,
+      "clip_ratio/low_mean": 0.0010797456052387133,
+      "clip_ratio/low_min": 6.000670327921398e-05,
+      "clip_ratio/region_mean": 0.002299795414728578,
+      "epoch": 4.466472303206997,
+      "grad_norm": 0.1346307098865509,
+      "learning_rate": 1e-06,
+      "loss": -0.0343,
+      "step": 434
+    },
+    {
+      "clip_ratio/high_max": 0.0023476571586797945,
+      "clip_ratio/high_mean": 0.0009690739752841182,
+      "clip_ratio/low_mean": 0.0009940544059645617,
+      "clip_ratio/low_min": 1.4269406165112741e-05,
+      "clip_ratio/region_mean": 0.001963128408533521,
+      "epoch": 4.475801749271137,
+      "grad_norm": 0.11996542662382126,
+      "learning_rate": 1e-06,
+      "loss": -0.0017,
+      "step": 435
+    },
+    {
+      "clip_ratio/high_max": 0.002613077071146108,
+      "clip_ratio/high_mean": 0.0009575607000442687,
+      "clip_ratio/low_mean": 0.0010433072020532563,
+      "clip_ratio/low_min": 2.778395173663739e-05,
+      "clip_ratio/region_mean": 0.002000867942115292,
+      "epoch": 4.485131195335277,
+      "grad_norm": 0.1359536349773407,
+      "learning_rate": 1e-06,
+      "loss": 0.0155,
+      "step": 436
+    },
+    {
+      "clip_ratio/high_max": 0.0025177682618959807,
+      "clip_ratio/high_mean": 0.001103411090298323,
+      "clip_ratio/low_mean": 0.0010170936766371597,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.002120504781487398,
+      "epoch": 4.494460641399417,
+      "grad_norm": 0.1333197057247162,
+      "learning_rate": 1e-06,
+      "loss": 0.0116,
+      "step": 437
+    },
+    {
+      "clip_ratio/high_max": 0.002562261142884381,
+      "clip_ratio/high_mean": 0.0010776235758385155,
+      "clip_ratio/low_mean": 0.000865392614286975,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001943016228324268,
+      "epoch": 4.503790087463557,
+      "grad_norm": 0.11565488576889038,
+      "learning_rate": 1e-06,
+      "loss": -0.0177,
+      "step": 438
+    },
+    {
+      "clip_ratio/high_max": 0.0025387959540239535,
+      "clip_ratio/high_mean": 0.00101970440846344,
+      "clip_ratio/low_mean": 0.0008223590139095904,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018420633932691999,
+      "epoch": 4.513119533527696,
+      "grad_norm": 0.11977821588516235,
+      "learning_rate": 1e-06,
+      "loss": -0.0517,
+      "step": 439
+    },
+    {
+      "clip_ratio/high_max": 0.0027230137056903914,
+      "clip_ratio/high_mean": 0.0010951757521979744,
+      "clip_ratio/low_mean": 0.0008794847781246062,
+      "clip_ratio/low_min": 4.546409400063567e-05,
+      "clip_ratio/region_mean": 0.0019746605175896548,
+      "epoch": 4.522448979591837,
+      "grad_norm": 0.12791550159454346,
+      "learning_rate": 1e-06,
+      "loss": 0.0017,
+      "step": 440
+    },
+    {
+      "clip_ratio/high_max": 0.002204522388637997,
+      "clip_ratio/high_mean": 0.0009462918096687645,
+      "clip_ratio/low_mean": 0.0009360012409160845,
+      "clip_ratio/low_min": 6.7650882556336e-05,
+      "clip_ratio/region_mean": 0.0018822930651367642,
+      "epoch": 4.531778425655976,
+      "grad_norm": 0.12334809452295303,
+      "learning_rate": 1e-06,
+      "loss": 0.0424,
+      "step": 441
+    },
+    {
+      "clip_ratio/high_max": 0.0024424805378657766,
+      "clip_ratio/high_mean": 0.0010569582154857926,
+      "clip_ratio/low_mean": 0.00098892788264493,
+      "clip_ratio/low_min": 0.00017586808553460287,
+      "clip_ratio/region_mean": 0.0020458860672079027,
+      "epoch": 4.541107871720117,
+      "grad_norm": 0.1356218457221985,
+      "learning_rate": 1e-06,
+      "loss": -0.0341,
+      "step": 442
+    },
+    {
+      "clip_ratio/high_max": 0.002113778260536492,
+      "clip_ratio/high_mean": 0.000924423870856117,
+      "clip_ratio/low_mean": 0.0011204129204998026,
+      "clip_ratio/low_min": 4.631707270164043e-05,
+      "clip_ratio/region_mean": 0.0020448367940844037,
+      "epoch": 4.550437317784256,
+      "grad_norm": 0.13238206505775452,
+      "learning_rate": 1e-06,
+      "loss": 0.0154,
+      "step": 443
+    },
+    {
+      "clip_ratio/high_max": 0.00255790411029011,
+      "clip_ratio/high_mean": 0.0010612423939164728,
+      "clip_ratio/low_mean": 0.0010988476660713786,
+      "clip_ratio/low_min": 0.00015866332523728488,
+      "clip_ratio/region_mean": 0.002160090058168862,
+      "epoch": 4.559766763848397,
+      "grad_norm": 0.13470828533172607,
+      "learning_rate": 1e-06,
+      "loss": 0.0117,
+      "step": 444
+    },
+    {
+      "clip_ratio/high_max": 0.0025511191561236046,
+      "clip_ratio/high_mean": 0.0010549079743213952,
+      "clip_ratio/low_mean": 0.00101842683943687,
+      "clip_ratio/low_min": 5.866337778570596e-05,
+      "clip_ratio/region_mean": 0.002073334762826562,
+      "epoch": 4.569096209912536,
+      "grad_norm": 0.13952285051345825,
+      "learning_rate": 1e-06,
+      "loss": -0.0541,
+      "step": 445
+    },
+    {
+      "clip_ratio/high_max": 0.002576140424935147,
+      "clip_ratio/high_mean": 0.0010571445636742283,
+      "clip_ratio/low_mean": 0.0009249123468180187,
+      "clip_ratio/low_min": 6.193603894644184e-05,
+      "clip_ratio/region_mean": 0.001982056928682141,
+      "epoch": 4.578425655976677,
+      "grad_norm": 0.13626143336296082,
+      "learning_rate": 1e-06,
+      "loss": -0.0096,
+      "step": 446
+    },
+    {
+      "clip_ratio/high_max": 0.0025502821008558385,
+      "clip_ratio/high_mean": 0.0010889160766964778,
+      "clip_ratio/low_mean": 0.0011650592969090212,
+      "clip_ratio/low_min": 4.91139835503418e-05,
+      "clip_ratio/region_mean": 0.0022539753481396474,
+      "epoch": 4.587755102040816,
+      "grad_norm": 0.13684609532356262,
+      "learning_rate": 1e-06,
+      "loss": 0.034,
+      "step": 447
+    },
+    {
+      "clip_ratio/high_max": 0.002484923774318304,
+      "clip_ratio/high_mean": 0.0011634185539151076,
+      "clip_ratio/low_mean": 0.0010489358355698641,
+      "clip_ratio/low_min": 9.727147880767006e-05,
+      "clip_ratio/region_mean": 0.0022123544331407174,
+      "epoch": 4.597084548104956,
+      "grad_norm": 0.14760440587997437,
+      "learning_rate": 1e-06,
+      "loss": -0.0015,
+      "step": 448
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0335170200892857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4081.0,
+      "completions/mean_length": 668.672119140625,
+      "completions/mean_terminated_length": 549.8145141601562,
+      "completions/min_length": 5.0,
+      "completions/min_terminated_length": 5.0,
+      "epoch": 4.606413994169096,
+      "grad_norm": 0.12598927319049835,
+      "learning_rate": 1e-06,
+      "loss": -0.0079,
+      "num_tokens": 277791545.0,
+      "reward": 0.598946750164032,
+      "reward_std": 0.18085338175296783,
+      "rewards/simpleverify_reward/mean": 0.5989466905593872,
+      "rewards/simpleverify_reward/std": 0.4901203513145447,
+      "step": 449
+    },
+    {
+      "clip_ratio/high_max": 0.0017125042431871407,
+      "clip_ratio/high_mean": 0.0007223903448903002,
+      "clip_ratio/low_mean": 0.0006181180142448284,
+      "clip_ratio/low_min": 2.3982824131962843e-05,
+      "clip_ratio/region_mean": 0.0013405084027908742,
+      "epoch": 4.615743440233236,
+      "grad_norm": 0.12622013688087463,
+      "learning_rate": 1e-06,
+      "loss": 0.0176,
+      "step": 450
+    },
+    {
+      "clip_ratio/high_max": 0.002197796042310074,
+      "clip_ratio/high_mean": 0.0008766838600422489,
+      "clip_ratio/low_mean": 0.0004888315133939614,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013655154180014506,
+      "epoch": 4.625072886297376,
+      "grad_norm": 0.13169193267822266,
+      "learning_rate": 1e-06,
+      "loss": -0.0409,
+      "step": 451
+    },
+    {
+      "clip_ratio/high_max": 0.001849649994255742,
+      "clip_ratio/high_mean": 0.0008005681011127308,
+      "clip_ratio/low_mean": 0.000547487103176536,
+      "clip_ratio/low_min": 2.8700626899080817e-05,
+      "clip_ratio/region_mean": 0.0013480551751854364,
+      "epoch": 4.634402332361516,
+      "grad_norm": 0.1239888146519661,
+      "learning_rate": 1e-06,
+      "loss": 0.0058,
+      "step": 452
+    },
+    {
+      "clip_ratio/high_max": 0.0020362655450298917,
+      "clip_ratio/high_mean": 0.0008718002318346407,
+      "clip_ratio/low_mean": 0.0006089451840125548,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014807453808316495,
+      "epoch": 4.643731778425656,
+      "grad_norm": 0.1250065714120865,
+      "learning_rate": 1e-06,
+      "loss": -0.0606,
+      "step": 453
+    },
+    {
+      "clip_ratio/high_max": 0.002317130725714378,
+      "clip_ratio/high_mean": 0.0008930037583922967,
+      "clip_ratio/low_mean": 0.0005384467485782807,
+      "clip_ratio/low_min": 5.730386692448519e-05,
+      "clip_ratio/region_mean": 0.0014314504660433158,
+      "epoch": 4.653061224489796,
+      "grad_norm": 0.12810203433036804,
+      "learning_rate": 1e-06,
+      "loss": -0.0191,
+      "step": 454
+    },
+    {
+      "clip_ratio/high_max": 0.0019615373894339427,
+      "clip_ratio/high_mean": 0.0008948678896558704,
+      "clip_ratio/low_mean": 0.0007931599084258778,
+      "clip_ratio/low_min": 2.9082890250720084e-05,
+      "clip_ratio/region_mean": 0.0016880278126336634,
+      "epoch": 4.662390670553936,
+      "grad_norm": 0.12833471596240997,
+      "learning_rate": 1e-06,
+      "loss": -0.0417,
+      "step": 455
+    },
+    {
+      "clip_ratio/high_max": 0.002077327713777777,
+      "clip_ratio/high_mean": 0.0008009091206986341,
+      "clip_ratio/low_mean": 0.0008868475961207878,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001687756710452959,
+      "epoch": 4.671720116618076,
+      "grad_norm": 0.12376552075147629,
+      "learning_rate": 1e-06,
+      "loss": -0.0187,
+      "step": 456
+    },
+    {
+      "clip_ratio/high_max": 0.002086305736156646,
+      "clip_ratio/high_mean": 0.000859870402564411,
+      "clip_ratio/low_mean": 0.0008675882581883343,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017274586898565758,
+      "epoch": 4.681049562682215,
+      "grad_norm": 0.12905630469322205,
+      "learning_rate": 1e-06,
+      "loss": -0.0077,
+      "step": 457
+    },
+    {
+      "clip_ratio/high_max": 0.001968920205399627,
+      "clip_ratio/high_mean": 0.0007975597563927295,
+      "clip_ratio/low_mean": 0.0008436562675342429,
+      "clip_ratio/low_min": 7.116145025065634e-05,
+      "clip_ratio/region_mean": 0.0016412159930041526,
+      "epoch": 4.690379008746356,
+      "grad_norm": 0.1319509744644165,
+      "learning_rate": 1e-06,
+      "loss": 0.0173,
+      "step": 458
+    },
+    {
+      "clip_ratio/high_max": 0.002110439232637873,
+      "clip_ratio/high_mean": 0.0009003827472042758,
+      "clip_ratio/low_mean": 0.0008657902708364418,
+      "clip_ratio/low_min": 3.33158886860474e-05,
+      "clip_ratio/region_mean": 0.0017661730453255586,
+      "epoch": 4.699708454810495,
+      "grad_norm": 0.12854984402656555,
+      "learning_rate": 1e-06,
+      "loss": 0.0002,
+      "step": 459
+    },
+    {
+      "clip_ratio/high_max": 0.002188599086366594,
+      "clip_ratio/high_mean": 0.0008464736656605965,
+      "clip_ratio/low_mean": 0.000951001449720934,
+      "clip_ratio/low_min": 0.00011905482642760035,
+      "clip_ratio/region_mean": 0.0017974751317524351,
+      "epoch": 4.709037900874636,
+      "grad_norm": 0.1228955015540123,
+      "learning_rate": 1e-06,
+      "loss": 0.0206,
+      "step": 460
+    },
+    {
+      "clip_ratio/high_max": 0.002151340391719714,
+      "clip_ratio/high_mean": 0.0008006596581253689,
+      "clip_ratio/low_mean": 0.0008776649119681679,
+      "clip_ratio/low_min": 0.00011294292562524788,
+      "clip_ratio/region_mean": 0.0016783245373517275,
+      "epoch": 4.718367346938775,
+      "grad_norm": 0.1404651403427124,
+      "learning_rate": 1e-06,
+      "loss": 0.0247,
+      "step": 461
+    },
+    {
+      "clip_ratio/high_max": 0.0021559811357292347,
+      "clip_ratio/high_mean": 0.0008401388895435957,
+      "clip_ratio/low_mean": 0.0007999353802006226,
+      "clip_ratio/low_min": 2.0051331375725567e-05,
+      "clip_ratio/region_mean": 0.0016400742897531018,
+      "epoch": 4.727696793002916,
+      "grad_norm": 0.12330736964941025,
+      "learning_rate": 1e-06,
+      "loss": -0.0252,
+      "step": 462
+    },
+    {
+      "clip_ratio/high_max": 0.002180338080506772,
+      "clip_ratio/high_mean": 0.0010245344492432196,
+      "clip_ratio/low_mean": 0.0010881009948207065,
+      "clip_ratio/low_min": 3.4293552744202316e-05,
+      "clip_ratio/region_mean": 0.0021126354404259473,
+      "epoch": 4.737026239067055,
+      "grad_norm": 0.14122845232486725,
+      "learning_rate": 1e-06,
+      "loss": -0.0143,
+      "step": 463
+    },
+    {
+      "clip_ratio/high_max": 0.001972054633370135,
+      "clip_ratio/high_mean": 0.0007304179962375201,
+      "clip_ratio/low_mean": 0.0009653888482716866,
+      "clip_ratio/low_min": 5.0627086238819174e-05,
+      "clip_ratio/region_mean": 0.0016958068299572915,
+      "epoch": 4.746355685131196,
+      "grad_norm": 0.12829378247261047,
+      "learning_rate": 1e-06,
+      "loss": 0.0416,
+      "step": 464
+    },
+    {
+      "clip_ratio/high_max": 0.002340914521482773,
+      "clip_ratio/high_mean": 0.0009467333129578037,
+      "clip_ratio/low_mean": 0.001042510961269727,
+      "clip_ratio/low_min": 0.00013667773419001605,
+      "clip_ratio/region_mean": 0.0019892442505806684,
+      "epoch": 4.755685131195335,
+      "grad_norm": 0.12634426355361938,
+      "learning_rate": 1e-06,
+      "loss": 0.0085,
+      "step": 465
+    },
+    {
+      "clip_ratio/high_max": 0.0024798499362077564,
+      "clip_ratio/high_mean": 0.0008589172230131226,
+      "clip_ratio/low_mean": 0.0008599830016464693,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017189002537634224,
+      "epoch": 4.765014577259475,
+      "grad_norm": 0.1335812211036682,
+      "learning_rate": 1e-06,
+      "loss": 0.0035,
+      "step": 466
+    },
+    {
+      "clip_ratio/high_max": 0.0020122003843425773,
+      "clip_ratio/high_mean": 0.0008412773731834022,
+      "clip_ratio/low_mean": 0.001073557401468861,
+      "clip_ratio/low_min": 5.362647607398685e-05,
+      "clip_ratio/region_mean": 0.0019148348001181148,
+      "epoch": 4.774344023323615,
+      "grad_norm": 0.13003897666931152,
+      "learning_rate": 1e-06,
+      "loss": -0.0069,
+      "step": 467
+    },
+    {
+      "clip_ratio/high_max": 0.0024020030941755977,
+      "clip_ratio/high_mean": 0.0009109533384616952,
+      "clip_ratio/low_mean": 0.0010316131920262706,
+      "clip_ratio/low_min": 9.37678760237759e-05,
+      "clip_ratio/region_mean": 0.001942566508660093,
+      "epoch": 4.783673469387755,
+      "grad_norm": 0.24173766374588013,
+      "learning_rate": 1e-06,
+      "loss": 0.0193,
+      "step": 468
+    },
+    {
+      "clip_ratio/high_max": 0.002175915622501634,
+      "clip_ratio/high_mean": 0.0008120271522784606,
+      "clip_ratio/low_mean": 0.0011344612612447236,
+      "clip_ratio/low_min": 0.00013007072539039655,
+      "clip_ratio/region_mean": 0.0019464884753688239,
+      "epoch": 4.793002915451895,
+      "grad_norm": 0.12008558213710785,
+      "learning_rate": 1e-06,
+      "loss": 0.0567,
+      "step": 469
+    },
+    {
+      "clip_ratio/high_max": 0.0027870368503499776,
+      "clip_ratio/high_mean": 0.0010877357472054427,
+      "clip_ratio/low_mean": 0.0009601725487300428,
+      "clip_ratio/low_min": 2.267368108732626e-05,
+      "clip_ratio/region_mean": 0.002047908303211443,
+      "epoch": 4.802332361516035,
+      "grad_norm": 0.13638655841350555,
+      "learning_rate": 1e-06,
+      "loss": -0.0196,
+      "step": 470
+    },
+    {
+      "clip_ratio/high_max": 0.0022125025425339118,
+      "clip_ratio/high_mean": 0.000991207820334239,
+      "clip_ratio/low_mean": 0.0009885704912449,
+      "clip_ratio/low_min": 2.6331716071581468e-05,
+      "clip_ratio/region_mean": 0.001979778360691853,
+      "epoch": 4.811661807580175,
+      "grad_norm": 0.1521136313676834,
+      "learning_rate": 1e-06,
+      "loss": -0.0188,
+      "step": 471
+    },
+    {
+      "clip_ratio/high_max": 0.002700241580896545,
+      "clip_ratio/high_mean": 0.00097161745725316,
+      "clip_ratio/low_mean": 0.0010361361401010072,
+      "clip_ratio/low_min": 8.029135642573237e-05,
+      "clip_ratio/region_mean": 0.0020077536028111354,
+      "epoch": 4.820991253644315,
+      "grad_norm": 0.12461855262517929,
+      "learning_rate": 1e-06,
+      "loss": 0.0036,
+      "step": 472
+    },
+    {
+      "clip_ratio/high_max": 0.0024424837101832964,
+      "clip_ratio/high_mean": 0.0008810730960249202,
+      "clip_ratio/low_mean": 0.0010888123833865393,
+      "clip_ratio/low_min": 9.56364983721869e-05,
+      "clip_ratio/region_mean": 0.0019698855030583218,
+      "epoch": 4.830320699708455,
+      "grad_norm": 0.12171190232038498,
+      "learning_rate": 1e-06,
+      "loss": 0.0292,
+      "step": 473
+    },
+    {
+      "clip_ratio/high_max": 0.0026358924151281826,
+      "clip_ratio/high_mean": 0.0009962538479157956,
+      "clip_ratio/low_mean": 0.0009987189896492055,
+      "clip_ratio/low_min": 0.00012843935201090062,
+      "clip_ratio/region_mean": 0.0019949728448409587,
+      "epoch": 4.839650145772595,
+      "grad_norm": 0.1612969934940338,
+      "learning_rate": 1e-06,
+      "loss": 0.018,
+      "step": 474
+    },
+    {
+      "clip_ratio/high_max": 0.002331630268599838,
+      "clip_ratio/high_mean": 0.0009881180340016726,
+      "clip_ratio/low_mean": 0.0009393481104780221,
+      "clip_ratio/low_min": 6.486190250143409e-05,
+      "clip_ratio/region_mean": 0.0019274661462986842,
+      "epoch": 4.848979591836734,
+      "grad_norm": 0.12759144604206085,
+      "learning_rate": 1e-06,
+      "loss": -0.0316,
+      "step": 475
+    },
+    {
+      "clip_ratio/high_max": 0.002235203282907605,
+      "clip_ratio/high_mean": 0.0009293396842622315,
+      "clip_ratio/low_mean": 0.000855879417485994,
+      "clip_ratio/low_min": 4.241055648890324e-05,
+      "clip_ratio/region_mean": 0.001785219086741563,
+      "epoch": 4.858309037900875,
+      "grad_norm": 0.12324231117963791,
+      "learning_rate": 1e-06,
+      "loss": -0.018,
+      "step": 476
+    },
+    {
+      "clip_ratio/high_max": 0.0026894152324530296,
+      "clip_ratio/high_mean": 0.0010632195844664238,
+      "clip_ratio/low_mean": 0.0009725290656206198,
+      "clip_ratio/low_min": 4.775536945089698e-05,
+      "clip_ratio/region_mean": 0.0020357486355351284,
+      "epoch": 4.867638483965014,
+      "grad_norm": 0.13588637113571167,
+      "learning_rate": 1e-06,
+      "loss": -0.0654,
+      "step": 477
+    },
+    {
+      "clip_ratio/high_max": 0.002452754713885952,
+      "clip_ratio/high_mean": 0.0009118791931541637,
+      "clip_ratio/low_mean": 0.0009465403363719815,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018584195204311982,
+      "epoch": 4.876967930029155,
+      "grad_norm": 0.1308816522359848,
+      "learning_rate": 1e-06,
+      "loss": 0.0089,
+      "step": 478
+    },
+    {
+      "clip_ratio/high_max": 0.002619025945023168,
+      "clip_ratio/high_mean": 0.0010607020703901071,
+      "clip_ratio/low_mean": 0.001078895766113419,
+      "clip_ratio/low_min": 6.010681318002753e-05,
+      "clip_ratio/region_mean": 0.002139597861969378,
+      "epoch": 4.886297376093294,
+      "grad_norm": 0.15502747893333435,
+      "learning_rate": 1e-06,
+      "loss": 0.0054,
+      "step": 479
+    },
+    {
+      "clip_ratio/high_max": 0.002136282102583209,
+      "clip_ratio/high_mean": 0.0008923701443563914,
+      "clip_ratio/low_mean": 0.0011423797222960275,
+      "clip_ratio/low_min": 9.303533170168521e-05,
+      "clip_ratio/region_mean": 0.0020347498430055566,
+      "epoch": 4.895626822157435,
+      "grad_norm": 0.11837761849164963,
+      "learning_rate": 1e-06,
+      "loss": 0.0086,
+      "step": 480
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0362723214285714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4086.0,
+      "completions/mean_length": 682.9603271484375,
+      "completions/mean_terminated_length": 554.501953125,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 5.0093294460641395,
+      "grad_norm": 0.13240015506744385,
+      "learning_rate": 1e-06,
+      "loss": 0.0043,
+      "num_tokens": 295939351.0,
+      "reward": 0.6090611219406128,
+      "reward_std": 0.18000535666942596,
+      "rewards/simpleverify_reward/mean": 0.6090611219406128,
+      "rewards/simpleverify_reward/std": 0.4879692494869232,
+      "step": 481
+    },
+    {
+      "clip_ratio/high_max": 0.0020576376482495107,
+      "clip_ratio/high_mean": 0.0008079357867245562,
+      "clip_ratio/low_mean": 0.0005491220399562735,
+      "clip_ratio/low_min": 2.6662172786018346e-05,
+      "clip_ratio/region_mean": 0.0013570578557846602,
+      "epoch": 5.01865889212828,
+      "grad_norm": 0.1365615427494049,
+      "learning_rate": 1e-06,
+      "loss": -0.0131,
+      "step": 482
+    },
+    {
+      "clip_ratio/high_max": 0.002011826218222268,
+      "clip_ratio/high_mean": 0.0008923103523557074,
+      "clip_ratio/low_mean": 0.0006162944546304061,
+      "clip_ratio/low_min": 2.3742658413539175e-05,
+      "clip_ratio/region_mean": 0.001508604793343693,
+      "epoch": 5.0279883381924195,
+      "grad_norm": 0.13470865786075592,
+      "learning_rate": 1e-06,
+      "loss": 0.0039,
+      "step": 483
+    },
+    {
+      "clip_ratio/high_max": 0.002328693233721424,
+      "clip_ratio/high_mean": 0.0010221383945463458,
+      "clip_ratio/low_mean": 0.0005530897924472811,
+      "clip_ratio/low_min": 4.729193460661918e-05,
+      "clip_ratio/region_mean": 0.0015752281797176693,
+      "epoch": 5.03731778425656,
+      "grad_norm": 0.14696210622787476,
+      "learning_rate": 1e-06,
+      "loss": -0.0645,
+      "step": 484
+    },
+    {
+      "clip_ratio/high_max": 0.0020421258959686384,
+      "clip_ratio/high_mean": 0.0007934162676974665,
+      "clip_ratio/low_mean": 0.0005984390418234398,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013918552940594964,
+      "epoch": 5.0466472303206995,
+      "grad_norm": 0.12379525601863861,
+      "learning_rate": 1e-06,
+      "loss": 0.0165,
+      "step": 485
+    },
+    {
+      "clip_ratio/high_max": 0.001749827843013918,
+      "clip_ratio/high_mean": 0.000792051461758092,
+      "clip_ratio/low_mean": 0.0007013388185441727,
+      "clip_ratio/low_min": 3.5201150240027346e-05,
+      "clip_ratio/region_mean": 0.0014933902530174237,
+      "epoch": 5.05597667638484,
+      "grad_norm": 0.1355205923318863,
+      "learning_rate": 1e-06,
+      "loss": -0.0002,
+      "step": 486
+    },
+    {
+      "clip_ratio/high_max": 0.0022914472137927078,
+      "clip_ratio/high_mean": 0.0009888761169349891,
+      "clip_ratio/low_mean": 0.0005872124193047057,
+      "clip_ratio/low_min": 4.3916755203099456e-05,
+      "clip_ratio/region_mean": 0.0015760885253257584,
+      "epoch": 5.0653061224489795,
+      "grad_norm": 0.15061749517917633,
+      "learning_rate": 1e-06,
+      "loss": 0.0022,
+      "step": 487
+    },
+    {
+      "clip_ratio/high_max": 0.002541314008340123,
+      "clip_ratio/high_mean": 0.0008550682805434917,
+      "clip_ratio/low_mean": 0.0008633553989056963,
+      "clip_ratio/low_min": 3.299023592262529e-05,
+      "clip_ratio/region_mean": 0.001718423667625757,
+      "epoch": 5.07463556851312,
+      "grad_norm": 0.12575991451740265,
+      "learning_rate": 1e-06,
+      "loss": 0.0186,
+      "step": 488
+    },
+    {
+      "clip_ratio/high_max": 0.0019343453641340602,
+      "clip_ratio/high_mean": 0.0008475117610942107,
+      "clip_ratio/low_mean": 0.0006883281857881229,
+      "clip_ratio/low_min": 1.659585723245982e-05,
+      "clip_ratio/region_mean": 0.001535839939606376,
+      "epoch": 5.0839650145772595,
+      "grad_norm": 0.10920840501785278,
+      "learning_rate": 1e-06,
+      "loss": -0.0217,
+      "step": 489
+    },
+    {
+      "clip_ratio/high_max": 0.001879649054899346,
+      "clip_ratio/high_mean": 0.0008659716040710919,
+      "clip_ratio/low_mean": 0.0006675041577182128,
+      "clip_ratio/low_min": 9.835097171162488e-05,
+      "clip_ratio/region_mean": 0.0015334757990785874,
+      "epoch": 5.093294460641399,
+      "grad_norm": 0.1407405585050583,
+      "learning_rate": 1e-06,
+      "loss": -0.0239,
+      "step": 490
+    },
+    {
+      "clip_ratio/high_max": 0.002007840506848879,
+      "clip_ratio/high_mean": 0.0008585395462432643,
+      "clip_ratio/low_mean": 0.0006898595765960636,
+      "clip_ratio/low_min": 1.3855021279596258e-05,
+      "clip_ratio/region_mean": 0.0015483990900975186,
+      "epoch": 5.1026239067055394,
+      "grad_norm": 0.13352954387664795,
+      "learning_rate": 1e-06,
+      "loss": 0.0022,
+      "step": 491
+    },
+    {
+      "clip_ratio/high_max": 0.0019643480045488104,
+      "clip_ratio/high_mean": 0.0008266790264315205,
+      "clip_ratio/low_mean": 0.0008368439093828783,
+      "clip_ratio/low_min": 3.501040828268742e-05,
+      "clip_ratio/region_mean": 0.0016635229330859147,
+      "epoch": 5.111953352769679,
+      "grad_norm": 0.13257625699043274,
+      "learning_rate": 1e-06,
+      "loss": -0.001,
+      "step": 492
+    },
+    {
+      "clip_ratio/high_max": 0.0018863148870877922,
+      "clip_ratio/high_mean": 0.0007869529854360735,
+      "clip_ratio/low_mean": 0.000801651411165949,
+      "clip_ratio/low_min": 1.3249947187432554e-05,
+      "clip_ratio/region_mean": 0.0015886044202488847,
+      "epoch": 5.121282798833819,
+      "grad_norm": 0.11635281145572662,
+      "learning_rate": 1e-06,
+      "loss": -0.0055,
+      "step": 493
+    },
+    {
+      "clip_ratio/high_max": 0.002032311327639036,
+      "clip_ratio/high_mean": 0.000924006675631972,
+      "clip_ratio/low_mean": 0.0007529759113822365,
+      "clip_ratio/low_min": 6.244476026040502e-05,
+      "clip_ratio/region_mean": 0.0016769826033851132,
+      "epoch": 5.130612244897959,
+      "grad_norm": 0.1377553641796112,
+      "learning_rate": 1e-06,
+      "loss": -0.0033,
+      "step": 494
+    },
+    {
+      "clip_ratio/high_max": 0.001949135312315775,
+      "clip_ratio/high_mean": 0.0009312449474236928,
+      "clip_ratio/low_mean": 0.0008835917869873811,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018148367162211798,
+      "epoch": 5.139941690962099,
+      "grad_norm": 0.13948386907577515,
+      "learning_rate": 1e-06,
+      "loss": -0.0362,
+      "step": 495
+    },
+    {
+      "clip_ratio/high_max": 0.002403590398898814,
+      "clip_ratio/high_mean": 0.0009901507073664106,
+      "clip_ratio/low_mean": 0.0007651728319615358,
+      "clip_ratio/low_min": 5.3115289119887166e-05,
+      "clip_ratio/region_mean": 0.0017553235447849147,
+      "epoch": 5.149271137026239,
+      "grad_norm": 0.12421394139528275,
+      "learning_rate": 1e-06,
+      "loss": -0.0375,
+      "step": 496
+    },
+    {
+      "clip_ratio/high_max": 0.0021713016758440062,
+      "clip_ratio/high_mean": 0.0009290425186918583,
+      "clip_ratio/low_mean": 0.0010167801592615433,
+      "clip_ratio/low_min": 6.576361283805454e-05,
+      "clip_ratio/region_mean": 0.0019458227034192532,
+      "epoch": 5.158600583090379,
+      "grad_norm": 0.12457162886857986,
+      "learning_rate": 1e-06,
+      "loss": 0.0328,
+      "step": 497
+    },
+    {
+      "clip_ratio/high_max": 0.0022624364792136475,
+      "clip_ratio/high_mean": 0.0009392213578394148,
+      "clip_ratio/low_mean": 0.001059698381141061,
+      "clip_ratio/low_min": 9.89463560472359e-05,
+      "clip_ratio/region_mean": 0.001998919771722285,
+      "epoch": 5.167930029154519,
+      "grad_norm": 0.1343548446893692,
+      "learning_rate": 1e-06,
+      "loss": 0.0023,
+      "step": 498
+    },
+    {
+      "clip_ratio/high_max": 0.0019533201193553396,
+      "clip_ratio/high_mean": 0.0008645405050629051,
+      "clip_ratio/low_mean": 0.0009914670808939263,
+      "clip_ratio/low_min": 4.188778984826058e-05,
+      "clip_ratio/region_mean": 0.0018560075986897573,
+      "epoch": 5.1772594752186585,
+      "grad_norm": 0.13913746178150177,
+      "learning_rate": 1e-06,
+      "loss": 0.0243,
+      "step": 499
+    },
+    {
+      "clip_ratio/high_max": 0.0020386755204526708,
+      "clip_ratio/high_mean": 0.0009038800017151516,
+      "clip_ratio/low_mean": 0.000914753754841513,
+      "clip_ratio/low_min": 1.6573852917645127e-05,
+      "clip_ratio/region_mean": 0.001818633776565548,
+      "epoch": 5.186588921282799,
+      "grad_norm": 0.12314088642597198,
+      "learning_rate": 1e-06,
+      "loss": 0.0028,
+      "step": 500
+    },
+    {
+      "clip_ratio/high_max": 0.0024852408678270876,
+      "clip_ratio/high_mean": 0.0010035527102445485,
+      "clip_ratio/low_mean": 0.0008701959159225225,
+      "clip_ratio/low_min": 2.7864467483595945e-05,
+      "clip_ratio/region_mean": 0.0018737486316240393,
+      "epoch": 5.1959183673469385,
+      "grad_norm": 0.1170763224363327,
+      "learning_rate": 1e-06,
+      "loss": -0.0443,
+      "step": 501
+    },
+    {
+      "clip_ratio/high_max": 0.0024521528321201913,
+      "clip_ratio/high_mean": 0.0009978024208976422,
+      "clip_ratio/low_mean": 0.0010160780257137958,
+      "clip_ratio/low_min": 6.657757512584794e-05,
+      "clip_ratio/region_mean": 0.0020138804757152684,
+      "epoch": 5.205247813411079,
+      "grad_norm": 0.1256285160779953,
+      "learning_rate": 1e-06,
+      "loss": 0.0035,
+      "step": 502
+    },
+    {
+      "clip_ratio/high_max": 0.002453805376717355,
+      "clip_ratio/high_mean": 0.001000544725684449,
+      "clip_ratio/low_mean": 0.0010675977282517124,
+      "clip_ratio/low_min": 0.00010594768718874548,
+      "clip_ratio/region_mean": 0.002068142421194352,
+      "epoch": 5.214577259475218,
+      "grad_norm": 0.12164364010095596,
+      "learning_rate": 1e-06,
+      "loss": 0.0276,
+      "step": 503
+    },
+    {
+      "clip_ratio/high_max": 0.0019454803150438238,
+      "clip_ratio/high_mean": 0.0009572258750267792,
+      "clip_ratio/low_mean": 0.0011246323301747907,
+      "clip_ratio/low_min": 7.069775529089384e-05,
+      "clip_ratio/region_mean": 0.0020818581906496547,
+      "epoch": 5.223906705539359,
+      "grad_norm": 0.124293752014637,
+      "learning_rate": 1e-06,
+      "loss": -0.0012,
+      "step": 504
+    },
+    {
+      "clip_ratio/high_max": 0.0023654099568375386,
+      "clip_ratio/high_mean": 0.0010104378598043695,
+      "clip_ratio/low_mean": 0.000990252610790776,
+      "clip_ratio/low_min": 7.599846685479861e-05,
+      "clip_ratio/region_mean": 0.002000690459681209,
+      "epoch": 5.233236151603498,
+      "grad_norm": 0.11535981297492981,
+      "learning_rate": 1e-06,
+      "loss": 0.0119,
+      "step": 505
+    },
+    {
+      "clip_ratio/high_max": 0.002256822554045357,
+      "clip_ratio/high_mean": 0.0010653167046257295,
+      "clip_ratio/low_mean": 0.0009366795766254654,
+      "clip_ratio/low_min": 5.9150375818717293e-05,
+      "clip_ratio/region_mean": 0.0020019962830701843,
+      "epoch": 5.242565597667639,
+      "grad_norm": 0.14496639370918274,
+      "learning_rate": 1e-06,
+      "loss": -0.0304,
+      "step": 506
+    },
+    {
+      "clip_ratio/high_max": 0.0024256028191302903,
+      "clip_ratio/high_mean": 0.0010171714257012354,
+      "clip_ratio/low_mean": 0.0010023805261880625,
+      "clip_ratio/low_min": 7.545882908743806e-05,
+      "clip_ratio/region_mean": 0.002019551939156372,
+      "epoch": 5.251895043731778,
+      "grad_norm": 0.11995477974414825,
+      "learning_rate": 1e-06,
+      "loss": -0.003,
+      "step": 507
+    },
+    {
+      "clip_ratio/high_max": 0.002414313275949098,
+      "clip_ratio/high_mean": 0.0010286249853379559,
+      "clip_ratio/low_mean": 0.001206538549013203,
+      "clip_ratio/low_min": 7.811899013177026e-05,
+      "clip_ratio/region_mean": 0.002235163548903074,
+      "epoch": 5.261224489795918,
+      "grad_norm": 0.13626430928707123,
+      "learning_rate": 1e-06,
+      "loss": 0.0301,
+      "step": 508
+    },
+    {
+      "clip_ratio/high_max": 0.0028657487346208654,
+      "clip_ratio/high_mean": 0.001197317789774388,
+      "clip_ratio/low_mean": 0.0010170679342991207,
+      "clip_ratio/low_min": 3.202979223715374e-05,
+      "clip_ratio/region_mean": 0.0022143856695038266,
+      "epoch": 5.270553935860058,
+      "grad_norm": 0.13054902851581573,
+      "learning_rate": 1e-06,
+      "loss": -0.0248,
+      "step": 509
+    },
+    {
+      "clip_ratio/high_max": 0.0024697585904505104,
+      "clip_ratio/high_mean": 0.0011102709831902757,
+      "clip_ratio/low_mean": 0.0008922044835344423,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020024753976031207,
+      "epoch": 5.279883381924198,
+      "grad_norm": 0.13530661165714264,
+      "learning_rate": 1e-06,
+      "loss": -0.0482,
+      "step": 510
+    },
+    {
+      "clip_ratio/high_max": 0.002492923813406378,
+      "clip_ratio/high_mean": 0.0010445061834616354,
+      "clip_ratio/low_mean": 0.00101987049310992,
+      "clip_ratio/low_min": 6.257984477997525e-05,
+      "clip_ratio/region_mean": 0.002064376720227301,
+      "epoch": 5.289212827988338,
+      "grad_norm": 0.13626393675804138,
+      "learning_rate": 1e-06,
+      "loss": -0.0071,
+      "step": 511
+    },
+    {
+      "clip_ratio/high_max": 0.00222903109897743,
+      "clip_ratio/high_mean": 0.0010143373947357759,
+      "clip_ratio/low_mean": 0.00103969967312878,
+      "clip_ratio/low_min": 3.111000478384085e-05,
+      "clip_ratio/region_mean": 0.0020540371260722168,
+      "epoch": 5.298542274052478,
+      "grad_norm": 0.12809443473815918,
+      "learning_rate": 1e-06,
+      "loss": -0.0203,
+      "step": 512
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0391322544642857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 3971.0,
+      "completions/mean_length": 688.7654418945312,
+      "completions/mean_terminated_length": 550.0025024414062,
+      "completions/min_length": 5.0,
+      "completions/min_terminated_length": 5.0,
+      "epoch": 5.307871720116618,
+      "grad_norm": 0.13560642302036285,
+      "learning_rate": 1e-06,
+      "loss": -0.012,
+      "num_tokens": 313967329.0,
+      "reward": 0.609375,
+      "reward_std": 0.175076425075531,
+      "rewards/simpleverify_reward/mean": 0.609375,
+      "rewards/simpleverify_reward/std": 0.48789897561073303,
+      "step": 513
+    },
+    {
+      "clip_ratio/high_max": 0.0017491583348601125,
+      "clip_ratio/high_mean": 0.0006491530384664657,
+      "clip_ratio/low_mean": 0.0005962008872302249,
+      "clip_ratio/low_min": 2.84956267933012e-05,
+      "clip_ratio/region_mean": 0.00124535392751568,
+      "epoch": 5.317201166180758,
+      "grad_norm": 0.12514084577560425,
+      "learning_rate": 1e-06,
+      "loss": 0.014,
+      "step": 514
+    },
+    {
+      "clip_ratio/high_max": 0.002272596779221203,
+      "clip_ratio/high_mean": 0.0009474223224970046,
+      "clip_ratio/low_mean": 0.0004419933588906133,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001389415680023376,
+      "epoch": 5.326530612244898,
+      "grad_norm": 0.1302099972963333,
+      "learning_rate": 1e-06,
+      "loss": -0.0756,
+      "step": 515
+    },
+    {
+      "clip_ratio/high_max": 0.001897988378914306,
+      "clip_ratio/high_mean": 0.0007603123631270137,
+      "clip_ratio/low_mean": 0.0006991100162849762,
+      "clip_ratio/low_min": 6.483782635768875e-05,
+      "clip_ratio/region_mean": 0.0014594223830499686,
+      "epoch": 5.335860058309038,
+      "grad_norm": 0.13479861617088318,
+      "learning_rate": 1e-06,
+      "loss": -0.01,
+      "step": 516
+    },
+    {
+      "clip_ratio/high_max": 0.0019074701631325297,
+      "clip_ratio/high_mean": 0.0007458546442649094,
+      "clip_ratio/low_mean": 0.000588814371440094,
+      "clip_ratio/low_min": 4.012657063867664e-05,
+      "clip_ratio/region_mean": 0.0013346690211619716,
+      "epoch": 5.345189504373177,
+      "grad_norm": 0.12178914994001389,
+      "learning_rate": 1e-06,
+      "loss": -0.0463,
+      "step": 517
+    },
+    {
+      "clip_ratio/high_max": 0.002265087685373146,
+      "clip_ratio/high_mean": 0.0008659941595396958,
+      "clip_ratio/low_mean": 0.0005778917366114911,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014438858852372505,
+      "epoch": 5.354518950437318,
+      "grad_norm": 0.12470357865095139,
+      "learning_rate": 1e-06,
+      "loss": -0.0124,
+      "step": 518
+    },
+    {
+      "clip_ratio/high_max": 0.002135470200300915,
+      "clip_ratio/high_mean": 0.0007226648194773588,
+      "clip_ratio/low_mean": 0.0006959315414860612,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014185963809723035,
+      "epoch": 5.363848396501457,
+      "grad_norm": 0.12351685017347336,
+      "learning_rate": 1e-06,
+      "loss": 0.015,
+      "step": 519
+    },
+    {
+      "clip_ratio/high_max": 0.0020754274737555534,
+      "clip_ratio/high_mean": 0.0008861361166054849,
+      "clip_ratio/low_mean": 0.0008542330779164331,
+      "clip_ratio/low_min": 4.0734303183853626e-05,
+      "clip_ratio/region_mean": 0.0017403691599611193,
+      "epoch": 5.373177842565598,
+      "grad_norm": 0.14237327873706818,
+      "learning_rate": 1e-06,
+      "loss": -0.0142,
+      "step": 520
+    },
+    {
+      "clip_ratio/high_max": 0.0020025275480293203,
+      "clip_ratio/high_mean": 0.0007684946249355562,
+      "clip_ratio/low_mean": 0.0007583003880426986,
+      "clip_ratio/low_min": 8.179931410268182e-05,
+      "clip_ratio/region_mean": 0.0015267949966073502,
+      "epoch": 5.382507288629737,
+      "grad_norm": 0.1756371259689331,
+      "learning_rate": 1e-06,
+      "loss": 0.04,
+      "step": 521
+    },
+    {
+      "clip_ratio/high_max": 0.00238994440951501,
+      "clip_ratio/high_mean": 0.0008855019750626525,
+      "clip_ratio/low_mean": 0.000931519203732023,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018170211551478133,
+      "epoch": 5.391836734693878,
+      "grad_norm": 0.14375469088554382,
+      "learning_rate": 1e-06,
+      "loss": -0.0052,
+      "step": 522
+    },
+    {
+      "clip_ratio/high_max": 0.002366450549743604,
+      "clip_ratio/high_mean": 0.0009654382156440988,
+      "clip_ratio/low_mean": 0.0007401532984658843,
+      "clip_ratio/low_min": 1.9266337403678335e-05,
+      "clip_ratio/region_mean": 0.0017055915122909937,
+      "epoch": 5.401166180758017,
+      "grad_norm": 0.13315987586975098,
+      "learning_rate": 1e-06,
+      "loss": -0.0612,
+      "step": 523
+    },
+    {
+      "clip_ratio/high_max": 0.002003363460971741,
+      "clip_ratio/high_mean": 0.0007940236864669714,
+      "clip_ratio/low_mean": 0.0008930124040489318,
+      "clip_ratio/low_min": 1.4381040273292456e-05,
+      "clip_ratio/region_mean": 0.0016870360777829774,
+      "epoch": 5.410495626822158,
+      "grad_norm": 0.1370738297700882,
+      "learning_rate": 1e-06,
+      "loss": 0.0271,
+      "step": 524
+    },
+    {
+      "clip_ratio/high_max": 0.0018541953249950893,
+      "clip_ratio/high_mean": 0.000729202223737957,
+      "clip_ratio/low_mean": 0.0009568829300405923,
+      "clip_ratio/low_min": 0.00011523921057232656,
+      "clip_ratio/region_mean": 0.0016860851537785493,
+      "epoch": 5.419825072886297,
+      "grad_norm": 0.1281536966562271,
+      "learning_rate": 1e-06,
+      "loss": 0.0394,
+      "step": 525
+    },
+    {
+      "clip_ratio/high_max": 0.0020467235444812104,
+      "clip_ratio/high_mean": 0.0008318013860844076,
+      "clip_ratio/low_mean": 0.0007885420172897284,
+      "clip_ratio/low_min": 5.9679015976144e-05,
+      "clip_ratio/region_mean": 0.0016203433915507048,
+      "epoch": 5.429154518950437,
+      "grad_norm": 0.11285793036222458,
+      "learning_rate": 1e-06,
+      "loss": -0.01,
+      "step": 526
+    },
+    {
+      "clip_ratio/high_max": 0.0021223134936008137,
+      "clip_ratio/high_mean": 0.0008833946394588565,
+      "clip_ratio/low_mean": 0.0008984029500425095,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017817975822254084,
+      "epoch": 5.438483965014577,
+      "grad_norm": 0.12460539489984512,
+      "learning_rate": 1e-06,
+      "loss": 0.0105,
+      "step": 527
+    },
+    {
+      "clip_ratio/high_max": 0.001999462958337972,
+      "clip_ratio/high_mean": 0.0008954146323958412,
+      "clip_ratio/low_mean": 0.0009278881589125376,
+      "clip_ratio/low_min": 2.8850068702013232e-05,
+      "clip_ratio/region_mean": 0.001823302738557686,
+      "epoch": 5.447813411078717,
+      "grad_norm": 0.13345134258270264,
+      "learning_rate": 1e-06,
+      "loss": -0.0419,
+      "step": 528
+    },
+    {
+      "clip_ratio/high_max": 0.002199254078732338,
+      "clip_ratio/high_mean": 0.0009600434696039883,
+      "clip_ratio/low_mean": 0.0008709008607183932,
+      "clip_ratio/low_min": 1.1042402547900565e-05,
+      "clip_ratio/region_mean": 0.0018309443039470352,
+      "epoch": 5.457142857142857,
+      "grad_norm": 0.13948243856430054,
+      "learning_rate": 1e-06,
+      "loss": -0.0244,
+      "step": 529
+    },
+    {
+      "clip_ratio/high_max": 0.002275763708894374,
+      "clip_ratio/high_mean": 0.0009221295531460783,
+      "clip_ratio/low_mean": 0.0006748406431142939,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015969702290021814,
+      "epoch": 5.466472303206997,
+      "grad_norm": 0.11956123262643814,
+      "learning_rate": 1e-06,
+      "loss": -0.0481,
+      "step": 530
+    },
+    {
+      "clip_ratio/high_max": 0.0022184816043591127,
+      "clip_ratio/high_mean": 0.000888157905137632,
+      "clip_ratio/low_mean": 0.0007502004336856771,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016383583315473516,
+      "epoch": 5.475801749271137,
+      "grad_norm": 0.11030435562133789,
+      "learning_rate": 1e-06,
+      "loss": -0.0435,
+      "step": 531
+    },
+    {
+      "clip_ratio/high_max": 0.0021036492253188044,
+      "clip_ratio/high_mean": 0.000867285640197224,
+      "clip_ratio/low_mean": 0.0007736313045825227,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016409169911639765,
+      "epoch": 5.485131195335277,
+      "grad_norm": 0.11707475781440735,
+      "learning_rate": 1e-06,
+      "loss": -0.0212,
+      "step": 532
+    },
+    {
+      "clip_ratio/high_max": 0.0020848766871495172,
+      "clip_ratio/high_mean": 0.000894044766027946,
+      "clip_ratio/low_mean": 0.0009081581538339378,
+      "clip_ratio/low_min": 9.204614980262704e-05,
+      "clip_ratio/region_mean": 0.0018022029689745978,
+      "epoch": 5.494460641399417,
+      "grad_norm": 0.14132405817508698,
+      "learning_rate": 1e-06,
+      "loss": -0.0265,
+      "step": 533
+    },
+    {
+      "clip_ratio/high_max": 0.002804819858283736,
+      "clip_ratio/high_mean": 0.001033475884469226,
+      "clip_ratio/low_mean": 0.0009524611705273855,
+      "clip_ratio/low_min": 1.6715699530323036e-05,
+      "clip_ratio/region_mean": 0.0019859370586345904,
+      "epoch": 5.503790087463557,
+      "grad_norm": 0.13796620070934296,
+      "learning_rate": 1e-06,
+      "loss": 0.0003,
+      "step": 534
+    },
+    {
+      "clip_ratio/high_max": 0.0021225897180556785,
+      "clip_ratio/high_mean": 0.0008588396703999024,
+      "clip_ratio/low_mean": 0.0008971254355856217,
+      "clip_ratio/low_min": 4.1293245885754004e-05,
+      "clip_ratio/region_mean": 0.0017559650805196725,
+      "epoch": 5.513119533527696,
+      "grad_norm": 0.12176398932933807,
+      "learning_rate": 1e-06,
+      "loss": 0.0177,
+      "step": 535
+    },
+    {
+      "clip_ratio/high_max": 0.0025111865616054274,
+      "clip_ratio/high_mean": 0.0010823341781360796,
+      "clip_ratio/low_mean": 0.0007295301929843845,
+      "clip_ratio/low_min": 4.5843520638300106e-05,
+      "clip_ratio/region_mean": 0.0018118642874469515,
+      "epoch": 5.522448979591837,
+      "grad_norm": 0.12989932298660278,
+      "learning_rate": 1e-06,
+      "loss": -0.0079,
+      "step": 536
+    },
+    {
+      "clip_ratio/high_max": 0.0026546081717242487,
+      "clip_ratio/high_mean": 0.001037176316458499,
+      "clip_ratio/low_mean": 0.0009605788654880598,
+      "clip_ratio/low_min": 1.3205155482864939e-05,
+      "clip_ratio/region_mean": 0.001997755200136453,
+      "epoch": 5.531778425655976,
+      "grad_norm": 0.13197563588619232,
+      "learning_rate": 1e-06,
+      "loss": -0.004,
+      "step": 537
+    },
+    {
+      "clip_ratio/high_max": 0.0022418158405344,
+      "clip_ratio/high_mean": 0.001036331923387479,
+      "clip_ratio/low_mean": 0.0008999032943393104,
+      "clip_ratio/low_min": 2.2045855075703003e-05,
+      "clip_ratio/region_mean": 0.001936235225002747,
+      "epoch": 5.541107871720117,
+      "grad_norm": 0.14139501750469208,
+      "learning_rate": 1e-06,
+      "loss": -0.0153,
+      "step": 538
+    },
+    {
+      "clip_ratio/high_max": 0.0024613587593194097,
+      "clip_ratio/high_mean": 0.0010359186217101524,
+      "clip_ratio/low_mean": 0.0009398598567713634,
+      "clip_ratio/low_min": 4.984726183465682e-05,
+      "clip_ratio/region_mean": 0.0019757784648390952,
+      "epoch": 5.550437317784256,
+      "grad_norm": 0.12041354924440384,
+      "learning_rate": 1e-06,
+      "loss": -0.0141,
+      "step": 539
+    },
+    {
+      "clip_ratio/high_max": 0.0025265402873628773,
+      "clip_ratio/high_mean": 0.0010248810413031606,
+      "clip_ratio/low_mean": 0.0009854290838120505,
+      "clip_ratio/low_min": 3.78825261577731e-05,
+      "clip_ratio/region_mean": 0.0020103101123822853,
+      "epoch": 5.559766763848397,
+      "grad_norm": 0.12725260853767395,
+      "learning_rate": 1e-06,
+      "loss": -0.0179,
+      "step": 540
+    },
+    {
+      "clip_ratio/high_max": 0.0021822889466420747,
+      "clip_ratio/high_mean": 0.0008952896350820083,
+      "clip_ratio/low_mean": 0.000962873557000421,
+      "clip_ratio/low_min": 8.405642256548163e-05,
+      "clip_ratio/region_mean": 0.0018581632321001962,
+      "epoch": 5.569096209912536,
+      "grad_norm": 0.13308624923229218,
+      "learning_rate": 1e-06,
+      "loss": -0.007,
+      "step": 541
+    },
+    {
+      "clip_ratio/high_max": 0.002624096829094924,
+      "clip_ratio/high_mean": 0.0011093954199168365,
+      "clip_ratio/low_mean": 0.001107957019485184,
+      "clip_ratio/low_min": 6.319619751593564e-05,
+      "clip_ratio/region_mean": 0.00221735244122101,
+      "epoch": 5.578425655976677,
+      "grad_norm": 0.13106843829154968,
+      "learning_rate": 1e-06,
+      "loss": 0.0057,
+      "step": 542
+    },
+    {
+      "clip_ratio/high_max": 0.002117255084158387,
+      "clip_ratio/high_mean": 0.0009360337826365139,
+      "clip_ratio/low_mean": 0.001085181789676426,
+      "clip_ratio/low_min": 0.00013736536129727028,
+      "clip_ratio/region_mean": 0.0020212155213812366,
+      "epoch": 5.587755102040816,
+      "grad_norm": 0.1337672919034958,
+      "learning_rate": 1e-06,
+      "loss": 0.0162,
+      "step": 543
+    },
+    {
+      "clip_ratio/high_max": 0.0020849234369961778,
+      "clip_ratio/high_mean": 0.0009420119222340873,
+      "clip_ratio/low_mean": 0.0009909714026434813,
+      "clip_ratio/low_min": 1.4351320714922622e-05,
+      "clip_ratio/region_mean": 0.0019329833594383672,
+      "epoch": 5.597084548104956,
+      "grad_norm": 0.1221359446644783,
+      "learning_rate": 1e-06,
+      "loss": -0.0048,
+      "step": 544
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0350167410714286,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4092.0,
+      "completions/mean_length": 671.4923706054688,
+      "completions/mean_terminated_length": 547.2257690429688,
+      "completions/min_length": 48.0,
+      "completions/min_terminated_length": 48.0,
+      "epoch": 5.606413994169096,
+      "grad_norm": 0.13670922815799713,
+      "learning_rate": 1e-06,
+      "loss": 0.0104,
+      "num_tokens": 331985892.0,
+      "reward": 0.6315569281578064,
+      "reward_std": 0.17832709848880768,
+      "rewards/simpleverify_reward/mean": 0.6315569281578064,
+      "rewards/simpleverify_reward/std": 0.482390820980072,
+      "step": 545
+    },
+    {
+      "clip_ratio/high_max": 0.001750637726217974,
+      "clip_ratio/high_mean": 0.0007403647496175836,
+      "clip_ratio/low_mean": 0.000600640892116644,
+      "clip_ratio/low_min": 3.950626432924764e-05,
+      "clip_ratio/region_mean": 0.0013410056490101852,
+      "epoch": 5.615743440233236,
+      "grad_norm": 0.12390350550413132,
+      "learning_rate": 1e-06,
+      "loss": -0.0075,
+      "step": 546
+    },
+    {
+      "clip_ratio/high_max": 0.002036589379713405,
+      "clip_ratio/high_mean": 0.0007229789825942134,
+      "clip_ratio/low_mean": 0.0006343035311147105,
+      "clip_ratio/low_min": 4.37473881902406e-05,
+      "clip_ratio/region_mean": 0.0013572824900620617,
+      "epoch": 5.625072886297376,
+      "grad_norm": 0.1806793510913849,
+      "learning_rate": 1e-06,
+      "loss": 0.02,
+      "step": 547
+    },
+    {
+      "clip_ratio/high_max": 0.0018286255617567804,
+      "clip_ratio/high_mean": 0.0007545240860054037,
+      "clip_ratio/low_mean": 0.0005959441023151157,
+      "clip_ratio/low_min": 1.720341242616996e-05,
+      "clip_ratio/region_mean": 0.001350468177406583,
+      "epoch": 5.634402332361516,
+      "grad_norm": 0.1340532749891281,
+      "learning_rate": 1e-06,
+      "loss": -0.0345,
+      "step": 548
+    },
+    {
+      "clip_ratio/high_max": 0.0022353808672050945,
+      "clip_ratio/high_mean": 0.0010070279677165672,
+      "clip_ratio/low_mean": 0.000661815238345298,
+      "clip_ratio/low_min": 1.5034880561870523e-05,
+      "clip_ratio/region_mean": 0.0016688432078808546,
+      "epoch": 5.643731778425656,
+      "grad_norm": 0.1435471475124359,
+      "learning_rate": 1e-06,
+      "loss": -0.0288,
+      "step": 549
+    },
+    {
+      "clip_ratio/high_max": 0.002421589808363933,
+      "clip_ratio/high_mean": 0.0009676043300714809,
+      "clip_ratio/low_mean": 0.0006054889258848561,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001573093242768664,
+      "epoch": 5.653061224489796,
+      "grad_norm": 0.14605525135993958,
+      "learning_rate": 1e-06,
+      "loss": -0.0196,
+      "step": 550
+    },
+    {
+      "clip_ratio/high_max": 0.0021120113524375483,
+      "clip_ratio/high_mean": 0.0008440371475444408,
+      "clip_ratio/low_mean": 0.0006554055444212281,
+      "clip_ratio/low_min": 1.055564916896401e-05,
+      "clip_ratio/region_mean": 0.0014994427183410153,
+      "epoch": 5.662390670553936,
+      "grad_norm": 0.13467949628829956,
+      "learning_rate": 1e-06,
+      "loss": 0.008,
+      "step": 551
+    },
+    {
+      "clip_ratio/high_max": 0.0022380819427780807,
+      "clip_ratio/high_mean": 0.0009138100904237945,
+      "clip_ratio/low_mean": 0.0008065852262006956,
+      "clip_ratio/low_min": 1.268262985831825e-05,
+      "clip_ratio/region_mean": 0.0017203953102580272,
+      "epoch": 5.671720116618076,
+      "grad_norm": 0.13332700729370117,
+      "learning_rate": 1e-06,
+      "loss": -0.0056,
+      "step": 552
+    },
+    {
+      "clip_ratio/high_max": 0.0018018883165495936,
+      "clip_ratio/high_mean": 0.0008263292038463987,
+      "clip_ratio/low_mean": 0.0007220802672236459,
+      "clip_ratio/low_min": 7.978991197887808e-05,
+      "clip_ratio/region_mean": 0.0015484094692510553,
+      "epoch": 5.681049562682215,
+      "grad_norm": 0.1289679855108261,
+      "learning_rate": 1e-06,
+      "loss": -0.006,
+      "step": 553
+    },
+    {
+      "clip_ratio/high_max": 0.0019725370257219765,
+      "clip_ratio/high_mean": 0.0008009821322048083,
+      "clip_ratio/low_mean": 0.0008184621183318086,
+      "clip_ratio/low_min": 4.022441953566158e-05,
+      "clip_ratio/region_mean": 0.0016194442650885321,
+      "epoch": 5.690379008746356,
+      "grad_norm": 0.586776852607727,
+      "learning_rate": 1e-06,
+      "loss": 0.0323,
+      "step": 554
+    },
+    {
+      "clip_ratio/high_max": 0.0020085054929950275,
+      "clip_ratio/high_mean": 0.000941243240959011,
+      "clip_ratio/low_mean": 0.0007120742120605428,
+      "clip_ratio/low_min": 9.390024388267193e-06,
+      "clip_ratio/region_mean": 0.0016533174639334902,
+      "epoch": 5.699708454810495,
+      "grad_norm": 0.12782545387744904,
+      "learning_rate": 1e-06,
+      "loss": -0.0608,
+      "step": 555
+    },
+    {
+      "clip_ratio/high_max": 0.0023442203273589257,
+      "clip_ratio/high_mean": 0.0010248871185467578,
+      "clip_ratio/low_mean": 0.000857668945172918,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018825560255208984,
+      "epoch": 5.709037900874636,
+      "grad_norm": 0.12847337126731873,
+      "learning_rate": 1e-06,
+      "loss": -0.0208,
+      "step": 556
+    },
+    {
+      "clip_ratio/high_max": 0.0024622225391794927,
+      "clip_ratio/high_mean": 0.0010462429090694059,
+      "clip_ratio/low_mean": 0.0008149801797117107,
+      "clip_ratio/low_min": 8.858046385284979e-05,
+      "clip_ratio/region_mean": 0.0018612230851431377,
+      "epoch": 5.718367346938775,
+      "grad_norm": 0.12106923758983612,
+      "learning_rate": 1e-06,
+      "loss": -0.0555,
+      "step": 557
+    },
+    {
+      "clip_ratio/high_max": 0.002218135188741144,
+      "clip_ratio/high_mean": 0.000883603253896581,
+      "clip_ratio/low_mean": 0.000936070253374055,
+      "clip_ratio/low_min": 4.945598266203888e-05,
+      "clip_ratio/region_mean": 0.0018196734235971235,
+      "epoch": 5.727696793002916,
+      "grad_norm": 0.13230378925800323,
+      "learning_rate": 1e-06,
+      "loss": 0.0094,
+      "step": 558
+    },
+    {
+      "clip_ratio/high_max": 0.0022100650385254994,
+      "clip_ratio/high_mean": 0.0009353403002023697,
+      "clip_ratio/low_mean": 0.0009273219384340337,
+      "clip_ratio/low_min": 8.200657975976355e-05,
+      "clip_ratio/region_mean": 0.0018626622259034775,
+      "epoch": 5.737026239067055,
+      "grad_norm": 0.11959223449230194,
+      "learning_rate": 1e-06,
+      "loss": 0.0044,
+      "step": 559
+    },
+    {
+      "clip_ratio/high_max": 0.002289699543325696,
+      "clip_ratio/high_mean": 0.0009810059309529606,
+      "clip_ratio/low_mean": 0.0009907797975756694,
+      "clip_ratio/low_min": 3.4939223041874357e-05,
+      "clip_ratio/region_mean": 0.0019717857358045876,
+      "epoch": 5.746355685131196,
+      "grad_norm": 0.14054135978221893,
+      "learning_rate": 1e-06,
+      "loss": -0.0248,
+      "step": 560
+    },
+    {
+      "clip_ratio/high_max": 0.002019625659158919,
+      "clip_ratio/high_mean": 0.0008171865774784237,
+      "clip_ratio/low_mean": 0.0008156140029313974,
+      "clip_ratio/low_min": 9.728010627441108e-05,
+      "clip_ratio/region_mean": 0.0016328005985997152,
+      "epoch": 5.755685131195335,
+      "grad_norm": 0.13992683589458466,
+      "learning_rate": 1e-06,
+      "loss": 0.0186,
+      "step": 561
+    },
+    {
+      "clip_ratio/high_max": 0.002730626001721248,
+      "clip_ratio/high_mean": 0.001053441308613401,
+      "clip_ratio/low_mean": 0.0008964314220065717,
+      "clip_ratio/low_min": 2.4108003344736062e-05,
+      "clip_ratio/region_mean": 0.0019498727269819938,
+      "epoch": 5.765014577259475,
+      "grad_norm": 0.13415361940860748,
+      "learning_rate": 1e-06,
+      "loss": -0.0119,
+      "step": 562
+    },
+    {
+      "clip_ratio/high_max": 0.002285698297782801,
+      "clip_ratio/high_mean": 0.0009506792448519263,
+      "clip_ratio/low_mean": 0.0009029839511640603,
+      "clip_ratio/low_min": 5.8279083532397635e-05,
+      "clip_ratio/region_mean": 0.0018536631760071032,
+      "epoch": 5.774344023323615,
+      "grad_norm": 0.12575222551822662,
+      "learning_rate": 1e-06,
+      "loss": -0.0048,
+      "step": 563
+    },
+    {
+      "clip_ratio/high_max": 0.0024820152029860765,
+      "clip_ratio/high_mean": 0.0010349018048145808,
+      "clip_ratio/low_mean": 0.0010373322420491604,
+      "clip_ratio/low_min": 0.000114900558401132,
+      "clip_ratio/region_mean": 0.002072234026854858,
+      "epoch": 5.783673469387755,
+      "grad_norm": 0.1460321545600891,
+      "learning_rate": 1e-06,
+      "loss": 0.0061,
+      "step": 564
+    },
+    {
+      "clip_ratio/high_max": 0.002482082723872736,
+      "clip_ratio/high_mean": 0.0009173096768790856,
+      "clip_ratio/low_mean": 0.0009734103441587649,
+      "clip_ratio/low_min": 9.505935304332525e-05,
+      "clip_ratio/region_mean": 0.0018907200064859353,
+      "epoch": 5.793002915451895,
+      "grad_norm": 0.14576870203018188,
+      "learning_rate": 1e-06,
+      "loss": 0.0102,
+      "step": 565
+    },
+    {
+      "clip_ratio/high_max": 0.0028556299657793716,
+      "clip_ratio/high_mean": 0.0010983677311742213,
+      "clip_ratio/low_mean": 0.001033614684274653,
+      "clip_ratio/low_min": 7.411484511976596e-05,
+      "clip_ratio/region_mean": 0.0021319823936210014,
+      "epoch": 5.802332361516035,
+      "grad_norm": 0.14002689719200134,
+      "learning_rate": 1e-06,
+      "loss": -0.0308,
+      "step": 566
+    },
+    {
+      "clip_ratio/high_max": 0.0023263703478733078,
+      "clip_ratio/high_mean": 0.0009396486493642442,
+      "clip_ratio/low_mean": 0.001030469493343844,
+      "clip_ratio/low_min": 6.231813995327684e-05,
+      "clip_ratio/region_mean": 0.0019701181881828234,
+      "epoch": 5.811661807580175,
+      "grad_norm": 0.14780262112617493,
+      "learning_rate": 1e-06,
+      "loss": 0.0073,
+      "step": 567
+    },
+    {
+      "clip_ratio/high_max": 0.0026978353707818314,
+      "clip_ratio/high_mean": 0.001165849062090274,
+      "clip_ratio/low_mean": 0.0010416831864858977,
+      "clip_ratio/low_min": 1.258558222616557e-05,
+      "clip_ratio/region_mean": 0.0022075322485761717,
+      "epoch": 5.820991253644315,
+      "grad_norm": 0.12568442523479462,
+      "learning_rate": 1e-06,
+      "loss": -0.0078,
+      "step": 568
+    },
+    {
+      "clip_ratio/high_max": 0.0023234408035932574,
+      "clip_ratio/high_mean": 0.0009093883036257466,
+      "clip_ratio/low_mean": 0.001033348786222632,
+      "clip_ratio/low_min": 2.622194188006688e-05,
+      "clip_ratio/region_mean": 0.0019427370789344423,
+      "epoch": 5.830320699708455,
+      "grad_norm": 0.13785843551158905,
+      "learning_rate": 1e-06,
+      "loss": -0.0242,
+      "step": 569
+    },
+    {
+      "clip_ratio/high_max": 0.002600420546514215,
+      "clip_ratio/high_mean": 0.0010785224658320658,
+      "clip_ratio/low_mean": 0.0010606590512907133,
+      "clip_ratio/low_min": 4.420942786964588e-05,
+      "clip_ratio/region_mean": 0.0021391815316746943,
+      "epoch": 5.839650145772595,
+      "grad_norm": 0.1736195981502533,
+      "learning_rate": 1e-06,
+      "loss": 0.0189,
+      "step": 570
+    },
+    {
+      "clip_ratio/high_max": 0.002762816053291317,
+      "clip_ratio/high_mean": 0.0010711998183978721,
+      "clip_ratio/low_mean": 0.0010755806742963614,
+      "clip_ratio/low_min": 2.829975164786447e-05,
+      "clip_ratio/region_mean": 0.0021467804617714137,
+      "epoch": 5.848979591836734,
+      "grad_norm": 0.14605410397052765,
+      "learning_rate": 1e-06,
+      "loss": 0.0059,
+      "step": 571
+    },
+    {
+      "clip_ratio/high_max": 0.0026545310392975807,
+      "clip_ratio/high_mean": 0.0011149754755024333,
+      "clip_ratio/low_mean": 0.0010497763596504228,
+      "clip_ratio/low_min": 5.000800229026936e-05,
+      "clip_ratio/region_mean": 0.0021647518369718455,
+      "epoch": 5.858309037900875,
+      "grad_norm": 0.14060768485069275,
+      "learning_rate": 1e-06,
+      "loss": -0.0148,
+      "step": 572
+    },
+    {
+      "clip_ratio/high_max": 0.00251997570740059,
+      "clip_ratio/high_mean": 0.0010765289152914193,
+      "clip_ratio/low_mean": 0.001015187895973213,
+      "clip_ratio/low_min": 4.920676292385906e-05,
+      "clip_ratio/region_mean": 0.00209171681854059,
+      "epoch": 5.867638483965014,
+      "grad_norm": 0.1242566704750061,
+      "learning_rate": 1e-06,
+      "loss": -0.0047,
+      "step": 573
+    },
+    {
+      "clip_ratio/high_max": 0.0029883282113587484,
+      "clip_ratio/high_mean": 0.0011473836930235848,
+      "clip_ratio/low_mean": 0.0010441396098030964,
+      "clip_ratio/low_min": 7.687041852477705e-05,
+      "clip_ratio/region_mean": 0.0021915233082836494,
+      "epoch": 5.876967930029155,
+      "grad_norm": 0.13064825534820557,
+      "learning_rate": 1e-06,
+      "loss": -0.0122,
+      "step": 574
+    },
+    {
+      "clip_ratio/high_max": 0.0027383708220440894,
+      "clip_ratio/high_mean": 0.0011308666489640018,
+      "clip_ratio/low_mean": 0.0010081860709760804,
+      "clip_ratio/low_min": 4.4768569750885945e-05,
+      "clip_ratio/region_mean": 0.0021390527035691775,
+      "epoch": 5.886297376093294,
+      "grad_norm": 0.1567859947681427,
+      "learning_rate": 1e-06,
+      "loss": -0.0072,
+      "step": 575
+    },
+    {
+      "clip_ratio/high_max": 0.0025184598998748697,
+      "clip_ratio/high_mean": 0.001111205951019656,
+      "clip_ratio/low_mean": 0.0009063593856808438,
+      "clip_ratio/low_min": 1.1480528883112129e-05,
+      "clip_ratio/region_mean": 0.002017565300775459,
+      "epoch": 5.895626822157435,
+      "grad_norm": 0.11835873872041702,
+      "learning_rate": 1e-06,
+      "loss": -0.0757,
+      "step": 576
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0405970982142857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4014.0,
+      "completions/mean_length": 689.2316284179688,
+      "completions/mean_terminated_length": 545.0742797851562,
+      "completions/min_length": 4.0,
+      "completions/min_terminated_length": 4.0,
+      "epoch": 6.0093294460641395,
+      "grad_norm": 0.13992027938365936,
+      "learning_rate": 1e-06,
+      "loss": 0.0034,
+      "num_tokens": 349849028.0,
+      "reward": 0.6221749782562256,
+      "reward_std": 0.16985295712947845,
+      "rewards/simpleverify_reward/mean": 0.6221749186515808,
+      "rewards/simpleverify_reward/std": 0.48485201597213745,
+      "step": 577
+    },
+    {
+      "clip_ratio/high_max": 0.001839089060013066,
+      "clip_ratio/high_mean": 0.0007203435447991069,
+      "clip_ratio/low_mean": 0.0005195758740228484,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001239919427462155,
+      "epoch": 6.01865889212828,
+      "grad_norm": 0.12756891548633575,
+      "learning_rate": 1e-06,
+      "loss": -0.0036,
+      "step": 578
+    },
+    {
+      "clip_ratio/high_max": 0.0018774046475300565,
+      "clip_ratio/high_mean": 0.0007390343562292401,
+      "clip_ratio/low_mean": 0.000552712145690748,
+      "clip_ratio/low_min": 2.9911461751908064e-05,
+      "clip_ratio/region_mean": 0.0012917465101054404,
+      "epoch": 6.0279883381924195,
+      "grad_norm": 0.14087362587451935,
+      "learning_rate": 1e-06,
+      "loss": -0.0046,
+      "step": 579
+    },
+    {
+      "clip_ratio/high_max": 0.0019639048987301067,
+      "clip_ratio/high_mean": 0.0007716389627603348,
+      "clip_ratio/low_mean": 0.0006355242221616209,
+      "clip_ratio/low_min": 1.1326567801006604e-05,
+      "clip_ratio/region_mean": 0.0014071632031118497,
+      "epoch": 6.03731778425656,
+      "grad_norm": 0.12709547579288483,
+      "learning_rate": 1e-06,
+      "loss": -0.0156,
+      "step": 580
+    },
+    {
+      "clip_ratio/high_max": 0.001984716254810337,
+      "clip_ratio/high_mean": 0.0008388578226004029,
+      "clip_ratio/low_mean": 0.0005622030585072935,
+      "clip_ratio/low_min": 3.081580962316366e-05,
+      "clip_ratio/region_mean": 0.0014010608938406222,
+      "epoch": 6.0466472303206995,
+      "grad_norm": 0.12824492156505585,
+      "learning_rate": 1e-06,
+      "loss": -0.0249,
+      "step": 581
+    },
+    {
+      "clip_ratio/high_max": 0.002107550681103021,
+      "clip_ratio/high_mean": 0.0008283532870336785,
+      "clip_ratio/low_mean": 0.0006303839909378439,
+      "clip_ratio/low_min": 5.199104634812102e-05,
+      "clip_ratio/region_mean": 0.0014587372861569747,
+      "epoch": 6.05597667638484,
+      "grad_norm": 0.12018285691738129,
+      "learning_rate": 1e-06,
+      "loss": 0.0127,
+      "step": 582
+    },
+    {
+      "clip_ratio/high_max": 0.002049487200565636,
+      "clip_ratio/high_mean": 0.0008545352739020018,
+      "clip_ratio/low_mean": 0.0006267588869377505,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00148129413719289,
+      "epoch": 6.0653061224489795,
+      "grad_norm": 0.12946173548698425,
+      "learning_rate": 1e-06,
+      "loss": 0.0087,
+      "step": 583
+    },
+    {
+      "clip_ratio/high_max": 0.0018140029460482765,
+      "clip_ratio/high_mean": 0.0008485489033773774,
+      "clip_ratio/low_mean": 0.000708300263795536,
+      "clip_ratio/low_min": 3.322311749798246e-05,
+      "clip_ratio/region_mean": 0.0015568491944577545,
+      "epoch": 6.07463556851312,
+      "grad_norm": 0.13875502347946167,
+      "learning_rate": 1e-06,
+      "loss": -0.011,
+      "step": 584
+    },
+    {
+      "clip_ratio/high_max": 0.002378743221925106,
+      "clip_ratio/high_mean": 0.0010164604191231774,
+      "clip_ratio/low_mean": 0.0007099261256371392,
+      "clip_ratio/low_min": 2.7026405405194964e-05,
+      "clip_ratio/region_mean": 0.0017263864865526557,
+      "epoch": 6.0839650145772595,
+      "grad_norm": 0.15075162053108215,
+      "learning_rate": 1e-06,
+      "loss": -0.0011,
+      "step": 585
+    },
+    {
+      "clip_ratio/high_max": 0.0020389452438394073,
+      "clip_ratio/high_mean": 0.0008910112519515678,
+      "clip_ratio/low_mean": 0.0007617377887072507,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001652749087952543,
+      "epoch": 6.093294460641399,
+      "grad_norm": 0.1448982208967209,
+      "learning_rate": 1e-06,
+      "loss": 0.0191,
+      "step": 586
+    },
+    {
+      "clip_ratio/high_max": 0.002088928649754962,
+      "clip_ratio/high_mean": 0.0007960854982229648,
+      "clip_ratio/low_mean": 0.0007277964423337835,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015238819178193808,
+      "epoch": 6.1026239067055394,
+      "grad_norm": 0.13771381974220276,
+      "learning_rate": 1e-06,
+      "loss": -0.0161,
+      "step": 587
+    },
+    {
+      "clip_ratio/high_max": 0.0022339819042827003,
+      "clip_ratio/high_mean": 0.0009189747433993034,
+      "clip_ratio/low_mean": 0.000862750326632522,
+      "clip_ratio/low_min": 4.3757292587542906e-05,
+      "clip_ratio/region_mean": 0.0017817250554799102,
+      "epoch": 6.111953352769679,
+      "grad_norm": 0.15242813527584076,
+      "learning_rate": 1e-06,
+      "loss": -0.0192,
+      "step": 588
+    },
+    {
+      "clip_ratio/high_max": 0.001995936741877813,
+      "clip_ratio/high_mean": 0.0008444324212177889,
+      "clip_ratio/low_mean": 0.000831053248475655,
+      "clip_ratio/low_min": 7.263965289894259e-05,
+      "clip_ratio/region_mean": 0.0016754856987972744,
+      "epoch": 6.121282798833819,
+      "grad_norm": 0.12993387877941132,
+      "learning_rate": 1e-06,
+      "loss": 0.003,
+      "step": 589
+    },
+    {
+      "clip_ratio/high_max": 0.002148512583517004,
+      "clip_ratio/high_mean": 0.0009271651069866493,
+      "clip_ratio/low_mean": 0.0007392742700176314,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016664393551764078,
+      "epoch": 6.130612244897959,
+      "grad_norm": 0.13784319162368774,
+      "learning_rate": 1e-06,
+      "loss": -0.0034,
+      "step": 590
+    },
+    {
+      "clip_ratio/high_max": 0.002533342529204674,
+      "clip_ratio/high_mean": 0.000949246366872103,
+      "clip_ratio/low_mean": 0.0007860487457946874,
+      "clip_ratio/low_min": 2.5911807824741118e-05,
+      "clip_ratio/region_mean": 0.0017352951253997162,
+      "epoch": 6.139941690962099,
+      "grad_norm": 0.16066806018352509,
+      "learning_rate": 1e-06,
+      "loss": -0.0404,
+      "step": 591
+    },
+    {
+      "clip_ratio/high_max": 0.0023953273412189446,
+      "clip_ratio/high_mean": 0.0009418380468559917,
+      "clip_ratio/low_mean": 0.000735820411136956,
+      "clip_ratio/low_min": 2.071594281005673e-05,
+      "clip_ratio/region_mean": 0.0016776584525359794,
+      "epoch": 6.149271137026239,
+      "grad_norm": 0.12303200364112854,
+      "learning_rate": 1e-06,
+      "loss": -0.0068,
+      "step": 592
+    },
+    {
+      "clip_ratio/high_max": 0.002475280132784974,
+      "clip_ratio/high_mean": 0.0009786700738914078,
+      "clip_ratio/low_mean": 0.0008750699344091117,
+      "clip_ratio/low_min": 1.9201228496967815e-05,
+      "clip_ratio/region_mean": 0.0018537400246714242,
+      "epoch": 6.158600583090379,
+      "grad_norm": 0.1335153728723526,
+      "learning_rate": 1e-06,
+      "loss": -0.0336,
+      "step": 593
+    },
+    {
+      "clip_ratio/high_max": 0.0021944416002952494,
+      "clip_ratio/high_mean": 0.0009211090437020175,
+      "clip_ratio/low_mean": 0.0008013766855583526,
+      "clip_ratio/low_min": 1.5232756595651153e-05,
+      "clip_ratio/region_mean": 0.0017224857365363277,
+      "epoch": 6.167930029154519,
+      "grad_norm": 0.1260383278131485,
+      "learning_rate": 1e-06,
+      "loss": -0.025,
+      "step": 594
+    },
+    {
+      "clip_ratio/high_max": 0.002393010574451182,
+      "clip_ratio/high_mean": 0.0008956637539085932,
+      "clip_ratio/low_mean": 0.0009093392845898052,
+      "clip_ratio/low_min": 0.00011197342428204138,
+      "clip_ratio/region_mean": 0.0018050030339509249,
+      "epoch": 6.1772594752186585,
+      "grad_norm": 0.13759589195251465,
+      "learning_rate": 1e-06,
+      "loss": 0.0157,
+      "step": 595
+    },
+    {
+      "clip_ratio/high_max": 0.00213131863711169,
+      "clip_ratio/high_mean": 0.0008908259896998061,
+      "clip_ratio/low_mean": 0.0008758629046496935,
+      "clip_ratio/low_min": 4.388017623568885e-05,
+      "clip_ratio/region_mean": 0.0017666888816165738,
+      "epoch": 6.186588921282799,
+      "grad_norm": 0.12837755680084229,
+      "learning_rate": 1e-06,
+      "loss": -0.0265,
+      "step": 596
+    },
+    {
+      "clip_ratio/high_max": 0.002409492401056923,
+      "clip_ratio/high_mean": 0.0010026691688835854,
+      "clip_ratio/low_mean": 0.0008720431505935267,
+      "clip_ratio/low_min": 1.5360039469669573e-05,
+      "clip_ratio/region_mean": 0.001874712310382165,
+      "epoch": 6.1959183673469385,
+      "grad_norm": 0.13673429191112518,
+      "learning_rate": 1e-06,
+      "loss": -0.0071,
+      "step": 597
+    },
+    {
+      "clip_ratio/high_max": 0.002369748406636063,
+      "clip_ratio/high_mean": 0.0010013857972808182,
+      "clip_ratio/low_mean": 0.000876778301972081,
+      "clip_ratio/low_min": 3.4616448829183355e-05,
+      "clip_ratio/region_mean": 0.0018781640974339098,
+      "epoch": 6.205247813411079,
+      "grad_norm": 0.1329900622367859,
+      "learning_rate": 1e-06,
+      "loss": 0.0188,
+      "step": 598
+    },
+    {
+      "clip_ratio/high_max": 0.0021368012894527055,
+      "clip_ratio/high_mean": 0.000836305545817595,
+      "clip_ratio/low_mean": 0.0009717360553622711,
+      "clip_ratio/low_min": 4.0644851651450153e-05,
+      "clip_ratio/region_mean": 0.0018080415975418873,
+      "epoch": 6.214577259475218,
+      "grad_norm": 0.1321057677268982,
+      "learning_rate": 1e-06,
+      "loss": 0.0373,
+      "step": 599
+    },
+    {
+      "clip_ratio/high_max": 0.0021515826738323085,
+      "clip_ratio/high_mean": 0.0008484534628223628,
+      "clip_ratio/low_mean": 0.0009161942289210856,
+      "clip_ratio/low_min": 1.8701375665841624e-05,
+      "clip_ratio/region_mean": 0.0017646476844674908,
+      "epoch": 6.223906705539359,
+      "grad_norm": 0.1322554051876068,
+      "learning_rate": 1e-06,
+      "loss": 0.0161,
+      "step": 600
+    },
+    {
+      "clip_ratio/high_max": 0.002358754485612735,
+      "clip_ratio/high_mean": 0.0009683257121650968,
+      "clip_ratio/low_mean": 0.0008753170877753291,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001843642836320214,
+      "epoch": 6.233236151603498,
+      "grad_norm": 0.11785576492547989,
+      "learning_rate": 1e-06,
+      "loss": -0.0332,
+      "step": 601
+    },
+    {
+      "clip_ratio/high_max": 0.0023365349807136226,
+      "clip_ratio/high_mean": 0.0010169639390369412,
+      "clip_ratio/low_mean": 0.0009689202806839603,
+      "clip_ratio/low_min": 5.7527481658325996e-05,
+      "clip_ratio/region_mean": 0.001985884242458269,
+      "epoch": 6.242565597667639,
+      "grad_norm": 0.12285005301237106,
+      "learning_rate": 1e-06,
+      "loss": 0.0047,
+      "step": 602
+    },
+    {
+      "clip_ratio/high_max": 0.0027111066883662716,
+      "clip_ratio/high_mean": 0.0010989059155690484,
+      "clip_ratio/low_mean": 0.0008540316393919056,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019529375931597315,
+      "epoch": 6.251895043731778,
+      "grad_norm": 0.11782839894294739,
+      "learning_rate": 1e-06,
+      "loss": -0.0259,
+      "step": 603
+    },
+    {
+      "clip_ratio/high_max": 0.0023481692041968927,
+      "clip_ratio/high_mean": 0.0010105891960847657,
+      "clip_ratio/low_mean": 0.0008507465663569747,
+      "clip_ratio/low_min": 1.814750248740893e-05,
+      "clip_ratio/region_mean": 0.0018613357678987086,
+      "epoch": 6.261224489795918,
+      "grad_norm": 0.12154977768659592,
+      "learning_rate": 1e-06,
+      "loss": -0.0293,
+      "step": 604
+    },
+    {
+      "clip_ratio/high_max": 0.0023895007907412946,
+      "clip_ratio/high_mean": 0.001023040087602567,
+      "clip_ratio/low_mean": 0.0008180166587408166,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018410567499813624,
+      "epoch": 6.270553935860058,
+      "grad_norm": 0.14218257367610931,
+      "learning_rate": 1e-06,
+      "loss": -0.0059,
+      "step": 605
+    },
+    {
+      "clip_ratio/high_max": 0.002362646468100138,
+      "clip_ratio/high_mean": 0.0010038689979410265,
+      "clip_ratio/low_mean": 0.000983446774625918,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001987315721635241,
+      "epoch": 6.279883381924198,
+      "grad_norm": 0.15273207426071167,
+      "learning_rate": 1e-06,
+      "loss": -0.0088,
+      "step": 606
+    },
+    {
+      "clip_ratio/high_max": 0.002689383734832518,
+      "clip_ratio/high_mean": 0.0010227087950624991,
+      "clip_ratio/low_mean": 0.0010481101562618278,
+      "clip_ratio/low_min": 6.923452019691467e-05,
+      "clip_ratio/region_mean": 0.0020708189040306024,
+      "epoch": 6.289212827988338,
+      "grad_norm": 0.12915882468223572,
+      "learning_rate": 1e-06,
+      "loss": -0.0116,
+      "step": 607
+    },
+    {
+      "clip_ratio/high_max": 0.002454601475619711,
+      "clip_ratio/high_mean": 0.0009301243590016384,
+      "clip_ratio/low_mean": 0.0009792717573873233,
+      "clip_ratio/low_min": 2.392802525719162e-05,
+      "clip_ratio/region_mean": 0.0019093961018370464,
+      "epoch": 6.298542274052478,
+      "grad_norm": 0.1344621181488037,
+      "learning_rate": 1e-06,
+      "loss": -0.0134,
+      "step": 608
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0467006138392857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4067.0,
+      "completions/mean_length": 719.3214111328125,
+      "completions/mean_terminated_length": 553.9032592773438,
+      "completions/min_length": 71.0,
+      "completions/min_terminated_length": 71.0,
+      "epoch": 6.307871720116618,
+      "grad_norm": 0.15148359537124634,
+      "learning_rate": 1e-06,
+      "loss": 0.011,
+      "num_tokens": 367821938.0,
+      "reward": 0.625558078289032,
+      "reward_std": 0.17333312332630157,
+      "rewards/simpleverify_reward/mean": 0.6255580186843872,
+      "rewards/simpleverify_reward/std": 0.4839869439601898,
+      "step": 609
+    },
+    {
+      "clip_ratio/high_max": 0.0022209672242752276,
+      "clip_ratio/high_mean": 0.000887323969436693,
+      "clip_ratio/low_mean": 0.000513251270604087,
+      "clip_ratio/low_min": 2.7151231734023895e-05,
+      "clip_ratio/region_mean": 0.001400575289153494,
+      "epoch": 6.317201166180758,
+      "grad_norm": 0.1446138471364975,
+      "learning_rate": 1e-06,
+      "loss": -0.0197,
+      "step": 610
+    },
+    {
+      "clip_ratio/high_max": 0.0021032415752415545,
+      "clip_ratio/high_mean": 0.0008024059316085186,
+      "clip_ratio/low_mean": 0.0005646726176564698,
+      "clip_ratio/low_min": 3.498779005894903e-05,
+      "clip_ratio/region_mean": 0.0013670785883732606,
+      "epoch": 6.326530612244898,
+      "grad_norm": 0.1274147778749466,
+      "learning_rate": 1e-06,
+      "loss": -0.0172,
+      "step": 611
+    },
+    {
+      "clip_ratio/high_max": 0.0020569848093145993,
+      "clip_ratio/high_mean": 0.0007649173021491151,
+      "clip_ratio/low_mean": 0.0006643257511314005,
+      "clip_ratio/low_min": 1.6322799638146535e-05,
+      "clip_ratio/region_mean": 0.0014292430569184944,
+      "epoch": 6.335860058309038,
+      "grad_norm": 0.1524488925933838,
+      "learning_rate": 1e-06,
+      "loss": 0.0126,
+      "step": 612
+    },
+    {
+      "clip_ratio/high_max": 0.0023838920387788676,
+      "clip_ratio/high_mean": 0.0009499993720964994,
+      "clip_ratio/low_mean": 0.0006925849338585977,
+      "clip_ratio/low_min": 3.3037757930287626e-05,
+      "clip_ratio/region_mean": 0.0016425842914031819,
+      "epoch": 6.345189504373177,
+      "grad_norm": 0.2502550780773163,
+      "learning_rate": 1e-06,
+      "loss": -0.019,
+      "step": 613
+    },
+    {
+      "clip_ratio/high_max": 0.0023228178106364794,
+      "clip_ratio/high_mean": 0.0008319845865116804,
+      "clip_ratio/low_mean": 0.0006052628432371421,
+      "clip_ratio/low_min": 1.4269406165112741e-05,
+      "clip_ratio/region_mean": 0.001437247425201349,
+      "epoch": 6.354518950437318,
+      "grad_norm": 0.14092649519443512,
+      "learning_rate": 1e-06,
+      "loss": -0.0288,
+      "step": 614
+    },
+    {
+      "clip_ratio/high_max": 0.002091754045977723,
+      "clip_ratio/high_mean": 0.0008345646383531857,
+      "clip_ratio/low_mean": 0.0007206213213066803,
+      "clip_ratio/low_min": 8.818550486466847e-05,
+      "clip_ratio/region_mean": 0.0015551859214610886,
+      "epoch": 6.363848396501457,
+      "grad_norm": 0.12686687707901,
+      "learning_rate": 1e-06,
+      "loss": 0.0251,
+      "step": 615
+    },
+    {
+      "clip_ratio/high_max": 0.0023100140315364115,
+      "clip_ratio/high_mean": 0.0009353762652608566,
+      "clip_ratio/low_mean": 0.0006434729712054832,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015788492310093716,
+      "epoch": 6.373177842565598,
+      "grad_norm": 0.14159606397151947,
+      "learning_rate": 1e-06,
+      "loss": -0.0249,
+      "step": 616
+    },
+    {
+      "clip_ratio/high_max": 0.0021749218212789856,
+      "clip_ratio/high_mean": 0.0008649941773910541,
+      "clip_ratio/low_mean": 0.0006086839803174371,
+      "clip_ratio/low_min": 1.512584731244715e-05,
+      "clip_ratio/region_mean": 0.0014736781631654594,
+      "epoch": 6.382507288629737,
+      "grad_norm": 0.12879303097724915,
+      "learning_rate": 1e-06,
+      "loss": -0.0302,
+      "step": 617
+    },
+    {
+      "clip_ratio/high_max": 0.001986747003684286,
+      "clip_ratio/high_mean": 0.0008684917156642769,
+      "clip_ratio/low_mean": 0.0006974654334044317,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015659571436117403,
+      "epoch": 6.391836734693878,
+      "grad_norm": 0.13047300279140472,
+      "learning_rate": 1e-06,
+      "loss": -0.0227,
+      "step": 618
+    },
+    {
+      "clip_ratio/high_max": 0.002185514196753502,
+      "clip_ratio/high_mean": 0.0008790957544988487,
+      "clip_ratio/low_mean": 0.0006804655877203913,
+      "clip_ratio/low_min": 1.7735528672346845e-05,
+      "clip_ratio/region_mean": 0.0015595613258483354,
+      "epoch": 6.401166180758017,
+      "grad_norm": 0.13169008493423462,
+      "learning_rate": 1e-06,
+      "loss": -0.0229,
+      "step": 619
+    },
+    {
+      "clip_ratio/high_max": 0.002048205979008344,
+      "clip_ratio/high_mean": 0.0008806505429674871,
+      "clip_ratio/low_mean": 0.0006990229676375748,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015796735024196096,
+      "epoch": 6.410495626822158,
+      "grad_norm": 0.12384453415870667,
+      "learning_rate": 1e-06,
+      "loss": -0.0254,
+      "step": 620
+    },
+    {
+      "clip_ratio/high_max": 0.001734846653562272,
+      "clip_ratio/high_mean": 0.0006966765240576933,
+      "clip_ratio/low_mean": 0.0008338587358593941,
+      "clip_ratio/low_min": 4.307616109144874e-05,
+      "clip_ratio/region_mean": 0.0015305352389987092,
+      "epoch": 6.419825072886297,
+      "grad_norm": 0.12743602693080902,
+      "learning_rate": 1e-06,
+      "loss": 0.0384,
+      "step": 621
+    },
+    {
+      "clip_ratio/high_max": 0.0022159938052936923,
+      "clip_ratio/high_mean": 0.0009187869982270058,
+      "clip_ratio/low_mean": 0.0007449616150552174,
+      "clip_ratio/low_min": 2.9599810659419745e-05,
+      "clip_ratio/region_mean": 0.001663748626015149,
+      "epoch": 6.429154518950437,
+      "grad_norm": 0.1390436589717865,
+      "learning_rate": 1e-06,
+      "loss": -0.0107,
+      "step": 622
+    },
+    {
+      "clip_ratio/high_max": 0.0025685537002573255,
+      "clip_ratio/high_mean": 0.0009798562405194389,
+      "clip_ratio/low_mean": 0.0008664640572533244,
+      "clip_ratio/low_min": 1.7660355297266506e-05,
+      "clip_ratio/region_mean": 0.0018463203086866997,
+      "epoch": 6.438483965014577,
+      "grad_norm": 0.1312895566225052,
+      "learning_rate": 1e-06,
+      "loss": -0.0101,
+      "step": 623
+    },
+    {
+      "clip_ratio/high_max": 0.0024969174119178206,
+      "clip_ratio/high_mean": 0.0008947170717874542,
+      "clip_ratio/low_mean": 0.0008471670316794189,
+      "clip_ratio/low_min": 5.392197272158228e-05,
+      "clip_ratio/region_mean": 0.0017418841089238413,
+      "epoch": 6.447813411078717,
+      "grad_norm": 0.1244460865855217,
+      "learning_rate": 1e-06,
+      "loss": -0.017,
+      "step": 624
+    },
+    {
+      "clip_ratio/high_max": 0.0019798041612375528,
+      "clip_ratio/high_mean": 0.0008563256888010073,
+      "clip_ratio/low_mean": 0.0007876590611886058,
+      "clip_ratio/low_min": 3.87471209251089e-05,
+      "clip_ratio/region_mean": 0.0016439847495348658,
+      "epoch": 6.457142857142857,
+      "grad_norm": 0.12919428944587708,
+      "learning_rate": 1e-06,
+      "loss": -0.0458,
+      "step": 625
+    },
+    {
+      "clip_ratio/high_max": 0.002259565269923769,
+      "clip_ratio/high_mean": 0.0010541889023443218,
+      "clip_ratio/low_mean": 0.0009211762953782454,
+      "clip_ratio/low_min": 4.037467806483619e-05,
+      "clip_ratio/region_mean": 0.0019753652159124613,
+      "epoch": 6.466472303206997,
+      "grad_norm": 0.13917073607444763,
+      "learning_rate": 1e-06,
+      "loss": -0.0062,
+      "step": 626
+    },
+    {
+      "clip_ratio/high_max": 0.0022725915259798057,
+      "clip_ratio/high_mean": 0.0009317070525867166,
+      "clip_ratio/low_mean": 0.0009989035042963224,
+      "clip_ratio/low_min": 1.6914749721763656e-05,
+      "clip_ratio/region_mean": 0.0019306105241412297,
+      "epoch": 6.475801749271137,
+      "grad_norm": 0.2032402753829956,
+      "learning_rate": 1e-06,
+      "loss": -0.0159,
+      "step": 627
+    },
+    {
+      "clip_ratio/high_max": 0.0022228288726182655,
+      "clip_ratio/high_mean": 0.0009595755964255659,
+      "clip_ratio/low_mean": 0.0009383168344356818,
+      "clip_ratio/low_min": 2.791659517242806e-05,
+      "clip_ratio/region_mean": 0.0018978924126713537,
+      "epoch": 6.485131195335277,
+      "grad_norm": 0.12489090114831924,
+      "learning_rate": 1e-06,
+      "loss": 0.0004,
+      "step": 628
+    },
+    {
+      "clip_ratio/high_max": 0.002745949721429497,
+      "clip_ratio/high_mean": 0.000985255235718796,
+      "clip_ratio/low_mean": 0.0010692366631701589,
+      "clip_ratio/low_min": 4.0198110582423396e-05,
+      "clip_ratio/region_mean": 0.0020544919098028913,
+      "epoch": 6.494460641399417,
+      "grad_norm": 0.14026491343975067,
+      "learning_rate": 1e-06,
+      "loss": -0.0038,
+      "step": 629
+    },
+    {
+      "clip_ratio/high_max": 0.0020863692261627875,
+      "clip_ratio/high_mean": 0.0008506243357260246,
+      "clip_ratio/low_mean": 0.0010225664809695445,
+      "clip_ratio/low_min": 5.534076990443282e-05,
+      "clip_ratio/region_mean": 0.0018731908130575903,
+      "epoch": 6.503790087463557,
+      "grad_norm": 0.13971810042858124,
+      "learning_rate": 1e-06,
+      "loss": 0.0003,
+      "step": 630
+    },
+    {
+      "clip_ratio/high_max": 0.002494731335900724,
+      "clip_ratio/high_mean": 0.0010490268068679143,
+      "clip_ratio/low_mean": 0.0008949859093263512,
+      "clip_ratio/low_min": 1.694685488473624e-05,
+      "clip_ratio/region_mean": 0.0019440127143752761,
+      "epoch": 6.513119533527696,
+      "grad_norm": 0.1381852924823761,
+      "learning_rate": 1e-06,
+      "loss": -0.0139,
+      "step": 631
+    },
+    {
+      "clip_ratio/high_max": 0.002818521599692758,
+      "clip_ratio/high_mean": 0.0010897832726186607,
+      "clip_ratio/low_mean": 0.0009193259065796155,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020091091719223186,
+      "epoch": 6.522448979591837,
+      "grad_norm": 0.13181720674037933,
+      "learning_rate": 1e-06,
+      "loss": -0.026,
+      "step": 632
+    },
+    {
+      "clip_ratio/high_max": 0.0020027623177156784,
+      "clip_ratio/high_mean": 0.0008709143876330927,
+      "clip_ratio/low_mean": 0.0011225317066418938,
+      "clip_ratio/low_min": 3.181201100233011e-05,
+      "clip_ratio/region_mean": 0.0019934461306547746,
+      "epoch": 6.531778425655976,
+      "grad_norm": 0.13129457831382751,
+      "learning_rate": 1e-06,
+      "loss": 0.0207,
+      "step": 633
+    },
+    {
+      "clip_ratio/high_max": 0.0022481358937511686,
+      "clip_ratio/high_mean": 0.0010323515944037354,
+      "clip_ratio/low_mean": 0.0008964085409388645,
+      "clip_ratio/low_min": 2.8242204280104488e-05,
+      "clip_ratio/region_mean": 0.0019287601608084515,
+      "epoch": 6.541107871720117,
+      "grad_norm": 0.14023590087890625,
+      "learning_rate": 1e-06,
+      "loss": -0.0376,
+      "step": 634
+    },
+    {
+      "clip_ratio/high_max": 0.0022737462750228588,
+      "clip_ratio/high_mean": 0.0009878130331344437,
+      "clip_ratio/low_mean": 0.0007715270003245678,
+      "clip_ratio/low_min": 3.7993919249856845e-05,
+      "clip_ratio/region_mean": 0.0017593400334590115,
+      "epoch": 6.550437317784256,
+      "grad_norm": 0.1318240761756897,
+      "learning_rate": 1e-06,
+      "loss": -0.0946,
+      "step": 635
+    },
+    {
+      "clip_ratio/high_max": 0.001854291847848799,
+      "clip_ratio/high_mean": 0.0008079188028204953,
+      "clip_ratio/low_mean": 0.001160078540124232,
+      "clip_ratio/low_min": 0.00015285129848052748,
+      "clip_ratio/region_mean": 0.0019679973629536107,
+      "epoch": 6.559766763848397,
+      "grad_norm": 0.1594737470149994,
+      "learning_rate": 1e-06,
+      "loss": 0.0375,
+      "step": 636
+    },
+    {
+      "clip_ratio/high_max": 0.002663663777639158,
+      "clip_ratio/high_mean": 0.0011117425838165218,
+      "clip_ratio/low_mean": 0.0009884372502710903,
+      "clip_ratio/low_min": 3.225418186048046e-05,
+      "clip_ratio/region_mean": 0.002100179895933252,
+      "epoch": 6.569096209912536,
+      "grad_norm": 0.13806471228599548,
+      "learning_rate": 1e-06,
+      "loss": -0.032,
+      "step": 637
+    },
+    {
+      "clip_ratio/high_max": 0.002303545923496131,
+      "clip_ratio/high_mean": 0.0009381746876897523,
+      "clip_ratio/low_mean": 0.0010181512025155826,
+      "clip_ratio/low_min": 1.7293858036282472e-05,
+      "clip_ratio/region_mean": 0.0019563259120332077,
+      "epoch": 6.578425655976677,
+      "grad_norm": 0.12728336453437805,
+      "learning_rate": 1e-06,
+      "loss": -0.006,
+      "step": 638
+    },
+    {
+      "clip_ratio/high_max": 0.002492465981049463,
+      "clip_ratio/high_mean": 0.0009622507150197634,
+      "clip_ratio/low_mean": 0.0009929457755788462,
+      "clip_ratio/low_min": 3.0569823138648644e-05,
+      "clip_ratio/region_mean": 0.0019551964942365885,
+      "epoch": 6.587755102040816,
+      "grad_norm": 0.12930989265441895,
+      "learning_rate": 1e-06,
+      "loss": 0.0055,
+      "step": 639
+    },
+    {
+      "clip_ratio/high_max": 0.002372198701777961,
+      "clip_ratio/high_mean": 0.0009705745032988489,
+      "clip_ratio/low_mean": 0.001233496735949302,
+      "clip_ratio/low_min": 2.739425872277934e-05,
+      "clip_ratio/region_mean": 0.0022040712283342145,
+      "epoch": 6.597084548104956,
+      "grad_norm": 0.12909016013145447,
+      "learning_rate": 1e-06,
+      "loss": 0.0112,
+      "step": 640
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0470145089285714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4080.0,
+      "completions/mean_length": 717.0166625976562,
+      "completions/mean_terminated_length": 550.318115234375,
+      "completions/min_length": 85.0,
+      "completions/min_terminated_length": 85.0,
+      "epoch": 6.606413994169096,
+      "grad_norm": 0.1321345865726471,
+      "learning_rate": 1e-06,
+      "loss": -0.0434,
+      "num_tokens": 385762678.0,
+      "reward": 0.6217564344406128,
+      "reward_std": 0.1684490144252777,
+      "rewards/simpleverify_reward/mean": 0.6217564344406128,
+      "rewards/simpleverify_reward/std": 0.4849573075771332,
+      "step": 641
+    },
+    {
+      "clip_ratio/high_max": 0.0019111405417788774,
+      "clip_ratio/high_mean": 0.0008823664575174917,
+      "clip_ratio/low_mean": 0.0004736024011435802,
+      "clip_ratio/low_min": 5.955027154413983e-05,
+      "clip_ratio/region_mean": 0.0013559688377426937,
+      "epoch": 6.615743440233236,
+      "grad_norm": 0.13952584564685822,
+      "learning_rate": 1e-06,
+      "loss": -0.0624,
+      "step": 642
+    },
+    {
+      "clip_ratio/high_max": 0.0022251204027270433,
+      "clip_ratio/high_mean": 0.0008111097849905491,
+      "clip_ratio/low_mean": 0.0004942562100040959,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013053659749857616,
+      "epoch": 6.625072886297376,
+      "grad_norm": 0.1269238442182541,
+      "learning_rate": 1e-06,
+      "loss": -0.0111,
+      "step": 643
+    },
+    {
+      "clip_ratio/high_max": 0.0023263603907253128,
+      "clip_ratio/high_mean": 0.0007895139106040006,
+      "clip_ratio/low_mean": 0.0004975230986019596,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012870370010205079,
+      "epoch": 6.634402332361516,
+      "grad_norm": 0.12807856500148773,
+      "learning_rate": 1e-06,
+      "loss": -0.0165,
+      "step": 644
+    },
+    {
+      "clip_ratio/high_max": 0.001852031626185635,
+      "clip_ratio/high_mean": 0.0007514072694903007,
+      "clip_ratio/low_mean": 0.0006156293147796532,
+      "clip_ratio/low_min": 3.296637714811368e-05,
+      "clip_ratio/region_mean": 0.0013670365915459115,
+      "epoch": 6.643731778425656,
+      "grad_norm": 0.14370988309383392,
+      "learning_rate": 1e-06,
+      "loss": -0.0373,
+      "step": 645
+    },
+    {
+      "clip_ratio/high_max": 0.0021303373614500742,
+      "clip_ratio/high_mean": 0.0008435951331193792,
+      "clip_ratio/low_mean": 0.0005227381061558845,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001366333231999306,
+      "epoch": 6.653061224489796,
+      "grad_norm": 0.13653628528118134,
+      "learning_rate": 1e-06,
+      "loss": -0.0438,
+      "step": 646
+    },
+    {
+      "clip_ratio/high_max": 0.0018736626516329125,
+      "clip_ratio/high_mean": 0.0007430201894749189,
+      "clip_ratio/low_mean": 0.0006139592060208088,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013569793954957277,
+      "epoch": 6.662390670553936,
+      "grad_norm": 0.13890062272548676,
+      "learning_rate": 1e-06,
+      "loss": 0.0208,
+      "step": 647
+    },
+    {
+      "clip_ratio/high_max": 0.002104283092194237,
+      "clip_ratio/high_mean": 0.0007425977601087652,
+      "clip_ratio/low_mean": 0.000717075407919765,
+      "clip_ratio/low_min": 3.3325113690807484e-05,
+      "clip_ratio/region_mean": 0.0014596731525671203,
+      "epoch": 6.671720116618076,
+      "grad_norm": 0.134938046336174,
+      "learning_rate": 1e-06,
+      "loss": 0.0182,
+      "step": 648
+    },
+    {
+      "clip_ratio/high_max": 0.0020582626675604843,
+      "clip_ratio/high_mean": 0.0008491899734508479,
+      "clip_ratio/low_mean": 0.0006601896284337272,
+      "clip_ratio/low_min": 2.7667108952300623e-05,
+      "clip_ratio/region_mean": 0.0015093796027940698,
+      "epoch": 6.681049562682215,
+      "grad_norm": 0.12558665871620178,
+      "learning_rate": 1e-06,
+      "loss": -0.0322,
+      "step": 649
+    },
+    {
+      "clip_ratio/high_max": 0.001985960712772794,
+      "clip_ratio/high_mean": 0.0007914558627817314,
+      "clip_ratio/low_mean": 0.0006645121648034547,
+      "clip_ratio/low_min": 1.5531808458035812e-05,
+      "clip_ratio/region_mean": 0.0014559680021193344,
+      "epoch": 6.690379008746356,
+      "grad_norm": 0.13106083869934082,
+      "learning_rate": 1e-06,
+      "loss": 0.0183,
+      "step": 650
+    },
+    {
+      "clip_ratio/high_max": 0.0019152088279952295,
+      "clip_ratio/high_mean": 0.0007018474379947293,
+      "clip_ratio/low_mean": 0.0007393714477075264,
+      "clip_ratio/low_min": 1.9272278223070316e-05,
+      "clip_ratio/region_mean": 0.0014412189048016444,
+      "epoch": 6.699708454810495,
+      "grad_norm": 0.13795951008796692,
+      "learning_rate": 1e-06,
+      "loss": -0.0041,
+      "step": 651
+    },
+    {
+      "clip_ratio/high_max": 0.0021834392828168347,
+      "clip_ratio/high_mean": 0.0009852310631686123,
+      "clip_ratio/low_mean": 0.0007133240669645602,
+      "clip_ratio/low_min": 2.5193605324602686e-05,
+      "clip_ratio/region_mean": 0.0016985550755634904,
+      "epoch": 6.709037900874636,
+      "grad_norm": 0.13325002789497375,
+      "learning_rate": 1e-06,
+      "loss": -0.0543,
+      "step": 652
+    },
+    {
+      "clip_ratio/high_max": 0.0024413528735749424,
+      "clip_ratio/high_mean": 0.001027618300213362,
+      "clip_ratio/low_mean": 0.0008024691069294931,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001830087428970728,
+      "epoch": 6.718367346938775,
+      "grad_norm": 0.13927236199378967,
+      "learning_rate": 1e-06,
+      "loss": -0.041,
+      "step": 653
+    },
+    {
+      "clip_ratio/high_max": 0.0024063595228653867,
+      "clip_ratio/high_mean": 0.0008185788965420215,
+      "clip_ratio/low_mean": 0.0008711892241990427,
+      "clip_ratio/low_min": 6.598297386517515e-05,
+      "clip_ratio/region_mean": 0.001689768127107527,
+      "epoch": 6.727696793002916,
+      "grad_norm": 0.15905453264713287,
+      "learning_rate": 1e-06,
+      "loss": 0.0362,
+      "step": 654
+    },
+    {
+      "clip_ratio/high_max": 0.0022277879106695764,
+      "clip_ratio/high_mean": 0.000880508352565812,
+      "clip_ratio/low_mean": 0.0008590872967033647,
+      "clip_ratio/low_min": 5.658558802679181e-05,
+      "clip_ratio/region_mean": 0.0017395956456311978,
+      "epoch": 6.737026239067055,
+      "grad_norm": 0.13501450419425964,
+      "learning_rate": 1e-06,
+      "loss": -0.0065,
+      "step": 655
+    },
+    {
+      "clip_ratio/high_max": 0.0020497181139944587,
+      "clip_ratio/high_mean": 0.0008242403855547309,
+      "clip_ratio/low_mean": 0.0008869453085935675,
+      "clip_ratio/low_min": 1.7346656022709794e-05,
+      "clip_ratio/region_mean": 0.001711185701424256,
+      "epoch": 6.746355685131196,
+      "grad_norm": 0.12874852120876312,
+      "learning_rate": 1e-06,
+      "loss": -0.0093,
+      "step": 656
+    },
+    {
+      "clip_ratio/high_max": 0.002331524548935704,
+      "clip_ratio/high_mean": 0.0009113175801758189,
+      "clip_ratio/low_mean": 0.0008681975141371367,
+      "clip_ratio/low_min": 3.488886613922659e-05,
+      "clip_ratio/region_mean": 0.0017795150924939662,
+      "epoch": 6.755685131195335,
+      "grad_norm": 0.12763261795043945,
+      "learning_rate": 1e-06,
+      "loss": 0.0088,
+      "step": 657
+    },
+    {
+      "clip_ratio/high_max": 0.00276705261785537,
+      "clip_ratio/high_mean": 0.0010822703552548774,
+      "clip_ratio/low_mean": 0.0008276878661490628,
+      "clip_ratio/low_min": 1.6864543795236386e-05,
+      "clip_ratio/region_mean": 0.001909958227770403,
+      "epoch": 6.765014577259475,
+      "grad_norm": 0.1331901103258133,
+      "learning_rate": 1e-06,
+      "loss": -0.0457,
+      "step": 658
+    },
+    {
+      "clip_ratio/high_max": 0.0021022208966314793,
+      "clip_ratio/high_mean": 0.0008931019328883849,
+      "clip_ratio/low_mean": 0.0008605613475083373,
+      "clip_ratio/low_min": 6.899977597640827e-05,
+      "clip_ratio/region_mean": 0.0017536632731207646,
+      "epoch": 6.774344023323615,
+      "grad_norm": 0.11489409953355789,
+      "learning_rate": 1e-06,
+      "loss": 0.0073,
+      "step": 659
+    },
+    {
+      "clip_ratio/high_max": 0.002463387296302244,
+      "clip_ratio/high_mean": 0.0009977245645131916,
+      "clip_ratio/low_mean": 0.0007485841033485485,
+      "clip_ratio/low_min": 2.4356975700356998e-05,
+      "clip_ratio/region_mean": 0.001746308640576899,
+      "epoch": 6.783673469387755,
+      "grad_norm": 0.11657286435365677,
+      "learning_rate": 1e-06,
+      "loss": -0.0459,
+      "step": 660
+    },
+    {
+      "clip_ratio/high_max": 0.0022582504607271403,
+      "clip_ratio/high_mean": 0.0009322661062469706,
+      "clip_ratio/low_mean": 0.0007675797896808945,
+      "clip_ratio/low_min": 4.44878378402791e-05,
+      "clip_ratio/region_mean": 0.0016998459104797803,
+      "epoch": 6.793002915451895,
+      "grad_norm": 0.1246897354722023,
+      "learning_rate": 1e-06,
+      "loss": -0.0173,
+      "step": 661
+    },
+    {
+      "clip_ratio/high_max": 0.00245572610583622,
+      "clip_ratio/high_mean": 0.0009250996590708382,
+      "clip_ratio/low_mean": 0.0010003117113228654,
+      "clip_ratio/low_min": 0.00011339371030771872,
+      "clip_ratio/region_mean": 0.0019254113140050322,
+      "epoch": 6.802332361516035,
+      "grad_norm": 0.13310350477695465,
+      "learning_rate": 1e-06,
+      "loss": 0.0137,
+      "step": 662
+    },
+    {
+      "clip_ratio/high_max": 0.0021200337941991165,
+      "clip_ratio/high_mean": 0.0009233693654095987,
+      "clip_ratio/low_mean": 0.0008585165523982141,
+      "clip_ratio/low_min": 6.877341365907341e-05,
+      "clip_ratio/region_mean": 0.0017818859341787174,
+      "epoch": 6.811661807580175,
+      "grad_norm": 0.13047559559345245,
+      "learning_rate": 1e-06,
+      "loss": 0.0144,
+      "step": 663
+    },
+    {
+      "clip_ratio/high_max": 0.001993808531551622,
+      "clip_ratio/high_mean": 0.0008405548724113032,
+      "clip_ratio/low_mean": 0.0008321946061187191,
+      "clip_ratio/low_min": 0.0001215322354255477,
+      "clip_ratio/region_mean": 0.0016727494803490117,
+      "epoch": 6.820991253644315,
+      "grad_norm": 0.11407936364412308,
+      "learning_rate": 1e-06,
+      "loss": -0.0322,
+      "step": 664
+    },
+    {
+      "clip_ratio/high_max": 0.0020853564492426813,
+      "clip_ratio/high_mean": 0.0009454951541556511,
+      "clip_ratio/low_mean": 0.0008293647879327182,
+      "clip_ratio/low_min": 6.899723666720092e-05,
+      "clip_ratio/region_mean": 0.001774859949364327,
+      "epoch": 6.830320699708455,
+      "grad_norm": 0.11815281957387924,
+      "learning_rate": 1e-06,
+      "loss": -0.0151,
+      "step": 665
+    },
+    {
+      "clip_ratio/high_max": 0.0025716686286614276,
+      "clip_ratio/high_mean": 0.001044850871039671,
+      "clip_ratio/low_mean": 0.0007745600814814679,
+      "clip_ratio/low_min": 3.21006664307788e-05,
+      "clip_ratio/region_mean": 0.0018194109507021494,
+      "epoch": 6.839650145772595,
+      "grad_norm": 0.13043001294136047,
+      "learning_rate": 1e-06,
+      "loss": -0.0444,
+      "step": 666
+    },
+    {
+      "clip_ratio/high_max": 0.002634268610563595,
+      "clip_ratio/high_mean": 0.0010826465440914035,
+      "clip_ratio/low_mean": 0.0009278506931877928,
+      "clip_ratio/low_min": 1.8927921701106243e-05,
+      "clip_ratio/region_mean": 0.00201049728639191,
+      "epoch": 6.848979591836734,
+      "grad_norm": 0.14265958964824677,
+      "learning_rate": 1e-06,
+      "loss": 0.0161,
+      "step": 667
+    },
+    {
+      "clip_ratio/high_max": 0.0021339081758924294,
+      "clip_ratio/high_mean": 0.0008609269116277574,
+      "clip_ratio/low_mean": 0.0009526016474410426,
+      "clip_ratio/low_min": 6.552633749379311e-05,
+      "clip_ratio/region_mean": 0.0018135285026801284,
+      "epoch": 6.858309037900875,
+      "grad_norm": 0.12092170119285583,
+      "learning_rate": 1e-06,
+      "loss": -0.0026,
+      "step": 668
+    },
+    {
+      "clip_ratio/high_max": 0.0026094949716934934,
+      "clip_ratio/high_mean": 0.0009530665192869492,
+      "clip_ratio/low_mean": 0.0009451533987885341,
+      "clip_ratio/low_min": 3.4131369829992764e-05,
+      "clip_ratio/region_mean": 0.0018982198744197376,
+      "epoch": 6.867638483965014,
+      "grad_norm": 0.1307818740606308,
+      "learning_rate": 1e-06,
+      "loss": 0.0059,
+      "step": 669
+    },
+    {
+      "clip_ratio/high_max": 0.002434329413517844,
+      "clip_ratio/high_mean": 0.0010277092878823169,
+      "clip_ratio/low_mean": 0.0010958412749459967,
+      "clip_ratio/low_min": 7.635772453795653e-05,
+      "clip_ratio/region_mean": 0.002123550591932144,
+      "epoch": 6.876967930029155,
+      "grad_norm": 0.14588437974452972,
+      "learning_rate": 1e-06,
+      "loss": 0.002,
+      "step": 670
+    },
+    {
+      "clip_ratio/high_max": 0.002530924561142456,
+      "clip_ratio/high_mean": 0.0010226413687632885,
+      "clip_ratio/low_mean": 0.0008716365955478977,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018942779424833134,
+      "epoch": 6.886297376093294,
+      "grad_norm": 0.1325405389070511,
+      "learning_rate": 1e-06,
+      "loss": -0.0008,
+      "step": 671
+    },
+    {
+      "clip_ratio/high_max": 0.0027137969445902854,
+      "clip_ratio/high_mean": 0.0011204696165805217,
+      "clip_ratio/low_mean": 0.0009062435274245217,
+      "clip_ratio/low_min": 0.00011564626584004145,
+      "clip_ratio/region_mean": 0.0020267131621949375,
+      "epoch": 6.895626822157435,
+      "grad_norm": 0.12995803356170654,
+      "learning_rate": 1e-06,
+      "loss": -0.0388,
+      "step": 672
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0511648995535714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4094.0,
+      "completions/mean_length": 733.5430297851562,
+      "completions/mean_terminated_length": 552.22607421875,
+      "completions/min_length": 73.0,
+      "completions/min_terminated_length": 73.0,
+      "epoch": 7.0093294460641395,
+      "grad_norm": 0.1446388065814972,
+      "learning_rate": 1e-06,
+      "loss": -0.0328,
+      "num_tokens": 403643101.0,
+      "reward": 0.631661593914032,
+      "reward_std": 0.17118562757968903,
+      "rewards/simpleverify_reward/mean": 0.6316615343093872,
+      "rewards/simpleverify_reward/std": 0.482362300157547,
+      "step": 673
+    },
+    {
+      "clip_ratio/high_max": 0.002274634229252115,
+      "clip_ratio/high_mean": 0.0008250863938883413,
+      "clip_ratio/low_mean": 0.0005860922447027406,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014111786767898593,
+      "epoch": 7.01865889212828,
+      "grad_norm": 0.14925158023834229,
+      "learning_rate": 1e-06,
+      "loss": 0.0071,
+      "step": 674
+    },
+    {
+      "clip_ratio/high_max": 0.002243863658804912,
+      "clip_ratio/high_mean": 0.000805944446256035,
+      "clip_ratio/low_mean": 0.0005229045946180122,
+      "clip_ratio/low_min": 1.3736264008912258e-05,
+      "clip_ratio/region_mean": 0.001328849026322132,
+      "epoch": 7.0279883381924195,
+      "grad_norm": 0.1439531147480011,
+      "learning_rate": 1e-06,
+      "loss": 0.0228,
+      "step": 675
+    },
+    {
+      "clip_ratio/high_max": 0.002420619966869708,
+      "clip_ratio/high_mean": 0.0008337804829352535,
+      "clip_ratio/low_mean": 0.0005500074266819865,
+      "clip_ratio/low_min": 1.165392495749984e-05,
+      "clip_ratio/region_mean": 0.0013837879268976394,
+      "epoch": 7.03731778425656,
+      "grad_norm": 0.14207249879837036,
+      "learning_rate": 1e-06,
+      "loss": -0.0277,
+      "step": 676
+    },
+    {
+      "clip_ratio/high_max": 0.002262996382341953,
+      "clip_ratio/high_mean": 0.000948994094869704,
+      "clip_ratio/low_mean": 0.0005671589369740104,
+      "clip_ratio/low_min": 4.60694382127258e-05,
+      "clip_ratio/region_mean": 0.0015161530172917992,
+      "epoch": 7.0466472303206995,
+      "grad_norm": 0.1413637399673462,
+      "learning_rate": 1e-06,
+      "loss": -0.0405,
+      "step": 677
+    },
+    {
+      "clip_ratio/high_max": 0.0017554597980051767,
+      "clip_ratio/high_mean": 0.0007474723761333735,
+      "clip_ratio/low_mean": 0.0006531819535666727,
+      "clip_ratio/low_min": 1.340626386081567e-05,
+      "clip_ratio/region_mean": 0.001400654324243078,
+      "epoch": 7.05597667638484,
+      "grad_norm": 0.14316169917583466,
+      "learning_rate": 1e-06,
+      "loss": 0.0451,
+      "step": 678
+    },
+    {
+      "clip_ratio/high_max": 0.002214854051999282,
+      "clip_ratio/high_mean": 0.0009222275821230141,
+      "clip_ratio/low_mean": 0.0006624479378842807,
+      "clip_ratio/low_min": 3.212185765733011e-05,
+      "clip_ratio/region_mean": 0.0015846755341044627,
+      "epoch": 7.0653061224489795,
+      "grad_norm": 0.13190414011478424,
+      "learning_rate": 1e-06,
+      "loss": -0.0666,
+      "step": 679
+    },
+    {
+      "clip_ratio/high_max": 0.002458992054016562,
+      "clip_ratio/high_mean": 0.0009588229640939971,
+      "clip_ratio/low_mean": 0.0006006334433550364,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015594563810736872,
+      "epoch": 7.07463556851312,
+      "grad_norm": 0.13769963383674622,
+      "learning_rate": 1e-06,
+      "loss": -0.0435,
+      "step": 680
+    },
+    {
+      "clip_ratio/high_max": 0.0019705963059095666,
+      "clip_ratio/high_mean": 0.0008263262461696286,
+      "clip_ratio/low_mean": 0.0006524786058434984,
+      "clip_ratio/low_min": 3.8958807635935955e-05,
+      "clip_ratio/region_mean": 0.0014788048538321164,
+      "epoch": 7.0839650145772595,
+      "grad_norm": 0.12982453405857086,
+      "learning_rate": 1e-06,
+      "loss": -0.024,
+      "step": 681
+    },
+    {
+      "clip_ratio/high_max": 0.002256254720123252,
+      "clip_ratio/high_mean": 0.0008849734849718516,
+      "clip_ratio/low_mean": 0.000692534000336309,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015775075044075493,
+      "epoch": 7.093294460641399,
+      "grad_norm": 0.1436099112033844,
+      "learning_rate": 1e-06,
+      "loss": -0.017,
+      "step": 682
+    },
+    {
+      "clip_ratio/high_max": 0.0022941908391658217,
+      "clip_ratio/high_mean": 0.0009941111602529418,
+      "clip_ratio/low_mean": 0.0006840911964900442,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016782023449195549,
+      "epoch": 7.1026239067055394,
+      "grad_norm": 0.13922156393527985,
+      "learning_rate": 1e-06,
+      "loss": -0.0516,
+      "step": 683
+    },
+    {
+      "clip_ratio/high_max": 0.002354970754822716,
+      "clip_ratio/high_mean": 0.0008975140444817953,
+      "clip_ratio/low_mean": 0.0006907357383170165,
+      "clip_ratio/low_min": 4.802758303412702e-05,
+      "clip_ratio/region_mean": 0.0015882497318671085,
+      "epoch": 7.111953352769679,
+      "grad_norm": 0.13131393492221832,
+      "learning_rate": 1e-06,
+      "loss": 0.0179,
+      "step": 684
+    },
+    {
+      "clip_ratio/high_max": 0.002078121993690729,
+      "clip_ratio/high_mean": 0.0008868593686202075,
+      "clip_ratio/low_mean": 0.0008656317368149757,
+      "clip_ratio/low_min": 3.122991256532259e-05,
+      "clip_ratio/region_mean": 0.0017524911236250773,
+      "epoch": 7.121282798833819,
+      "grad_norm": 0.15135273337364197,
+      "learning_rate": 1e-06,
+      "loss": 0.0063,
+      "step": 685
+    },
+    {
+      "clip_ratio/high_max": 0.0020751736919919495,
+      "clip_ratio/high_mean": 0.0007557308708783239,
+      "clip_ratio/low_mean": 0.000793259750935249,
+      "clip_ratio/low_min": 1.581477772560902e-05,
+      "clip_ratio/region_mean": 0.0015489905963477213,
+      "epoch": 7.130612244897959,
+      "grad_norm": 0.1345197707414627,
+      "learning_rate": 1e-06,
+      "loss": -0.0156,
+      "step": 686
+    },
+    {
+      "clip_ratio/high_max": 0.002680592573597096,
+      "clip_ratio/high_mean": 0.0010690451599657536,
+      "clip_ratio/low_mean": 0.0007496202579204692,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018186654197052121,
+      "epoch": 7.139941690962099,
+      "grad_norm": 0.13936971127986908,
+      "learning_rate": 1e-06,
+      "loss": -0.0451,
+      "step": 687
+    },
+    {
+      "clip_ratio/high_max": 0.002169146573578473,
+      "clip_ratio/high_mean": 0.000876297472132137,
+      "clip_ratio/low_mean": 0.0009492349327047123,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001825532381189987,
+      "epoch": 7.149271137026239,
+      "grad_norm": 0.1437898576259613,
+      "learning_rate": 1e-06,
+      "loss": 0.0306,
+      "step": 688
+    },
+    {
+      "clip_ratio/high_max": 0.0022963358896959107,
+      "clip_ratio/high_mean": 0.0009715426658658544,
+      "clip_ratio/low_mean": 0.0008095532939478289,
+      "clip_ratio/low_min": 3.749865209101699e-05,
+      "clip_ratio/region_mean": 0.001781095976184588,
+      "epoch": 7.158600583090379,
+      "grad_norm": 0.1644812375307083,
+      "learning_rate": 1e-06,
+      "loss": -0.0328,
+      "step": 689
+    },
+    {
+      "clip_ratio/high_max": 0.0020644167161663063,
+      "clip_ratio/high_mean": 0.0009001021589938318,
+      "clip_ratio/low_mean": 0.001011189477139851,
+      "clip_ratio/low_min": 1.4920028661435936e-05,
+      "clip_ratio/region_mean": 0.0019112915979349054,
+      "epoch": 7.167930029154519,
+      "grad_norm": 0.15941566228866577,
+      "learning_rate": 1e-06,
+      "loss": -0.0075,
+      "step": 690
+    },
+    {
+      "clip_ratio/high_max": 0.0023595267775817774,
+      "clip_ratio/high_mean": 0.0009081766766030341,
+      "clip_ratio/low_mean": 0.0007945056386233773,
+      "clip_ratio/low_min": 3.2765401556389406e-05,
+      "clip_ratio/region_mean": 0.001702682362520136,
+      "epoch": 7.1772594752186585,
+      "grad_norm": 0.13080213963985443,
+      "learning_rate": 1e-06,
+      "loss": 0.0068,
+      "step": 691
+    },
+    {
+      "clip_ratio/high_max": 0.0022986296971794218,
+      "clip_ratio/high_mean": 0.0010158856930502225,
+      "clip_ratio/low_mean": 0.0010339694599679206,
+      "clip_ratio/low_min": 6.743405174347572e-05,
+      "clip_ratio/region_mean": 0.00204985513119027,
+      "epoch": 7.186588921282799,
+      "grad_norm": 0.1372389942407608,
+      "learning_rate": 1e-06,
+      "loss": 0.0044,
+      "step": 692
+    },
+    {
+      "clip_ratio/high_max": 0.0023204920944408514,
+      "clip_ratio/high_mean": 0.0008961426792666316,
+      "clip_ratio/low_mean": 0.0010075816626340384,
+      "clip_ratio/low_min": 1.6525647879461758e-05,
+      "clip_ratio/region_mean": 0.0019037243546335958,
+      "epoch": 7.1959183673469385,
+      "grad_norm": 0.1330796182155609,
+      "learning_rate": 1e-06,
+      "loss": 0.022,
+      "step": 693
+    },
+    {
+      "clip_ratio/high_max": 0.0028003066036035307,
+      "clip_ratio/high_mean": 0.000988101564871613,
+      "clip_ratio/low_mean": 0.0009583205028320663,
+      "clip_ratio/low_min": 1.648641591600608e-05,
+      "clip_ratio/region_mean": 0.001946422089531552,
+      "epoch": 7.205247813411079,
+      "grad_norm": 0.1313985288143158,
+      "learning_rate": 1e-06,
+      "loss": -0.0274,
+      "step": 694
+    },
+    {
+      "clip_ratio/high_max": 0.002214182008174248,
+      "clip_ratio/high_mean": 0.0010327479685656726,
+      "clip_ratio/low_mean": 0.0008715986059542047,
+      "clip_ratio/low_min": 2.9267150239320472e-05,
+      "clip_ratio/region_mean": 0.0019043465363210998,
+      "epoch": 7.214577259475218,
+      "grad_norm": 0.13056787848472595,
+      "learning_rate": 1e-06,
+      "loss": -0.0765,
+      "step": 695
+    },
+    {
+      "clip_ratio/high_max": 0.0021691152069251984,
+      "clip_ratio/high_mean": 0.0009348489729745779,
+      "clip_ratio/low_mean": 0.000981265957307187,
+      "clip_ratio/low_min": 1.5683814126532525e-05,
+      "clip_ratio/region_mean": 0.0019161148666171357,
+      "epoch": 7.223906705539359,
+      "grad_norm": 0.12841342389583588,
+      "learning_rate": 1e-06,
+      "loss": -0.0112,
+      "step": 696
+    },
+    {
+      "clip_ratio/high_max": 0.0021718939606216736,
+      "clip_ratio/high_mean": 0.0009283142389904242,
+      "clip_ratio/low_mean": 0.0007290018111234531,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001657316075579729,
+      "epoch": 7.233236151603498,
+      "grad_norm": 0.13874033093452454,
+      "learning_rate": 1e-06,
+      "loss": -0.0493,
+      "step": 697
+    },
+    {
+      "clip_ratio/high_max": 0.002405275768978754,
+      "clip_ratio/high_mean": 0.0009633324098103913,
+      "clip_ratio/low_mean": 0.0007987799581314903,
+      "clip_ratio/low_min": 5.313033307174919e-05,
+      "clip_ratio/region_mean": 0.0017621123333810829,
+      "epoch": 7.242565597667639,
+      "grad_norm": 0.12721934914588928,
+      "learning_rate": 1e-06,
+      "loss": -0.047,
+      "step": 698
+    },
+    {
+      "clip_ratio/high_max": 0.0022063128199079074,
+      "clip_ratio/high_mean": 0.0008847082754073199,
+      "clip_ratio/low_mean": 0.0009335355862276629,
+      "clip_ratio/low_min": 1.8712575183599256e-05,
+      "clip_ratio/region_mean": 0.001818243857997004,
+      "epoch": 7.251895043731778,
+      "grad_norm": 0.1464638113975525,
+      "learning_rate": 1e-06,
+      "loss": -0.0474,
+      "step": 699
+    },
+    {
+      "clip_ratio/high_max": 0.002669274304935243,
+      "clip_ratio/high_mean": 0.0010141936545551289,
+      "clip_ratio/low_mean": 0.0008192513323592721,
+      "clip_ratio/low_min": 3.173394361510873e-05,
+      "clip_ratio/region_mean": 0.0018334450287511572,
+      "epoch": 7.261224489795918,
+      "grad_norm": 0.12942858040332794,
+      "learning_rate": 1e-06,
+      "loss": -0.0304,
+      "step": 700
+    },
+    {
+      "clip_ratio/high_max": 0.0025161706580547616,
+      "clip_ratio/high_mean": 0.0010398143494967371,
+      "clip_ratio/low_mean": 0.0010020996051025577,
+      "clip_ratio/low_min": 4.560462821245892e-05,
+      "clip_ratio/region_mean": 0.0020419140200829133,
+      "epoch": 7.270553935860058,
+      "grad_norm": 0.11450876295566559,
+      "learning_rate": 1e-06,
+      "loss": -0.0262,
+      "step": 701
+    },
+    {
+      "clip_ratio/high_max": 0.0022585059705306776,
+      "clip_ratio/high_mean": 0.0009332661811640719,
+      "clip_ratio/low_mean": 0.0011183156238985248,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020515818177955225,
+      "epoch": 7.279883381924198,
+      "grad_norm": 0.13772554695606232,
+      "learning_rate": 1e-06,
+      "loss": 0.0031,
+      "step": 702
+    },
+    {
+      "clip_ratio/high_max": 0.0026186193281318992,
+      "clip_ratio/high_mean": 0.0010648052593751345,
+      "clip_ratio/low_mean": 0.001090158299120958,
+      "clip_ratio/low_min": 8.075695222942159e-05,
+      "clip_ratio/region_mean": 0.002154963578504976,
+      "epoch": 7.289212827988338,
+      "grad_norm": 0.18790072202682495,
+      "learning_rate": 1e-06,
+      "loss": 0.0152,
+      "step": 703
+    },
+    {
+      "clip_ratio/high_max": 0.002458573682815768,
+      "clip_ratio/high_mean": 0.0010990938244503923,
+      "clip_ratio/low_mean": 0.0008878713979356689,
+      "clip_ratio/low_min": 2.874884921766352e-05,
+      "clip_ratio/region_mean": 0.0019869652242050506,
+      "epoch": 7.298542274052478,
+      "grad_norm": 0.13063327968120575,
+      "learning_rate": 1e-06,
+      "loss": -0.0327,
+      "step": 704
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0545479910714286,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4090.0,
+      "completions/mean_length": 749.7044067382812,
+      "completions/mean_terminated_length": 556.639404296875,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 7.307871720116618,
+      "grad_norm": 0.1654537320137024,
+      "learning_rate": 1e-06,
+      "loss": -0.0528,
+      "num_tokens": 421575001.0,
+      "reward": 0.6293247938156128,
+      "reward_std": 0.16807661950588226,
+      "rewards/simpleverify_reward/mean": 0.6293247938156128,
+      "rewards/simpleverify_reward/std": 0.48299407958984375,
+      "step": 705
+    },
+    {
+      "clip_ratio/high_max": 0.0018265460385009646,
+      "clip_ratio/high_mean": 0.0007896211682236753,
+      "clip_ratio/low_mean": 0.0004435829960129922,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012332041915215086,
+      "epoch": 7.317201166180758,
+      "grad_norm": 0.13172782957553864,
+      "learning_rate": 1e-06,
+      "loss": -0.0518,
+      "step": 706
+    },
+    {
+      "clip_ratio/high_max": 0.0019675174025906017,
+      "clip_ratio/high_mean": 0.000811134652394685,
+      "clip_ratio/low_mean": 0.0005115683743497357,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001322703055848251,
+      "epoch": 7.326530612244898,
+      "grad_norm": 0.1299964338541031,
+      "learning_rate": 1e-06,
+      "loss": -0.0396,
+      "step": 707
+    },
+    {
+      "clip_ratio/high_max": 0.0019363403334864415,
+      "clip_ratio/high_mean": 0.0007731013447482837,
+      "clip_ratio/low_mean": 0.0005323018531271373,
+      "clip_ratio/low_min": 2.1118432414368726e-05,
+      "clip_ratio/region_mean": 0.001305403231526725,
+      "epoch": 7.335860058309038,
+      "grad_norm": 0.15011441707611084,
+      "learning_rate": 1e-06,
+      "loss": -0.0005,
+      "step": 708
+    },
+    {
+      "clip_ratio/high_max": 0.0020807597538805567,
+      "clip_ratio/high_mean": 0.0007859736288082786,
+      "clip_ratio/low_mean": 0.0004989729723092751,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001284946600208059,
+      "epoch": 7.345189504373177,
+      "grad_norm": 0.12670396268367767,
+      "learning_rate": 1e-06,
+      "loss": -0.0189,
+      "step": 709
+    },
+    {
+      "clip_ratio/high_max": 0.00171028740805923,
+      "clip_ratio/high_mean": 0.0006777316666557454,
+      "clip_ratio/low_mean": 0.0005668956837325823,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012446273758541793,
+      "epoch": 7.354518950437318,
+      "grad_norm": 0.12118559330701828,
+      "learning_rate": 1e-06,
+      "loss": -0.0101,
+      "step": 710
+    },
+    {
+      "clip_ratio/high_max": 0.0021373688505264,
+      "clip_ratio/high_mean": 0.0007869376004236983,
+      "clip_ratio/low_mean": 0.0005991235275359941,
+      "clip_ratio/low_min": 5.3041017963550985e-05,
+      "clip_ratio/region_mean": 0.001386061132507166,
+      "epoch": 7.363848396501457,
+      "grad_norm": 0.15470343828201294,
+      "learning_rate": 1e-06,
+      "loss": -0.0066,
+      "step": 711
+    },
+    {
+      "clip_ratio/high_max": 0.0020572402245306876,
+      "clip_ratio/high_mean": 0.0008137016102409689,
+      "clip_ratio/low_mean": 0.0005353947508410783,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013490963719959836,
+      "epoch": 7.373177842565598,
+      "grad_norm": 0.12657277286052704,
+      "learning_rate": 1e-06,
+      "loss": -0.0631,
+      "step": 712
+    },
+    {
+      "clip_ratio/high_max": 0.002108126394887222,
+      "clip_ratio/high_mean": 0.0008721666526980698,
+      "clip_ratio/low_mean": 0.0005689751374120533,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014411418087547645,
+      "epoch": 7.382507288629737,
+      "grad_norm": 0.1506998986005783,
+      "learning_rate": 1e-06,
+      "loss": -0.0117,
+      "step": 713
+    },
+    {
+      "clip_ratio/high_max": 0.0021407490348792635,
+      "clip_ratio/high_mean": 0.0007874676266510505,
+      "clip_ratio/low_mean": 0.0006852654159956728,
+      "clip_ratio/low_min": 3.16756540996721e-05,
+      "clip_ratio/region_mean": 0.0014727330490131862,
+      "epoch": 7.391836734693878,
+      "grad_norm": 0.14170649647712708,
+      "learning_rate": 1e-06,
+      "loss": 0.0095,
+      "step": 714
+    },
+    {
+      "clip_ratio/high_max": 0.001763935240887804,
+      "clip_ratio/high_mean": 0.0008219140472647268,
+      "clip_ratio/low_mean": 0.0006489899096777663,
+      "clip_ratio/low_min": 2.61451586993644e-05,
+      "clip_ratio/region_mean": 0.001470903956942493,
+      "epoch": 7.401166180758017,
+      "grad_norm": 0.13472793996334076,
+      "learning_rate": 1e-06,
+      "loss": 0.0177,
+      "step": 715
+    },
+    {
+      "clip_ratio/high_max": 0.0021679992059944198,
+      "clip_ratio/high_mean": 0.0008356792923223111,
+      "clip_ratio/low_mean": 0.0007308494523385889,
+      "clip_ratio/low_min": 3.1248534469341394e-05,
+      "clip_ratio/region_mean": 0.0015665287464798894,
+      "epoch": 7.410495626822158,
+      "grad_norm": 0.14047972857952118,
+      "learning_rate": 1e-06,
+      "loss": -0.012,
+      "step": 716
+    },
+    {
+      "clip_ratio/high_max": 0.0022411099416785873,
+      "clip_ratio/high_mean": 0.0007892521362009575,
+      "clip_ratio/low_mean": 0.0007932402222650126,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015824923502805177,
+      "epoch": 7.419825072886297,
+      "grad_norm": 0.13195101916790009,
+      "learning_rate": 1e-06,
+      "loss": 0.0245,
+      "step": 717
+    },
+    {
+      "clip_ratio/high_max": 0.002300900043337606,
+      "clip_ratio/high_mean": 0.0009294850769947516,
+      "clip_ratio/low_mean": 0.0006163782636576798,
+      "clip_ratio/low_min": 1.4884496522427071e-05,
+      "clip_ratio/region_mean": 0.0015458633279195055,
+      "epoch": 7.429154518950437,
+      "grad_norm": 0.12871462106704712,
+      "learning_rate": 1e-06,
+      "loss": -0.0491,
+      "step": 718
+    },
+    {
+      "clip_ratio/high_max": 0.002381900525506353,
+      "clip_ratio/high_mean": 0.0010049530519609107,
+      "clip_ratio/low_mean": 0.0007968530899233883,
+      "clip_ratio/low_min": 1.3516436411009636e-05,
+      "clip_ratio/region_mean": 0.0018018061346083414,
+      "epoch": 7.438483965014577,
+      "grad_norm": 0.13149583339691162,
+      "learning_rate": 1e-06,
+      "loss": -0.0338,
+      "step": 719
+    },
+    {
+      "clip_ratio/high_max": 0.0022384045259968843,
+      "clip_ratio/high_mean": 0.0009122575393121224,
+      "clip_ratio/low_mean": 0.0008931772263167659,
+      "clip_ratio/low_min": 4.29996543971356e-05,
+      "clip_ratio/region_mean": 0.0018054347456200048,
+      "epoch": 7.447813411078717,
+      "grad_norm": 0.1355581134557724,
+      "learning_rate": 1e-06,
+      "loss": -0.0048,
+      "step": 720
+    },
+    {
+      "clip_ratio/high_max": 0.0022939459486224223,
+      "clip_ratio/high_mean": 0.0008962302754298435,
+      "clip_ratio/low_mean": 0.0007868133252486587,
+      "clip_ratio/low_min": 2.021999353019055e-05,
+      "clip_ratio/region_mean": 0.0016830435997690074,
+      "epoch": 7.457142857142857,
+      "grad_norm": 0.12592262029647827,
+      "learning_rate": 1e-06,
+      "loss": -0.0571,
+      "step": 721
+    },
+    {
+      "clip_ratio/high_max": 0.0022696035593980923,
+      "clip_ratio/high_mean": 0.0008649930314277299,
+      "clip_ratio/low_mean": 0.00077668963058386,
+      "clip_ratio/low_min": 3.1017370929475874e-05,
+      "clip_ratio/region_mean": 0.0016416826292697806,
+      "epoch": 7.466472303206997,
+      "grad_norm": 0.13758066296577454,
+      "learning_rate": 1e-06,
+      "loss": -0.0146,
+      "step": 722
+    },
+    {
+      "clip_ratio/high_max": 0.002263756272441242,
+      "clip_ratio/high_mean": 0.0008887717467587208,
+      "clip_ratio/low_mean": 0.0009162615606328472,
+      "clip_ratio/low_min": 3.6778646062884945e-05,
+      "clip_ratio/region_mean": 0.001805033316486515,
+      "epoch": 7.475801749271137,
+      "grad_norm": 0.1429128646850586,
+      "learning_rate": 1e-06,
+      "loss": -0.0103,
+      "step": 723
+    },
+    {
+      "clip_ratio/high_max": 0.002174385757825803,
+      "clip_ratio/high_mean": 0.0008165840026777005,
+      "clip_ratio/low_mean": 0.001072266481060069,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018888504710048437,
+      "epoch": 7.485131195335277,
+      "grad_norm": 0.1299685388803482,
+      "learning_rate": 1e-06,
+      "loss": -0.0031,
+      "step": 724
+    },
+    {
+      "clip_ratio/high_max": 0.0025791304724407382,
+      "clip_ratio/high_mean": 0.0009592306014383212,
+      "clip_ratio/low_mean": 0.0008740468474570662,
+      "clip_ratio/low_min": 9.472469355387148e-05,
+      "clip_ratio/region_mean": 0.0018332774852751754,
+      "epoch": 7.494460641399417,
+      "grad_norm": 0.13293859362602234,
+      "learning_rate": 1e-06,
+      "loss": -0.0253,
+      "step": 725
+    },
+    {
+      "clip_ratio/high_max": 0.0021132149777258746,
+      "clip_ratio/high_mean": 0.0008707468296051957,
+      "clip_ratio/low_mean": 0.0008908439904189436,
+      "clip_ratio/low_min": 5.6357079301960766e-05,
+      "clip_ratio/region_mean": 0.0017615908582229167,
+      "epoch": 7.503790087463557,
+      "grad_norm": 0.16601741313934326,
+      "learning_rate": 1e-06,
+      "loss": -0.0005,
+      "step": 726
+    },
+    {
+      "clip_ratio/high_max": 0.00237304413531092,
+      "clip_ratio/high_mean": 0.0009437441749469144,
+      "clip_ratio/low_mean": 0.0008811013576632831,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018248454907734413,
+      "epoch": 7.513119533527696,
+      "grad_norm": 0.12976625561714172,
+      "learning_rate": 1e-06,
+      "loss": 0.0076,
+      "step": 727
+    },
+    {
+      "clip_ratio/high_max": 0.0020073395935469307,
+      "clip_ratio/high_mean": 0.000853819359690533,
+      "clip_ratio/low_mean": 0.0010332491765439045,
+      "clip_ratio/low_min": 0.00010556827328400686,
+      "clip_ratio/region_mean": 0.0018870685453293845,
+      "epoch": 7.522448979591837,
+      "grad_norm": 0.1306147277355194,
+      "learning_rate": 1e-06,
+      "loss": 0.0245,
+      "step": 728
+    },
+    {
+      "clip_ratio/high_max": 0.0022646796387562063,
+      "clip_ratio/high_mean": 0.0009346798888145713,
+      "clip_ratio/low_mean": 0.0009723316361487377,
+      "clip_ratio/low_min": 3.4032127587124705e-05,
+      "clip_ratio/region_mean": 0.001907011515868362,
+      "epoch": 7.531778425655976,
+      "grad_norm": 0.1261693239212036,
+      "learning_rate": 1e-06,
+      "loss": 0.0064,
+      "step": 729
+    },
+    {
+      "clip_ratio/high_max": 0.0021227979741524905,
+      "clip_ratio/high_mean": 0.0009229148636222817,
+      "clip_ratio/low_mean": 0.0009381707404827466,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001861085562268272,
+      "epoch": 7.541107871720117,
+      "grad_norm": 0.1381930261850357,
+      "learning_rate": 1e-06,
+      "loss": -0.0369,
+      "step": 730
+    },
+    {
+      "clip_ratio/high_max": 0.002701890691241715,
+      "clip_ratio/high_mean": 0.0010717016557464376,
+      "clip_ratio/low_mean": 0.0009449970402783947,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020166987233096734,
+      "epoch": 7.550437317784256,
+      "grad_norm": 0.12776899337768555,
+      "learning_rate": 1e-06,
+      "loss": -0.009,
+      "step": 731
+    },
+    {
+      "clip_ratio/high_max": 0.0025922139902831987,
+      "clip_ratio/high_mean": 0.0010527947997616138,
+      "clip_ratio/low_mean": 0.0009150450914603425,
+      "clip_ratio/low_min": 7.38550370442681e-05,
+      "clip_ratio/region_mean": 0.0019678398821270093,
+      "epoch": 7.559766763848397,
+      "grad_norm": 0.12704047560691833,
+      "learning_rate": 1e-06,
+      "loss": -0.0294,
+      "step": 732
+    },
+    {
+      "clip_ratio/high_max": 0.0025765405007405207,
+      "clip_ratio/high_mean": 0.0010714326563174836,
+      "clip_ratio/low_mean": 0.000773973153627594,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018454057790222578,
+      "epoch": 7.569096209912536,
+      "grad_norm": 0.12981204688549042,
+      "learning_rate": 1e-06,
+      "loss": -0.0459,
+      "step": 733
+    },
+    {
+      "clip_ratio/high_max": 0.0025848354889603797,
+      "clip_ratio/high_mean": 0.0010267427805956686,
+      "clip_ratio/low_mean": 0.0009916157578118145,
+      "clip_ratio/low_min": 7.33508177290787e-05,
+      "clip_ratio/region_mean": 0.002018358551140409,
+      "epoch": 7.578425655976677,
+      "grad_norm": 0.12835033237934113,
+      "learning_rate": 1e-06,
+      "loss": -0.0476,
+      "step": 734
+    },
+    {
+      "clip_ratio/high_max": 0.0029205042446847074,
+      "clip_ratio/high_mean": 0.0010689858536352403,
+      "clip_ratio/low_mean": 0.0009311280464316951,
+      "clip_ratio/low_min": 3.9735346945235506e-05,
+      "clip_ratio/region_mean": 0.002000113898247946,
+      "epoch": 7.587755102040816,
+      "grad_norm": 0.13488556444644928,
+      "learning_rate": 1e-06,
+      "loss": -0.0107,
+      "step": 735
+    },
+    {
+      "clip_ratio/high_max": 0.0026054226036649197,
+      "clip_ratio/high_mean": 0.0011152154329465702,
+      "clip_ratio/low_mean": 0.0009277506087528309,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020429660435183905,
+      "epoch": 7.597084548104956,
+      "grad_norm": 0.14065741002559662,
+      "learning_rate": 1e-06,
+      "loss": -0.0273,
+      "step": 736
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0523507254464286,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4093.0,
+      "completions/mean_length": 726.693115234375,
+      "completions/mean_terminated_length": 540.5634155273438,
+      "completions/min_length": 44.0,
+      "completions/min_terminated_length": 44.0,
+      "epoch": 7.606413994169096,
+      "grad_norm": 0.14061662554740906,
+      "learning_rate": 1e-06,
+      "loss": -0.0664,
+      "num_tokens": 439141330.0,
+      "reward": 0.6386370062828064,
+      "reward_std": 0.1572299748659134,
+      "rewards/simpleverify_reward/mean": 0.6386370062828064,
+      "rewards/simpleverify_reward/std": 0.4804038405418396,
+      "step": 737
+    },
+    {
+      "clip_ratio/high_max": 0.0021191291889408603,
+      "clip_ratio/high_mean": 0.000694068457960384,
+      "clip_ratio/low_mean": 0.0005553743349082652,
+      "clip_ratio/low_min": 2.4898302399378736e-05,
+      "clip_ratio/region_mean": 0.0012494427974161226,
+      "epoch": 7.615743440233236,
+      "grad_norm": 0.13872262835502625,
+      "learning_rate": 1e-06,
+      "loss": -0.0296,
+      "step": 738
+    },
+    {
+      "clip_ratio/high_max": 0.001839825723436661,
+      "clip_ratio/high_mean": 0.0007257124880197807,
+      "clip_ratio/low_mean": 0.0005697241695088451,
+      "clip_ratio/low_min": 1.4695508980366867e-05,
+      "clip_ratio/region_mean": 0.0012954366466146894,
+      "epoch": 7.625072886297376,
+      "grad_norm": 0.1459311693906784,
+      "learning_rate": 1e-06,
+      "loss": 0.0056,
+      "step": 739
+    },
+    {
+      "clip_ratio/high_max": 0.0017843237947090529,
+      "clip_ratio/high_mean": 0.0007224646633403609,
+      "clip_ratio/low_mean": 0.00046399886696235626,
+      "clip_ratio/low_min": 1.4387661394721363e-05,
+      "clip_ratio/region_mean": 0.0011864635271194857,
+      "epoch": 7.634402332361516,
+      "grad_norm": 0.13359272480010986,
+      "learning_rate": 1e-06,
+      "loss": -0.0104,
+      "step": 740
+    },
+    {
+      "clip_ratio/high_max": 0.0019549584576452617,
+      "clip_ratio/high_mean": 0.0007238358693939517,
+      "clip_ratio/low_mean": 0.0005262705772111076,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001250106462975964,
+      "epoch": 7.643731778425656,
+      "grad_norm": 0.13564907014369965,
+      "learning_rate": 1e-06,
+      "loss": 0.0085,
+      "step": 741
+    },
+    {
+      "clip_ratio/high_max": 0.0019569512478483375,
+      "clip_ratio/high_mean": 0.0007869689179642592,
+      "clip_ratio/low_mean": 0.0005607096295534575,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013476785497914534,
+      "epoch": 7.653061224489796,
+      "grad_norm": 0.13925661146640778,
+      "learning_rate": 1e-06,
+      "loss": -0.013,
+      "step": 742
+    },
+    {
+      "clip_ratio/high_max": 0.002155585403670557,
+      "clip_ratio/high_mean": 0.0007721753099758644,
+      "clip_ratio/low_mean": 0.0006401544342224952,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014123297041805927,
+      "epoch": 7.662390670553936,
+      "grad_norm": 0.1261594295501709,
+      "learning_rate": 1e-06,
+      "loss": 0.0021,
+      "step": 743
+    },
+    {
+      "clip_ratio/high_max": 0.0020438058563740924,
+      "clip_ratio/high_mean": 0.0007585656567243859,
+      "clip_ratio/low_mean": 0.000594217734033009,
+      "clip_ratio/low_min": 5.420122579380404e-05,
+      "clip_ratio/region_mean": 0.0013527834016713314,
+      "epoch": 7.671720116618076,
+      "grad_norm": 0.14886973798274994,
+      "learning_rate": 1e-06,
+      "loss": 0.0097,
+      "step": 744
+    },
+    {
+      "clip_ratio/high_max": 0.002141269600542728,
+      "clip_ratio/high_mean": 0.0007296354851860087,
+      "clip_ratio/low_mean": 0.0005828428238601191,
+      "clip_ratio/low_min": 3.678119173855521e-05,
+      "clip_ratio/region_mean": 0.0013124783035891596,
+      "epoch": 7.681049562682215,
+      "grad_norm": 0.1279527097940445,
+      "learning_rate": 1e-06,
+      "loss": -0.0107,
+      "step": 745
+    },
+    {
+      "clip_ratio/high_max": 0.0018894861641456373,
+      "clip_ratio/high_mean": 0.0008073787248576991,
+      "clip_ratio/low_mean": 0.0006899479803905706,
+      "clip_ratio/low_min": 1.3778659194940701e-05,
+      "clip_ratio/region_mean": 0.001497326695243828,
+      "epoch": 7.690379008746356,
+      "grad_norm": 0.12833702564239502,
+      "learning_rate": 1e-06,
+      "loss": -0.0189,
+      "step": 746
+    },
+    {
+      "clip_ratio/high_max": 0.0021815282816533,
+      "clip_ratio/high_mean": 0.0008724681774765486,
+      "clip_ratio/low_mean": 0.00075677987842937,
+      "clip_ratio/low_min": 6.36046752333641e-05,
+      "clip_ratio/region_mean": 0.0016292480286210775,
+      "epoch": 7.699708454810495,
+      "grad_norm": 0.14017045497894287,
+      "learning_rate": 1e-06,
+      "loss": -0.0239,
+      "step": 747
+    },
+    {
+      "clip_ratio/high_max": 0.002242789210868068,
+      "clip_ratio/high_mean": 0.000980944896582514,
+      "clip_ratio/low_mean": 0.0006289490929702879,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016098940395750105,
+      "epoch": 7.709037900874636,
+      "grad_norm": 0.1378800868988037,
+      "learning_rate": 1e-06,
+      "loss": -0.069,
+      "step": 748
+    },
+    {
+      "clip_ratio/high_max": 0.0022026304941391572,
+      "clip_ratio/high_mean": 0.0008512298409186769,
+      "clip_ratio/low_mean": 0.0007073326523823198,
+      "clip_ratio/low_min": 5.845650503033539e-05,
+      "clip_ratio/region_mean": 0.0015585625224048272,
+      "epoch": 7.718367346938775,
+      "grad_norm": 0.15025298297405243,
+      "learning_rate": 1e-06,
+      "loss": -0.0259,
+      "step": 749
+    },
+    {
+      "clip_ratio/high_max": 0.0021277290979924146,
+      "clip_ratio/high_mean": 0.0008715810854482697,
+      "clip_ratio/low_mean": 0.0008244460714195156,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016960271932475735,
+      "epoch": 7.727696793002916,
+      "grad_norm": 0.12647944688796997,
+      "learning_rate": 1e-06,
+      "loss": -0.0316,
+      "step": 750
+    },
+    {
+      "clip_ratio/high_max": 0.0019349800022609998,
+      "clip_ratio/high_mean": 0.0007705838361289352,
+      "clip_ratio/low_mean": 0.0007449139393429505,
+      "clip_ratio/low_min": 1.825883737183176e-05,
+      "clip_ratio/region_mean": 0.0015154977700149175,
+      "epoch": 7.737026239067055,
+      "grad_norm": 0.12538565695285797,
+      "learning_rate": 1e-06,
+      "loss": -0.0166,
+      "step": 751
+    },
+    {
+      "clip_ratio/high_max": 0.002175618767068954,
+      "clip_ratio/high_mean": 0.0007867826789151877,
+      "clip_ratio/low_mean": 0.0006965943339309888,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014833769891993143,
+      "epoch": 7.746355685131196,
+      "grad_norm": 0.11999280750751495,
+      "learning_rate": 1e-06,
+      "loss": -0.0122,
+      "step": 752
+    },
+    {
+      "clip_ratio/high_max": 0.002342407766263932,
+      "clip_ratio/high_mean": 0.0008849341065797489,
+      "clip_ratio/low_mean": 0.0007787324575474486,
+      "clip_ratio/low_min": 6.030642907717265e-05,
+      "clip_ratio/region_mean": 0.0016636665386613458,
+      "epoch": 7.755685131195335,
+      "grad_norm": 0.12813863158226013,
+      "learning_rate": 1e-06,
+      "loss": 0.0155,
+      "step": 753
+    },
+    {
+      "clip_ratio/high_max": 0.002402739835815737,
+      "clip_ratio/high_mean": 0.0009222707431035815,
+      "clip_ratio/low_mean": 0.0007032123385215527,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016254830807156395,
+      "epoch": 7.765014577259475,
+      "grad_norm": 0.12236373871564865,
+      "learning_rate": 1e-06,
+      "loss": -0.0404,
+      "step": 754
+    },
+    {
+      "clip_ratio/high_max": 0.0023693379189353436,
+      "clip_ratio/high_mean": 0.0009158612647297559,
+      "clip_ratio/low_mean": 0.0006638583918174845,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001579719697474502,
+      "epoch": 7.774344023323615,
+      "grad_norm": 0.14664119482040405,
+      "learning_rate": 1e-06,
+      "loss": -0.02,
+      "step": 755
+    },
+    {
+      "clip_ratio/high_max": 0.0024966469463834073,
+      "clip_ratio/high_mean": 0.0009141220871242695,
+      "clip_ratio/low_mean": 0.0008772214750933927,
+      "clip_ratio/low_min": 4.241121860104613e-05,
+      "clip_ratio/region_mean": 0.0017913435731315985,
+      "epoch": 7.783673469387755,
+      "grad_norm": 0.1441788375377655,
+      "learning_rate": 1e-06,
+      "loss": -0.0246,
+      "step": 756
+    },
+    {
+      "clip_ratio/high_max": 0.002183746179071022,
+      "clip_ratio/high_mean": 0.0008900575940060662,
+      "clip_ratio/low_mean": 0.0007474913682017359,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016375489758502226,
+      "epoch": 7.793002915451895,
+      "grad_norm": 0.1311672180891037,
+      "learning_rate": 1e-06,
+      "loss": -0.0248,
+      "step": 757
+    },
+    {
+      "clip_ratio/high_max": 0.0023040065098030027,
+      "clip_ratio/high_mean": 0.0009115982938965317,
+      "clip_ratio/low_mean": 0.0009188011445075972,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018303994220332243,
+      "epoch": 7.802332361516035,
+      "grad_norm": 0.12278234958648682,
+      "learning_rate": 1e-06,
+      "loss": -0.0019,
+      "step": 758
+    },
+    {
+      "clip_ratio/high_max": 0.0024888798579922877,
+      "clip_ratio/high_mean": 0.0009157267086266074,
+      "clip_ratio/low_mean": 0.0009965313620341476,
+      "clip_ratio/low_min": 3.1063114874996245e-05,
+      "clip_ratio/region_mean": 0.0019122580779367127,
+      "epoch": 7.811661807580175,
+      "grad_norm": 0.16130737960338593,
+      "learning_rate": 1e-06,
+      "loss": -0.0011,
+      "step": 759
+    },
+    {
+      "clip_ratio/high_max": 0.0022073256404837593,
+      "clip_ratio/high_mean": 0.0008899613239918835,
+      "clip_ratio/low_mean": 0.0007628938619745895,
+      "clip_ratio/low_min": 1.6237983800238e-05,
+      "clip_ratio/region_mean": 0.0016528551605006214,
+      "epoch": 7.820991253644315,
+      "grad_norm": 0.14124083518981934,
+      "learning_rate": 1e-06,
+      "loss": -0.0324,
+      "step": 760
+    },
+    {
+      "clip_ratio/high_max": 0.0023219711292767897,
+      "clip_ratio/high_mean": 0.0009657984555815347,
+      "clip_ratio/low_mean": 0.0008797907903499436,
+      "clip_ratio/low_min": 2.1922132873442024e-05,
+      "clip_ratio/region_mean": 0.0018455891840858385,
+      "epoch": 7.830320699708455,
+      "grad_norm": 0.13350093364715576,
+      "learning_rate": 1e-06,
+      "loss": -0.0612,
+      "step": 761
+    },
+    {
+      "clip_ratio/high_max": 0.0023900306259747595,
+      "clip_ratio/high_mean": 0.000912744899324025,
+      "clip_ratio/low_mean": 0.0008640388550702482,
+      "clip_ratio/low_min": 4.823771450901404e-05,
+      "clip_ratio/region_mean": 0.0017767837634892203,
+      "epoch": 7.839650145772595,
+      "grad_norm": 0.1436462700366974,
+      "learning_rate": 1e-06,
+      "loss": 0.0099,
+      "step": 762
+    },
+    {
+      "clip_ratio/high_max": 0.0022774578974349424,
+      "clip_ratio/high_mean": 0.0009866814325505402,
+      "clip_ratio/low_mean": 0.0009157908698398387,
+      "clip_ratio/low_min": 7.198559251264669e-05,
+      "clip_ratio/region_mean": 0.00190247232967522,
+      "epoch": 7.848979591836734,
+      "grad_norm": 0.15024107694625854,
+      "learning_rate": 1e-06,
+      "loss": -0.0429,
+      "step": 763
+    },
+    {
+      "clip_ratio/high_max": 0.0024024269805522636,
+      "clip_ratio/high_mean": 0.0009643874273024267,
+      "clip_ratio/low_mean": 0.0010378498391219182,
+      "clip_ratio/low_min": 4.0650407754583284e-05,
+      "clip_ratio/region_mean": 0.002002237299166154,
+      "epoch": 7.858309037900875,
+      "grad_norm": 0.14950677752494812,
+      "learning_rate": 1e-06,
+      "loss": 0.0092,
+      "step": 764
+    },
+    {
+      "clip_ratio/high_max": 0.0025013785816554446,
+      "clip_ratio/high_mean": 0.0010364690988353686,
+      "clip_ratio/low_mean": 0.0008471385317534441,
+      "clip_ratio/low_min": 3.644208936748328e-05,
+      "clip_ratio/region_mean": 0.0018836076342267916,
+      "epoch": 7.867638483965014,
+      "grad_norm": 0.1644466072320938,
+      "learning_rate": 1e-06,
+      "loss": -0.0366,
+      "step": 765
+    },
+    {
+      "clip_ratio/high_max": 0.002225145770353265,
+      "clip_ratio/high_mean": 0.0009225417343259323,
+      "clip_ratio/low_mean": 0.000907451358216349,
+      "clip_ratio/low_min": 4.641090527002234e-05,
+      "clip_ratio/region_mean": 0.0018299930889043026,
+      "epoch": 7.876967930029155,
+      "grad_norm": 0.14500048756599426,
+      "learning_rate": 1e-06,
+      "loss": 0.0122,
+      "step": 766
+    },
+    {
+      "clip_ratio/high_max": 0.0023329905816353858,
+      "clip_ratio/high_mean": 0.0009116015207837336,
+      "clip_ratio/low_mean": 0.001174640399767668,
+      "clip_ratio/low_min": 9.06680324987974e-05,
+      "clip_ratio/region_mean": 0.0020862419478362426,
+      "epoch": 7.886297376093294,
+      "grad_norm": 0.14140664041042328,
+      "learning_rate": 1e-06,
+      "loss": 0.0479,
+      "step": 767
+    },
+    {
+      "clip_ratio/high_max": 0.002283850350067951,
+      "clip_ratio/high_mean": 0.0010036260773631511,
+      "clip_ratio/low_mean": 0.000938110258175584,
+      "clip_ratio/low_min": 3.4181022783741355e-05,
+      "clip_ratio/region_mean": 0.0019417363364482298,
+      "epoch": 7.895626822157435,
+      "grad_norm": 0.12925556302070618,
+      "learning_rate": 1e-06,
+      "loss": -0.0362,
+      "step": 768
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0570591517857143,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4073.0,
+      "completions/mean_length": 751.2974853515625,
+      "completions/mean_terminated_length": 548.903076171875,
+      "completions/min_length": 6.0,
+      "completions/min_terminated_length": 6.0,
+      "epoch": 8.00932944606414,
+      "grad_norm": 0.13675914704799652,
+      "learning_rate": 1e-06,
+      "loss": -0.0454,
+      "num_tokens": 456881505.0,
+      "reward": 0.6374163031578064,
+      "reward_std": 0.16118381917476654,
+      "rewards/simpleverify_reward/mean": 0.6374163031578064,
+      "rewards/simpleverify_reward/std": 0.48075446486473083,
+      "step": 769
+    },
+    {
+      "clip_ratio/high_max": 0.0016781171871116385,
+      "clip_ratio/high_mean": 0.0006622902747039916,
+      "clip_ratio/low_mean": 0.0004056672162278119,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010679574879759457,
+      "epoch": 8.018658892128279,
+      "grad_norm": 0.12612979114055634,
+      "learning_rate": 1e-06,
+      "loss": -0.0346,
+      "step": 770
+    },
+    {
+      "clip_ratio/high_max": 0.0017111044380726526,
+      "clip_ratio/high_mean": 0.0006947221700102091,
+      "clip_ratio/low_mean": 0.0003641227492607868,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010588449295028113,
+      "epoch": 8.02798833819242,
+      "grad_norm": 0.13261425495147705,
+      "learning_rate": 1e-06,
+      "loss": -0.0395,
+      "step": 771
+    },
+    {
+      "clip_ratio/high_max": 0.0014102601453487296,
+      "clip_ratio/high_mean": 0.0006350980161187181,
+      "clip_ratio/low_mean": 0.0005383820771385217,
+      "clip_ratio/low_min": 1.190249440696789e-05,
+      "clip_ratio/region_mean": 0.0011734801119018812,
+      "epoch": 8.03731778425656,
+      "grad_norm": 0.1318708062171936,
+      "learning_rate": 1e-06,
+      "loss": -0.0049,
+      "step": 772
+    },
+    {
+      "clip_ratio/high_max": 0.001980179651582148,
+      "clip_ratio/high_mean": 0.000854958598210942,
+      "clip_ratio/low_mean": 0.0005672808019880904,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014222394020180218,
+      "epoch": 8.0466472303207,
+      "grad_norm": 0.13942308723926544,
+      "learning_rate": 1e-06,
+      "loss": -0.0381,
+      "step": 773
+    },
+    {
+      "clip_ratio/high_max": 0.0018664007293409668,
+      "clip_ratio/high_mean": 0.0008455624174530385,
+      "clip_ratio/low_mean": 0.0006330710912152426,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014786335341341328,
+      "epoch": 8.055976676384839,
+      "grad_norm": 0.16271492838859558,
+      "learning_rate": 1e-06,
+      "loss": -0.0264,
+      "step": 774
+    },
+    {
+      "clip_ratio/high_max": 0.0018644443043740466,
+      "clip_ratio/high_mean": 0.0007480310832761461,
+      "clip_ratio/low_mean": 0.0005536467474485107,
+      "clip_ratio/low_min": 2.112557012878824e-05,
+      "clip_ratio/region_mean": 0.001301677868468687,
+      "epoch": 8.06530612244898,
+      "grad_norm": 0.12925513088703156,
+      "learning_rate": 1e-06,
+      "loss": -0.0489,
+      "step": 775
+    },
+    {
+      "clip_ratio/high_max": 0.0023161625758802984,
+      "clip_ratio/high_mean": 0.0009264021828130353,
+      "clip_ratio/low_mean": 0.0006490904725069413,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015754926898807753,
+      "epoch": 8.07463556851312,
+      "grad_norm": 0.13870471715927124,
+      "learning_rate": 1e-06,
+      "loss": -0.033,
+      "step": 776
+    },
+    {
+      "clip_ratio/high_max": 0.0019738978226087056,
+      "clip_ratio/high_mean": 0.0007244862554216525,
+      "clip_ratio/low_mean": 0.0007899869851826224,
+      "clip_ratio/low_min": 6.817312169005163e-05,
+      "clip_ratio/region_mean": 0.0015144732169574127,
+      "epoch": 8.08396501457726,
+      "grad_norm": 0.12384528666734695,
+      "learning_rate": 1e-06,
+      "loss": 0.0128,
+      "step": 777
+    },
+    {
+      "clip_ratio/high_max": 0.0023334121251537,
+      "clip_ratio/high_mean": 0.0008317334923049202,
+      "clip_ratio/low_mean": 0.0006092061266826931,
+      "clip_ratio/low_min": 1.4541647033183835e-05,
+      "clip_ratio/region_mean": 0.0014409396462724544,
+      "epoch": 8.093294460641399,
+      "grad_norm": 0.1514134556055069,
+      "learning_rate": 1e-06,
+      "loss": -0.0476,
+      "step": 778
+    },
+    {
+      "clip_ratio/high_max": 0.00216744260978885,
+      "clip_ratio/high_mean": 0.0009052204914041795,
+      "clip_ratio/low_mean": 0.0006727804830006789,
+      "clip_ratio/low_min": 2.3656320990994573e-05,
+      "clip_ratio/region_mean": 0.00157800099987071,
+      "epoch": 8.102623906705539,
+      "grad_norm": 0.19655881822109222,
+      "learning_rate": 1e-06,
+      "loss": -0.0622,
+      "step": 779
+    },
+    {
+      "clip_ratio/high_max": 0.0019985899634775706,
+      "clip_ratio/high_mean": 0.0008344051639141981,
+      "clip_ratio/low_mean": 0.0007848602344893152,
+      "clip_ratio/low_min": 1.8502072634873912e-05,
+      "clip_ratio/region_mean": 0.0016192654002225026,
+      "epoch": 8.11195335276968,
+      "grad_norm": 0.1499754637479782,
+      "learning_rate": 1e-06,
+      "loss": 0.0097,
+      "step": 780
+    },
+    {
+      "clip_ratio/high_max": 0.001745296161971055,
+      "clip_ratio/high_mean": 0.0007103092029865365,
+      "clip_ratio/low_mean": 0.0007737592368357582,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014840684198134113,
+      "epoch": 8.12128279883382,
+      "grad_norm": 0.12729890644550323,
+      "learning_rate": 1e-06,
+      "loss": -0.0037,
+      "step": 781
+    },
+    {
+      "clip_ratio/high_max": 0.0021704562386730686,
+      "clip_ratio/high_mean": 0.0008493876994180027,
+      "clip_ratio/low_mean": 0.0008668792506796308,
+      "clip_ratio/low_min": 3.071253013331443e-05,
+      "clip_ratio/region_mean": 0.00171626696101157,
+      "epoch": 8.130612244897959,
+      "grad_norm": 0.1464262455701828,
+      "learning_rate": 1e-06,
+      "loss": -0.0003,
+      "step": 782
+    },
+    {
+      "clip_ratio/high_max": 0.00218276544183027,
+      "clip_ratio/high_mean": 0.0009033630940393778,
+      "clip_ratio/low_mean": 0.0007663823653274449,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016697454775567167,
+      "epoch": 8.139941690962099,
+      "grad_norm": 0.15434731543064117,
+      "learning_rate": 1e-06,
+      "loss": -0.0201,
+      "step": 783
+    },
+    {
+      "clip_ratio/high_max": 0.0021151314649614505,
+      "clip_ratio/high_mean": 0.0007573006714665098,
+      "clip_ratio/low_mean": 0.000899002931873838,
+      "clip_ratio/low_min": 9.366901849716669e-05,
+      "clip_ratio/region_mean": 0.0016563036188017577,
+      "epoch": 8.14927113702624,
+      "grad_norm": 0.27759093046188354,
+      "learning_rate": 1e-06,
+      "loss": 0.0238,
+      "step": 784
+    },
+    {
+      "clip_ratio/high_max": 0.002407892912742682,
+      "clip_ratio/high_mean": 0.0009699102156446315,
+      "clip_ratio/low_mean": 0.0008394020132982405,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018093122307618614,
+      "epoch": 8.15860058309038,
+      "grad_norm": 0.12997399270534515,
+      "learning_rate": 1e-06,
+      "loss": -0.0494,
+      "step": 785
+    },
+    {
+      "clip_ratio/high_max": 0.0022966804062889423,
+      "clip_ratio/high_mean": 0.0009885221370495856,
+      "clip_ratio/low_mean": 0.0008676420920892269,
+      "clip_ratio/low_min": 6.781096453778446e-05,
+      "clip_ratio/region_mean": 0.0018561642282293178,
+      "epoch": 8.167930029154519,
+      "grad_norm": 0.12709791958332062,
+      "learning_rate": 1e-06,
+      "loss": -0.0519,
+      "step": 786
+    },
+    {
+      "clip_ratio/high_max": 0.0021771273750346154,
+      "clip_ratio/high_mean": 0.0009113561172853224,
+      "clip_ratio/low_mean": 0.0008371757976419758,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017485319331171922,
+      "epoch": 8.177259475218658,
+      "grad_norm": 0.19155512750148773,
+      "learning_rate": 1e-06,
+      "loss": -0.0444,
+      "step": 787
+    },
+    {
+      "clip_ratio/high_max": 0.001957065673195757,
+      "clip_ratio/high_mean": 0.0007083869313646574,
+      "clip_ratio/low_mean": 0.0008589770914113615,
+      "clip_ratio/low_min": 5.75090161873959e-05,
+      "clip_ratio/region_mean": 0.0015673640082241036,
+      "epoch": 8.186588921282798,
+      "grad_norm": 0.1266499161720276,
+      "learning_rate": 1e-06,
+      "loss": 0.0401,
+      "step": 788
+    },
+    {
+      "clip_ratio/high_max": 0.0024794062628643587,
+      "clip_ratio/high_mean": 0.0008836291708576027,
+      "clip_ratio/low_mean": 0.0009431578437215649,
+      "clip_ratio/low_min": 3.724418820638675e-05,
+      "clip_ratio/region_mean": 0.0018267870036652312,
+      "epoch": 8.19591836734694,
+      "grad_norm": 0.1369238942861557,
+      "learning_rate": 1e-06,
+      "loss": -0.0126,
+      "step": 789
+    },
+    {
+      "clip_ratio/high_max": 0.0021670952482963912,
+      "clip_ratio/high_mean": 0.0008826126722851768,
+      "clip_ratio/low_mean": 0.0008767210147198057,
+      "clip_ratio/low_min": 1.9060689737671055e-05,
+      "clip_ratio/region_mean": 0.00175933372520376,
+      "epoch": 8.205247813411079,
+      "grad_norm": 0.13367097079753876,
+      "learning_rate": 1e-06,
+      "loss": -0.0295,
+      "step": 790
+    },
+    {
+      "clip_ratio/high_max": 0.0023366394598269835,
+      "clip_ratio/high_mean": 0.001017025560940965,
+      "clip_ratio/low_mean": 0.0008878636945155449,
+      "clip_ratio/low_min": 1.8701375665841624e-05,
+      "clip_ratio/region_mean": 0.0019048892863793299,
+      "epoch": 8.214577259475218,
+      "grad_norm": 0.14439278841018677,
+      "learning_rate": 1e-06,
+      "loss": -0.008,
+      "step": 791
+    },
+    {
+      "clip_ratio/high_max": 0.00213179298225441,
+      "clip_ratio/high_mean": 0.0008897867573978147,
+      "clip_ratio/low_mean": 0.0009545487991999835,
+      "clip_ratio/low_min": 4.5175278501119465e-05,
+      "clip_ratio/region_mean": 0.0018443355584167875,
+      "epoch": 8.223906705539358,
+      "grad_norm": 0.1411471962928772,
+      "learning_rate": 1e-06,
+      "loss": 0.0154,
+      "step": 792
+    },
+    {
+      "clip_ratio/high_max": 0.0025781976110010874,
+      "clip_ratio/high_mean": 0.0010088125764013967,
+      "clip_ratio/low_mean": 0.0006651837893514312,
+      "clip_ratio/low_min": 2.7734635295928456e-05,
+      "clip_ratio/region_mean": 0.001673996368481312,
+      "epoch": 8.2332361516035,
+      "grad_norm": 0.1256917268037796,
+      "learning_rate": 1e-06,
+      "loss": -0.0382,
+      "step": 793
+    },
+    {
+      "clip_ratio/high_max": 0.002320486633834662,
+      "clip_ratio/high_mean": 0.0009754740822245367,
+      "clip_ratio/low_mean": 0.0007639980030944571,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017394720598531421,
+      "epoch": 8.242565597667639,
+      "grad_norm": 0.1294924020767212,
+      "learning_rate": 1e-06,
+      "loss": -0.0585,
+      "step": 794
+    },
+    {
+      "clip_ratio/high_max": 0.0025719086042954586,
+      "clip_ratio/high_mean": 0.0011508624993439298,
+      "clip_ratio/low_mean": 0.0007402381897918531,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001891100691864267,
+      "epoch": 8.251895043731778,
+      "grad_norm": 0.14637455344200134,
+      "learning_rate": 1e-06,
+      "loss": -0.0483,
+      "step": 795
+    },
+    {
+      "clip_ratio/high_max": 0.0021692998343496583,
+      "clip_ratio/high_mean": 0.0009030763594637392,
+      "clip_ratio/low_mean": 0.000868905979586998,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017719823736115359,
+      "epoch": 8.261224489795918,
+      "grad_norm": 0.12796564400196075,
+      "learning_rate": 1e-06,
+      "loss": -0.0128,
+      "step": 796
+    },
+    {
+      "clip_ratio/high_max": 0.002855377344531007,
+      "clip_ratio/high_mean": 0.0010504756482987432,
+      "clip_ratio/low_mean": 0.0009795163714443333,
+      "clip_ratio/low_min": 0.00010217772432952188,
+      "clip_ratio/region_mean": 0.0020299920142861083,
+      "epoch": 8.270553935860057,
+      "grad_norm": 0.14632874727249146,
+      "learning_rate": 1e-06,
+      "loss": 0.0078,
+      "step": 797
+    },
+    {
+      "clip_ratio/high_max": 0.0026055614071083255,
+      "clip_ratio/high_mean": 0.0009758277210494271,
+      "clip_ratio/low_mean": 0.0008346058320967131,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018104335467796773,
+      "epoch": 8.279883381924199,
+      "grad_norm": 0.14142385125160217,
+      "learning_rate": 1e-06,
+      "loss": -0.0234,
+      "step": 798
+    },
+    {
+      "clip_ratio/high_max": 0.0023912920150905848,
+      "clip_ratio/high_mean": 0.0010060902204713784,
+      "clip_ratio/low_mean": 0.0007692072795180138,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017752974963514134,
+      "epoch": 8.289212827988338,
+      "grad_norm": 0.13493581116199493,
+      "learning_rate": 1e-06,
+      "loss": -0.0447,
+      "step": 799
+    },
+    {
+      "clip_ratio/high_max": 0.0025832414685282856,
+      "clip_ratio/high_mean": 0.0009292196482419968,
+      "clip_ratio/low_mean": 0.0010997168410540326,
+      "clip_ratio/low_min": 5.557256918109488e-05,
+      "clip_ratio/region_mean": 0.002028936447459273,
+      "epoch": 8.298542274052478,
+      "grad_norm": 0.1228645071387291,
+      "learning_rate": 1e-06,
+      "loss": 0.007,
+      "step": 800
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0580008370535714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4072.0,
+      "completions/mean_length": 747.948974609375,
+      "completions/mean_terminated_length": 541.802490234375,
+      "completions/min_length": 95.0,
+      "completions/min_terminated_length": 95.0,
+      "epoch": 8.307871720116617,
+      "grad_norm": 0.1516033560037613,
+      "learning_rate": 1e-06,
+      "loss": -0.0105,
+      "num_tokens": 474366680.0,
+      "reward": 0.6474260687828064,
+      "reward_std": 0.1622728705406189,
+      "rewards/simpleverify_reward/mean": 0.6474260687828064,
+      "rewards/simpleverify_reward/std": 0.4777797758579254,
+      "step": 801
+    },
+    {
+      "clip_ratio/high_max": 0.0017537986641400494,
+      "clip_ratio/high_mean": 0.0006630350926570827,
+      "clip_ratio/low_mean": 0.00047387506856466644,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011369101484888233,
+      "epoch": 8.317201166180759,
+      "grad_norm": 0.14466431736946106,
+      "learning_rate": 1e-06,
+      "loss": 0.0116,
+      "step": 802
+    },
+    {
+      "clip_ratio/high_max": 0.0019687302847160026,
+      "clip_ratio/high_mean": 0.0008052143439272186,
+      "clip_ratio/low_mean": 0.0005499929629877442,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013552073251048569,
+      "epoch": 8.326530612244898,
+      "grad_norm": 0.1408139318227768,
+      "learning_rate": 1e-06,
+      "loss": -0.0174,
+      "step": 803
+    },
+    {
+      "clip_ratio/high_max": 0.002125952061760472,
+      "clip_ratio/high_mean": 0.0007655736026208615,
+      "clip_ratio/low_mean": 0.00043082632464575,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011963999022555072,
+      "epoch": 8.335860058309038,
+      "grad_norm": 0.12387403845787048,
+      "learning_rate": 1e-06,
+      "loss": -0.0215,
+      "step": 804
+    },
+    {
+      "clip_ratio/high_max": 0.0019115101858915295,
+      "clip_ratio/high_mean": 0.0008138671455526492,
+      "clip_ratio/low_mean": 0.0005153647325641941,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001329231894487748,
+      "epoch": 8.345189504373177,
+      "grad_norm": 0.13465812802314758,
+      "learning_rate": 1e-06,
+      "loss": -0.0301,
+      "step": 805
+    },
+    {
+      "clip_ratio/high_max": 0.0018984133057529107,
+      "clip_ratio/high_mean": 0.0007280560739673092,
+      "clip_ratio/low_mean": 0.0006429082936847408,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013709643790207338,
+      "epoch": 8.354518950437317,
+      "grad_norm": 0.14861463010311127,
+      "learning_rate": 1e-06,
+      "loss": 0.0125,
+      "step": 806
+    },
+    {
+      "clip_ratio/high_max": 0.0021903294109506533,
+      "clip_ratio/high_mean": 0.00087721385898476,
+      "clip_ratio/low_mean": 0.0005823174324177671,
+      "clip_ratio/low_min": 1.0235833542537875e-05,
+      "clip_ratio/region_mean": 0.0014595313004974741,
+      "epoch": 8.363848396501458,
+      "grad_norm": 0.13971786201000214,
+      "learning_rate": 1e-06,
+      "loss": -0.0437,
+      "step": 807
+    },
+    {
+      "clip_ratio/high_max": 0.0019012930497410707,
+      "clip_ratio/high_mean": 0.00080004764640762,
+      "clip_ratio/low_mean": 0.0007180351167335175,
+      "clip_ratio/low_min": 6.560621386597631e-05,
+      "clip_ratio/region_mean": 0.0015180827758740634,
+      "epoch": 8.373177842565598,
+      "grad_norm": 0.1959521472454071,
+      "learning_rate": 1e-06,
+      "loss": 0.0021,
+      "step": 808
+    },
+    {
+      "clip_ratio/high_max": 0.0021668398039764725,
+      "clip_ratio/high_mean": 0.0008851054335536901,
+      "clip_ratio/low_mean": 0.0007341410037042806,
+      "clip_ratio/low_min": 2.4323799152625725e-05,
+      "clip_ratio/region_mean": 0.0016192464827327058,
+      "epoch": 8.382507288629737,
+      "grad_norm": 0.16336694359779358,
+      "learning_rate": 1e-06,
+      "loss": -0.0438,
+      "step": 809
+    },
+    {
+      "clip_ratio/high_max": 0.002415544498944655,
+      "clip_ratio/high_mean": 0.0009118056823353982,
+      "clip_ratio/low_mean": 0.0005797651588181907,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014915708525222726,
+      "epoch": 8.391836734693877,
+      "grad_norm": 0.1311698853969574,
+      "learning_rate": 1e-06,
+      "loss": -0.0214,
+      "step": 810
+    },
+    {
+      "clip_ratio/high_max": 0.0023332574783125892,
+      "clip_ratio/high_mean": 0.0008412513579969527,
+      "clip_ratio/low_mean": 0.0006205119052538066,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014617633059970103,
+      "epoch": 8.401166180758018,
+      "grad_norm": 0.1968589723110199,
+      "learning_rate": 1e-06,
+      "loss": -0.025,
+      "step": 811
+    },
+    {
+      "clip_ratio/high_max": 0.0020871867855021264,
+      "clip_ratio/high_mean": 0.0008100756458588876,
+      "clip_ratio/low_mean": 0.0009123428335442441,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017224184848600999,
+      "epoch": 8.410495626822158,
+      "grad_norm": 0.17827925086021423,
+      "learning_rate": 1e-06,
+      "loss": 0.0001,
+      "step": 812
+    },
+    {
+      "clip_ratio/high_max": 0.0022598059367737733,
+      "clip_ratio/high_mean": 0.0009513376862742007,
+      "clip_ratio/low_mean": 0.000689091301865119,
+      "clip_ratio/low_min": 3.802284572884673e-05,
+      "clip_ratio/region_mean": 0.0016404290072387084,
+      "epoch": 8.419825072886297,
+      "grad_norm": 0.12976522743701935,
+      "learning_rate": 1e-06,
+      "loss": -0.0408,
+      "step": 813
+    },
+    {
+      "clip_ratio/high_max": 0.002414627371763345,
+      "clip_ratio/high_mean": 0.0009501463482592953,
+      "clip_ratio/low_mean": 0.0007094226825756778,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016595690321992151,
+      "epoch": 8.429154518950437,
+      "grad_norm": 0.13414177298545837,
+      "learning_rate": 1e-06,
+      "loss": -0.0379,
+      "step": 814
+    },
+    {
+      "clip_ratio/high_max": 0.0020716543585876934,
+      "clip_ratio/high_mean": 0.0009414587057108292,
+      "clip_ratio/low_mean": 0.0007051264256006107,
+      "clip_ratio/low_min": 2.5699013349367306e-05,
+      "clip_ratio/region_mean": 0.0016465851222164929,
+      "epoch": 8.438483965014576,
+      "grad_norm": 0.131557434797287,
+      "learning_rate": 1e-06,
+      "loss": -0.0194,
+      "step": 815
+    },
+    {
+      "clip_ratio/high_max": 0.0020102107628190424,
+      "clip_ratio/high_mean": 0.000899550527719839,
+      "clip_ratio/low_mean": 0.0007386505203612614,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016382010471716058,
+      "epoch": 8.447813411078718,
+      "grad_norm": 0.14207573235034943,
+      "learning_rate": 1e-06,
+      "loss": -0.0376,
+      "step": 816
+    },
+    {
+      "clip_ratio/high_max": 0.0020864536418230273,
+      "clip_ratio/high_mean": 0.0008691776092746295,
+      "clip_ratio/low_mean": 0.0009405826713191345,
+      "clip_ratio/low_min": 1.6648908058414236e-05,
+      "clip_ratio/region_mean": 0.001809760280593764,
+      "epoch": 8.457142857142857,
+      "grad_norm": 0.13197103142738342,
+      "learning_rate": 1e-06,
+      "loss": -0.0235,
+      "step": 817
+    },
+    {
+      "clip_ratio/high_max": 0.0022347010344674345,
+      "clip_ratio/high_mean": 0.0008625889913673745,
+      "clip_ratio/low_mean": 0.0008217213689931668,
+      "clip_ratio/low_min": 3.0652281566290185e-05,
+      "clip_ratio/region_mean": 0.0016843103876453824,
+      "epoch": 8.466472303206997,
+      "grad_norm": 0.1277029812335968,
+      "learning_rate": 1e-06,
+      "loss": -0.0221,
+      "step": 818
+    },
+    {
+      "clip_ratio/high_max": 0.0024130923120537773,
+      "clip_ratio/high_mean": 0.001104865896195406,
+      "clip_ratio/low_mean": 0.0008431018450210104,
+      "clip_ratio/low_min": 1.2138279089413118e-05,
+      "clip_ratio/region_mean": 0.001947967779415194,
+      "epoch": 8.475801749271136,
+      "grad_norm": 0.15858964622020721,
+      "learning_rate": 1e-06,
+      "loss": -0.0702,
+      "step": 819
+    },
+    {
+      "clip_ratio/high_max": 0.0026420907088322565,
+      "clip_ratio/high_mean": 0.0010906054685619893,
+      "clip_ratio/low_mean": 0.000783914281782927,
+      "clip_ratio/low_min": 1.3368983672989998e-05,
+      "clip_ratio/region_mean": 0.0018745197376119904,
+      "epoch": 8.485131195335278,
+      "grad_norm": 0.15172161161899567,
+      "learning_rate": 1e-06,
+      "loss": -0.0132,
+      "step": 820
+    },
+    {
+      "clip_ratio/high_max": 0.002111362449795706,
+      "clip_ratio/high_mean": 0.0009811183545025415,
+      "clip_ratio/low_mean": 0.0009079625669983216,
+      "clip_ratio/low_min": 1.2178487850178499e-05,
+      "clip_ratio/region_mean": 0.0018890809296863154,
+      "epoch": 8.494460641399417,
+      "grad_norm": 0.14209824800491333,
+      "learning_rate": 1e-06,
+      "loss": -0.0104,
+      "step": 821
+    },
+    {
+      "clip_ratio/high_max": 0.0021146593426237814,
+      "clip_ratio/high_mean": 0.0009244757911801571,
+      "clip_ratio/low_mean": 0.0008680353193994961,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017925111278600525,
+      "epoch": 8.503790087463557,
+      "grad_norm": 0.13806912302970886,
+      "learning_rate": 1e-06,
+      "loss": 0.0008,
+      "step": 822
+    },
+    {
+      "clip_ratio/high_max": 0.001988493932003621,
+      "clip_ratio/high_mean": 0.0008030673452594783,
+      "clip_ratio/low_mean": 0.0009028253880387638,
+      "clip_ratio/low_min": 4.0526813791075256e-05,
+      "clip_ratio/region_mean": 0.0017058927405741997,
+      "epoch": 8.513119533527696,
+      "grad_norm": 0.13951145112514496,
+      "learning_rate": 1e-06,
+      "loss": 0.0021,
+      "step": 823
+    },
+    {
+      "clip_ratio/high_max": 0.0024955428452813067,
+      "clip_ratio/high_mean": 0.0009691582217783434,
+      "clip_ratio/low_mean": 0.0009108888134505833,
+      "clip_ratio/low_min": 4.585593705996871e-05,
+      "clip_ratio/region_mean": 0.0018800470352289267,
+      "epoch": 8.522448979591836,
+      "grad_norm": 0.14607743918895721,
+      "learning_rate": 1e-06,
+      "loss": -0.0383,
+      "step": 824
+    },
+    {
+      "clip_ratio/high_max": 0.00227100537813385,
+      "clip_ratio/high_mean": 0.0009037887102749664,
+      "clip_ratio/low_mean": 0.0007138600303733256,
+      "clip_ratio/low_min": 2.584780850156676e-05,
+      "clip_ratio/region_mean": 0.0016176487442862708,
+      "epoch": 8.531778425655977,
+      "grad_norm": 0.12231608480215073,
+      "learning_rate": 1e-06,
+      "loss": -0.0312,
+      "step": 825
+    },
+    {
+      "clip_ratio/high_max": 0.002586353599326685,
+      "clip_ratio/high_mean": 0.001039758000842994,
+      "clip_ratio/low_mean": 0.0008937632101151394,
+      "clip_ratio/low_min": 3.640069917310029e-05,
+      "clip_ratio/region_mean": 0.0019335211982252076,
+      "epoch": 8.541107871720117,
+      "grad_norm": 0.13818369805812836,
+      "learning_rate": 1e-06,
+      "loss": -0.0298,
+      "step": 826
+    },
+    {
+      "clip_ratio/high_max": 0.0021612934942822903,
+      "clip_ratio/high_mean": 0.0010622190238791518,
+      "clip_ratio/low_mean": 0.0010118979116668925,
+      "clip_ratio/low_min": 4.829984391108155e-05,
+      "clip_ratio/region_mean": 0.002074116862786468,
+      "epoch": 8.550437317784256,
+      "grad_norm": 0.1717173010110855,
+      "learning_rate": 1e-06,
+      "loss": -0.0393,
+      "step": 827
+    },
+    {
+      "clip_ratio/high_max": 0.0023573423459311016,
+      "clip_ratio/high_mean": 0.0009583396094967611,
+      "clip_ratio/low_mean": 0.0009784350786503637,
+      "clip_ratio/low_min": 7.57310390326893e-05,
+      "clip_ratio/region_mean": 0.0019367746681382414,
+      "epoch": 8.559766763848396,
+      "grad_norm": 0.17738987505435944,
+      "learning_rate": 1e-06,
+      "loss": 0.0059,
+      "step": 828
+    },
+    {
+      "clip_ratio/high_max": 0.0025482142955297604,
+      "clip_ratio/high_mean": 0.000998525614704704,
+      "clip_ratio/low_mean": 0.0008901449473341927,
+      "clip_ratio/low_min": 2.2599891963182017e-05,
+      "clip_ratio/region_mean": 0.0018886705656768754,
+      "epoch": 8.569096209912537,
+      "grad_norm": 0.14093036949634552,
+      "learning_rate": 1e-06,
+      "loss": -0.0091,
+      "step": 829
+    },
+    {
+      "clip_ratio/high_max": 0.0021053235614090227,
+      "clip_ratio/high_mean": 0.0009183491893054452,
+      "clip_ratio/low_mean": 0.000790588781455881,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017089379689423367,
+      "epoch": 8.578425655976677,
+      "grad_norm": 0.126109778881073,
+      "learning_rate": 1e-06,
+      "loss": -0.0194,
+      "step": 830
+    },
+    {
+      "clip_ratio/high_max": 0.002523310438846238,
+      "clip_ratio/high_mean": 0.0009514377707091626,
+      "clip_ratio/low_mean": 0.0009702378229121678,
+      "clip_ratio/low_min": 4.616568094206741e-05,
+      "clip_ratio/region_mean": 0.001921675582707394,
+      "epoch": 8.587755102040816,
+      "grad_norm": 0.14047396183013916,
+      "learning_rate": 1e-06,
+      "loss": -0.0247,
+      "step": 831
+    },
+    {
+      "clip_ratio/high_max": 0.003317179507575929,
+      "clip_ratio/high_mean": 0.0011828214010165539,
+      "clip_ratio/low_mean": 0.000908896574856044,
+      "clip_ratio/low_min": 5.2624192903749645e-05,
+      "clip_ratio/region_mean": 0.002091717913572211,
+      "epoch": 8.597084548104956,
+      "grad_norm": 0.14829492568969727,
+      "learning_rate": 1e-06,
+      "loss": -0.0135,
+      "step": 832
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0633370535714286,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4092.0,
+      "completions/mean_length": 771.0543212890625,
+      "completions/mean_terminated_length": 546.2217407226562,
+      "completions/min_length": 5.0,
+      "completions/min_terminated_length": 5.0,
+      "epoch": 8.606413994169095,
+      "grad_norm": 0.1405651569366455,
+      "learning_rate": 1e-06,
+      "loss": -0.0467,
+      "num_tokens": 491854571.0,
+      "reward": 0.6417410969734192,
+      "reward_std": 0.15736746788024902,
+      "rewards/simpleverify_reward/mean": 0.6417410969734192,
+      "rewards/simpleverify_reward/std": 0.47949716448783875,
+      "step": 833
+    },
+    {
+      "clip_ratio/high_max": 0.0020139113257755525,
+      "clip_ratio/high_mean": 0.0008633421166450717,
+      "clip_ratio/low_mean": 0.00046989334896352375,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001333235461061122,
+      "epoch": 8.615743440233237,
+      "grad_norm": 0.13367526233196259,
+      "learning_rate": 1e-06,
+      "loss": -0.0218,
+      "step": 834
+    },
+    {
+      "clip_ratio/high_max": 0.002141184486390557,
+      "clip_ratio/high_mean": 0.0008294206363643752,
+      "clip_ratio/low_mean": 0.00038089149211373297,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012103121262043715,
+      "epoch": 8.625072886297376,
+      "grad_norm": 0.13916561007499695,
+      "learning_rate": 1e-06,
+      "loss": -0.0267,
+      "step": 835
+    },
+    {
+      "clip_ratio/high_max": 0.001740491181408288,
+      "clip_ratio/high_mean": 0.0007440051322191721,
+      "clip_ratio/low_mean": 0.0005984672025078908,
+      "clip_ratio/low_min": 4.800307215191424e-05,
+      "clip_ratio/region_mean": 0.001342472358373925,
+      "epoch": 8.634402332361516,
+      "grad_norm": 0.15485632419586182,
+      "learning_rate": 1e-06,
+      "loss": -0.0397,
+      "step": 836
+    },
+    {
+      "clip_ratio/high_max": 0.002032392942055594,
+      "clip_ratio/high_mean": 0.0007078141152305761,
+      "clip_ratio/low_mean": 0.0005370698117985739,
+      "clip_ratio/low_min": 0.0001028032766043907,
+      "clip_ratio/region_mean": 0.0012448839224816766,
+      "epoch": 8.643731778425655,
+      "grad_norm": 0.13908182084560394,
+      "learning_rate": 1e-06,
+      "loss": -0.0203,
+      "step": 837
+    },
+    {
+      "clip_ratio/high_max": 0.0017979219264816493,
+      "clip_ratio/high_mean": 0.0007961944247654174,
+      "clip_ratio/low_mean": 0.0005261639134914731,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013223583191575017,
+      "epoch": 8.653061224489797,
+      "grad_norm": 0.12595003843307495,
+      "learning_rate": 1e-06,
+      "loss": -0.0343,
+      "step": 838
+    },
+    {
+      "clip_ratio/high_max": 0.002292386212502606,
+      "clip_ratio/high_mean": 0.0009561050537740812,
+      "clip_ratio/low_mean": 0.0005483322347572539,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015044372557895258,
+      "epoch": 8.662390670553936,
+      "grad_norm": 0.14232099056243896,
+      "learning_rate": 1e-06,
+      "loss": -0.0391,
+      "step": 839
+    },
+    {
+      "clip_ratio/high_max": 0.001992901903577149,
+      "clip_ratio/high_mean": 0.0009374034398206277,
+      "clip_ratio/low_mean": 0.0004780116523761535,
+      "clip_ratio/low_min": 1.5972400433383882e-05,
+      "clip_ratio/region_mean": 0.0014154150921967812,
+      "epoch": 8.671720116618076,
+      "grad_norm": 0.17381148040294647,
+      "learning_rate": 1e-06,
+      "loss": -0.0247,
+      "step": 840
+    },
+    {
+      "clip_ratio/high_max": 0.0020656290325860027,
+      "clip_ratio/high_mean": 0.0008125649655994494,
+      "clip_ratio/low_mean": 0.0005825711668876465,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013951361397630535,
+      "epoch": 8.681049562682215,
+      "grad_norm": 0.1460714489221573,
+      "learning_rate": 1e-06,
+      "loss": 0.0085,
+      "step": 841
+    },
+    {
+      "clip_ratio/high_max": 0.0019548289055819623,
+      "clip_ratio/high_mean": 0.0007950219624035526,
+      "clip_ratio/low_mean": 0.0007099590866346261,
+      "clip_ratio/low_min": 2.2587639250559732e-05,
+      "clip_ratio/region_mean": 0.0015049810644995887,
+      "epoch": 8.690379008746355,
+      "grad_norm": 0.15236659348011017,
+      "learning_rate": 1e-06,
+      "loss": -0.0208,
+      "step": 842
+    },
+    {
+      "clip_ratio/high_max": 0.0020667382632382214,
+      "clip_ratio/high_mean": 0.0007644728430022951,
+      "clip_ratio/low_mean": 0.0007049965988699114,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014694694291392807,
+      "epoch": 8.699708454810496,
+      "grad_norm": 0.13557924330234528,
+      "learning_rate": 1e-06,
+      "loss": 0.0126,
+      "step": 843
+    },
+    {
+      "clip_ratio/high_max": 0.0024244647356681526,
+      "clip_ratio/high_mean": 0.0009140740894508781,
+      "clip_ratio/low_mean": 0.0006336965607260936,
+      "clip_ratio/low_min": 2.9460286896210164e-05,
+      "clip_ratio/region_mean": 0.0015477706692763604,
+      "epoch": 8.709037900874636,
+      "grad_norm": 0.12969401478767395,
+      "learning_rate": 1e-06,
+      "loss": -0.0196,
+      "step": 844
+    },
+    {
+      "clip_ratio/high_max": 0.001969263033970492,
+      "clip_ratio/high_mean": 0.0007806028243066976,
+      "clip_ratio/low_mean": 0.0006583489393960917,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001438951752788853,
+      "epoch": 8.718367346938775,
+      "grad_norm": 0.12705589830875397,
+      "learning_rate": 1e-06,
+      "loss": -0.0223,
+      "step": 845
+    },
+    {
+      "clip_ratio/high_max": 0.0023710055902483873,
+      "clip_ratio/high_mean": 0.0009322769874415826,
+      "clip_ratio/low_mean": 0.0007197825716502848,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016520596036571078,
+      "epoch": 8.727696793002915,
+      "grad_norm": 0.14861077070236206,
+      "learning_rate": 1e-06,
+      "loss": -0.0123,
+      "step": 846
+    },
+    {
+      "clip_ratio/high_max": 0.002109671004291158,
+      "clip_ratio/high_mean": 0.0008622139730505296,
+      "clip_ratio/low_mean": 0.0007059752788336482,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015681892873544712,
+      "epoch": 8.737026239067056,
+      "grad_norm": 0.13615943491458893,
+      "learning_rate": 1e-06,
+      "loss": -0.0244,
+      "step": 847
+    },
+    {
+      "clip_ratio/high_max": 0.0022711220517521724,
+      "clip_ratio/high_mean": 0.0009589893797965487,
+      "clip_ratio/low_mean": 0.0008089785569609376,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017679679294815287,
+      "epoch": 8.746355685131196,
+      "grad_norm": 0.14134782552719116,
+      "learning_rate": 1e-06,
+      "loss": -0.0467,
+      "step": 848
+    },
+    {
+      "clip_ratio/high_max": 0.0025540151946188416,
+      "clip_ratio/high_mean": 0.0009891991085169138,
+      "clip_ratio/low_mean": 0.0007917930579424137,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017809922064770944,
+      "epoch": 8.755685131195335,
+      "grad_norm": 0.13083255290985107,
+      "learning_rate": 1e-06,
+      "loss": -0.0139,
+      "step": 849
+    },
+    {
+      "clip_ratio/high_max": 0.0023866647825343534,
+      "clip_ratio/high_mean": 0.0009160980953311082,
+      "clip_ratio/low_mean": 0.0006603442161576822,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001576442340592621,
+      "epoch": 8.765014577259475,
+      "grad_norm": 0.1456131786108017,
+      "learning_rate": 1e-06,
+      "loss": -0.0592,
+      "step": 850
+    },
+    {
+      "clip_ratio/high_max": 0.002494656575436238,
+      "clip_ratio/high_mean": 0.0009352633715025149,
+      "clip_ratio/low_mean": 0.0007715230840403819,
+      "clip_ratio/low_min": 2.8331822250038385e-05,
+      "clip_ratio/region_mean": 0.0017067864682758227,
+      "epoch": 8.774344023323614,
+      "grad_norm": 0.12012716382741928,
+      "learning_rate": 1e-06,
+      "loss": 0.0012,
+      "step": 851
+    },
+    {
+      "clip_ratio/high_max": 0.0022803933534305543,
+      "clip_ratio/high_mean": 0.0009966295747290133,
+      "clip_ratio/low_mean": 0.0007608390424138634,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017574685989529826,
+      "epoch": 8.783673469387756,
+      "grad_norm": 0.1424596756696701,
+      "learning_rate": 1e-06,
+      "loss": -0.0378,
+      "step": 852
+    },
+    {
+      "clip_ratio/high_max": 0.002275281192851253,
+      "clip_ratio/high_mean": 0.0009698879621282686,
+      "clip_ratio/low_mean": 0.000846161432491499,
+      "clip_ratio/low_min": 1.1913839443877805e-05,
+      "clip_ratio/region_mean": 0.0018160494291805662,
+      "epoch": 8.793002915451895,
+      "grad_norm": 0.13369524478912354,
+      "learning_rate": 1e-06,
+      "loss": -0.04,
+      "step": 853
+    },
+    {
+      "clip_ratio/high_max": 0.0024288226341013797,
+      "clip_ratio/high_mean": 0.0010673616961867083,
+      "clip_ratio/low_mean": 0.000930726502701873,
+      "clip_ratio/low_min": 4.8840844101505354e-05,
+      "clip_ratio/region_mean": 0.001998088257096242,
+      "epoch": 8.802332361516035,
+      "grad_norm": 0.12861855328083038,
+      "learning_rate": 1e-06,
+      "loss": -0.031,
+      "step": 854
+    },
+    {
+      "clip_ratio/high_max": 0.002092519549478311,
+      "clip_ratio/high_mean": 0.0008946466059569502,
+      "clip_ratio/low_mean": 0.00072539451252851,
+      "clip_ratio/low_min": 2.8026905056321993e-05,
+      "clip_ratio/region_mean": 0.001620041140995454,
+      "epoch": 8.811661807580174,
+      "grad_norm": 0.1658993512392044,
+      "learning_rate": 1e-06,
+      "loss": -0.0357,
+      "step": 855
+    },
+    {
+      "clip_ratio/high_max": 0.002359290658205282,
+      "clip_ratio/high_mean": 0.0009629812084313016,
+      "clip_ratio/low_mean": 0.0009085323217732366,
+      "clip_ratio/low_min": 3.4712578781181946e-05,
+      "clip_ratio/region_mean": 0.0018715135083766654,
+      "epoch": 8.820991253644316,
+      "grad_norm": 0.16082775592803955,
+      "learning_rate": 1e-06,
+      "loss": -0.006,
+      "step": 856
+    },
+    {
+      "clip_ratio/high_max": 0.0028039188619004562,
+      "clip_ratio/high_mean": 0.00101169560366543,
+      "clip_ratio/low_mean": 0.0009264432374038734,
+      "clip_ratio/low_min": 5.167424387764186e-05,
+      "clip_ratio/region_mean": 0.0019381388410693035,
+      "epoch": 8.830320699708455,
+      "grad_norm": 0.13666211068630219,
+      "learning_rate": 1e-06,
+      "loss": -0.0073,
+      "step": 857
+    },
+    {
+      "clip_ratio/high_max": 0.002292525088705588,
+      "clip_ratio/high_mean": 0.0009539639413560508,
+      "clip_ratio/low_mean": 0.0007223619941214565,
+      "clip_ratio/low_min": 2.9164722946006805e-05,
+      "clip_ratio/region_mean": 0.0016763259409344755,
+      "epoch": 8.839650145772595,
+      "grad_norm": 0.12612462043762207,
+      "learning_rate": 1e-06,
+      "loss": -0.0168,
+      "step": 858
+    },
+    {
+      "clip_ratio/high_max": 0.002549354379880242,
+      "clip_ratio/high_mean": 0.0008722816619410878,
+      "clip_ratio/low_mean": 0.0008769092128204647,
+      "clip_ratio/low_min": 3.9382481190841645e-05,
+      "clip_ratio/region_mean": 0.0017491908874944784,
+      "epoch": 8.848979591836734,
+      "grad_norm": 0.13609710335731506,
+      "learning_rate": 1e-06,
+      "loss": 0.0032,
+      "step": 859
+    },
+    {
+      "clip_ratio/high_max": 0.0024448685035167728,
+      "clip_ratio/high_mean": 0.0009766333550942363,
+      "clip_ratio/low_mean": 0.0007997267639439087,
+      "clip_ratio/low_min": 6.836960710643325e-05,
+      "clip_ratio/region_mean": 0.0017763601208571345,
+      "epoch": 8.858309037900874,
+      "grad_norm": 0.13199026882648468,
+      "learning_rate": 1e-06,
+      "loss": -0.0264,
+      "step": 860
+    },
+    {
+      "clip_ratio/high_max": 0.002441056858515367,
+      "clip_ratio/high_mean": 0.0010800373784149997,
+      "clip_ratio/low_mean": 0.0008245835870184237,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019046209490625188,
+      "epoch": 8.867638483965015,
+      "grad_norm": 0.14391858875751495,
+      "learning_rate": 1e-06,
+      "loss": -0.0303,
+      "step": 861
+    },
+    {
+      "clip_ratio/high_max": 0.0022580993027077056,
+      "clip_ratio/high_mean": 0.0009184454575006384,
+      "clip_ratio/low_mean": 0.0008429733952652896,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017614188473089598,
+      "epoch": 8.876967930029155,
+      "grad_norm": 0.13431082665920258,
+      "learning_rate": 1e-06,
+      "loss": -0.0231,
+      "step": 862
+    },
+    {
+      "clip_ratio/high_max": 0.0022402869908546563,
+      "clip_ratio/high_mean": 0.0009044435319083277,
+      "clip_ratio/low_mean": 0.0009824261505855247,
+      "clip_ratio/low_min": 6.632088843616657e-05,
+      "clip_ratio/region_mean": 0.0018868696497520432,
+      "epoch": 8.886297376093294,
+      "grad_norm": 0.12800784409046173,
+      "learning_rate": 1e-06,
+      "loss": -0.0093,
+      "step": 863
+    },
+    {
+      "clip_ratio/high_max": 0.0023474494300899096,
+      "clip_ratio/high_mean": 0.0009831388015300035,
+      "clip_ratio/low_mean": 0.0009691117556940299,
+      "clip_ratio/low_min": 2.5206694772350602e-05,
+      "clip_ratio/region_mean": 0.0019522505317581818,
+      "epoch": 8.895626822157434,
+      "grad_norm": 0.15675704181194305,
+      "learning_rate": 1e-06,
+      "loss": -0.0218,
+      "step": 864
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0659528459821429,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4088.0,
+      "completions/mean_length": 784.3197631835938,
+      "completions/mean_terminated_length": 550.4828491210938,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "epoch": 9.00932944606414,
+      "grad_norm": 0.14521193504333496,
+      "learning_rate": 1e-06,
+      "loss": 0.0117,
+      "num_tokens": 509499180.0,
+      "reward": 0.641671359539032,
+      "reward_std": 0.16063126921653748,
+      "rewards/simpleverify_reward/mean": 0.6416712999343872,
+      "rewards/simpleverify_reward/std": 0.47951772809028625,
+      "step": 865
+    },
+    {
+      "clip_ratio/high_max": 0.0024364194250665605,
+      "clip_ratio/high_mean": 0.0009134492465818767,
+      "clip_ratio/low_mean": 0.00040719718981563346,
+      "clip_ratio/low_min": 1.4695508980366867e-05,
+      "clip_ratio/region_mean": 0.0013206464282120578,
+      "epoch": 9.018658892128279,
+      "grad_norm": 0.16433118283748627,
+      "learning_rate": 1e-06,
+      "loss": -0.0653,
+      "step": 866
+    },
+    {
+      "clip_ratio/high_max": 0.0019430474858381785,
+      "clip_ratio/high_mean": 0.0007149408756959019,
+      "clip_ratio/low_mean": 0.0004546089248833596,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001169549777841894,
+      "epoch": 9.02798833819242,
+      "grad_norm": 0.13100212812423706,
+      "learning_rate": 1e-06,
+      "loss": -0.0202,
+      "step": 867
+    },
+    {
+      "clip_ratio/high_max": 0.0018072276361635886,
+      "clip_ratio/high_mean": 0.0007499644798372174,
+      "clip_ratio/low_mean": 0.0005435647397007415,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001293529203394428,
+      "epoch": 9.03731778425656,
+      "grad_norm": 0.1497403234243393,
+      "learning_rate": 1e-06,
+      "loss": -0.0411,
+      "step": 868
+    },
+    {
+      "clip_ratio/high_max": 0.0017921896287589334,
+      "clip_ratio/high_mean": 0.000762457370001357,
+      "clip_ratio/low_mean": 0.0005329809991962975,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001295438389206538,
+      "epoch": 9.0466472303207,
+      "grad_norm": 0.1435271054506302,
+      "learning_rate": 1e-06,
+      "loss": -0.0211,
+      "step": 869
+    },
+    {
+      "clip_ratio/high_max": 0.002030614428804256,
+      "clip_ratio/high_mean": 0.000821765683213016,
+      "clip_ratio/low_mean": 0.000541309720119898,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013630753892357461,
+      "epoch": 9.055976676384839,
+      "grad_norm": 0.1355751007795334,
+      "learning_rate": 1e-06,
+      "loss": -0.037,
+      "step": 870
+    },
+    {
+      "clip_ratio/high_max": 0.002037711939919973,
+      "clip_ratio/high_mean": 0.000821711926619173,
+      "clip_ratio/low_mean": 0.0006188861989357974,
+      "clip_ratio/low_min": 1.001121290755691e-05,
+      "clip_ratio/region_mean": 0.0014405981200980023,
+      "epoch": 9.06530612244898,
+      "grad_norm": 0.1574908196926117,
+      "learning_rate": 1e-06,
+      "loss": -0.0352,
+      "step": 871
+    },
+    {
+      "clip_ratio/high_max": 0.0021282535781210754,
+      "clip_ratio/high_mean": 0.000931289270738489,
+      "clip_ratio/low_mean": 0.0006665466107733664,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015978358787833713,
+      "epoch": 9.07463556851312,
+      "grad_norm": 0.15857058763504028,
+      "learning_rate": 1e-06,
+      "loss": -0.002,
+      "step": 872
+    },
+    {
+      "clip_ratio/high_max": 0.002413314927252941,
+      "clip_ratio/high_mean": 0.0008469961994705955,
+      "clip_ratio/low_mean": 0.0008364817949768621,
+      "clip_ratio/low_min": 2.830615994753316e-05,
+      "clip_ratio/region_mean": 0.0016834779817145318,
+      "epoch": 9.08396501457726,
+      "grad_norm": 0.13699011504650116,
+      "learning_rate": 1e-06,
+      "loss": 0.0012,
+      "step": 873
+    },
+    {
+      "clip_ratio/high_max": 0.001854141904914286,
+      "clip_ratio/high_mean": 0.0007477157305402216,
+      "clip_ratio/low_mean": 0.0006288414078881033,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001376557171170134,
+      "epoch": 9.093294460641399,
+      "grad_norm": 0.13408535718917847,
+      "learning_rate": 1e-06,
+      "loss": -0.0174,
+      "step": 874
+    },
+    {
+      "clip_ratio/high_max": 0.001909519916807767,
+      "clip_ratio/high_mean": 0.0007421680829793331,
+      "clip_ratio/low_mean": 0.0007377129140877514,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001479881026170915,
+      "epoch": 9.102623906705539,
+      "grad_norm": 0.1423283964395523,
+      "learning_rate": 1e-06,
+      "loss": 0.0104,
+      "step": 875
+    },
+    {
+      "clip_ratio/high_max": 0.0023590355485794134,
+      "clip_ratio/high_mean": 0.0010331771973142168,
+      "clip_ratio/low_mean": 0.0006847015729363193,
+      "clip_ratio/low_min": 8.903273919713683e-05,
+      "clip_ratio/region_mean": 0.0017178787747980095,
+      "epoch": 9.11195335276968,
+      "grad_norm": 0.12830820679664612,
+      "learning_rate": 1e-06,
+      "loss": -0.0576,
+      "step": 876
+    },
+    {
+      "clip_ratio/high_max": 0.002145286292943638,
+      "clip_ratio/high_mean": 0.0009106013840209926,
+      "clip_ratio/low_mean": 0.000685119616719021,
+      "clip_ratio/low_min": 2.841599052771926e-05,
+      "clip_ratio/region_mean": 0.0015957209579937626,
+      "epoch": 9.12128279883382,
+      "grad_norm": 0.1287694126367569,
+      "learning_rate": 1e-06,
+      "loss": -0.0183,
+      "step": 877
+    },
+    {
+      "clip_ratio/high_max": 0.0021193885659158695,
+      "clip_ratio/high_mean": 0.0009188613312289817,
+      "clip_ratio/low_mean": 0.0008369850747840246,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001755846431478858,
+      "epoch": 9.130612244897959,
+      "grad_norm": 0.1833963245153427,
+      "learning_rate": 1e-06,
+      "loss": -0.0609,
+      "step": 878
+    },
+    {
+      "clip_ratio/high_max": 0.002296690221555764,
+      "clip_ratio/high_mean": 0.001003718052743352,
+      "clip_ratio/low_mean": 0.0008277043680209317,
+      "clip_ratio/low_min": 6.366182788042352e-05,
+      "clip_ratio/region_mean": 0.001831422410759842,
+      "epoch": 9.139941690962099,
+      "grad_norm": 0.1349063366651535,
+      "learning_rate": 1e-06,
+      "loss": -0.0221,
+      "step": 879
+    },
+    {
+      "clip_ratio/high_max": 0.002385257885180181,
+      "clip_ratio/high_mean": 0.0009333109192084521,
+      "clip_ratio/low_mean": 0.0007122137849364663,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016455246986879501,
+      "epoch": 9.14927113702624,
+      "grad_norm": 0.12756817042827606,
+      "learning_rate": 1e-06,
+      "loss": -0.0318,
+      "step": 880
+    },
+    {
+      "clip_ratio/high_max": 0.00244465890864376,
+      "clip_ratio/high_mean": 0.0009807757523958571,
+      "clip_ratio/low_mean": 0.0009915324444591533,
+      "clip_ratio/low_min": 1.4259639101510402e-05,
+      "clip_ratio/region_mean": 0.0019723081932170317,
+      "epoch": 9.15860058309038,
+      "grad_norm": 0.1390228271484375,
+      "learning_rate": 1e-06,
+      "loss": -0.0122,
+      "step": 881
+    },
+    {
+      "clip_ratio/high_max": 0.002297080136486329,
+      "clip_ratio/high_mean": 0.000955383791733766,
+      "clip_ratio/low_mean": 0.0009365522073494503,
+      "clip_ratio/low_min": 5.11177750013303e-05,
+      "clip_ratio/region_mean": 0.0018919359499705024,
+      "epoch": 9.167930029154519,
+      "grad_norm": 0.13791263103485107,
+      "learning_rate": 1e-06,
+      "loss": 0.0105,
+      "step": 882
+    },
+    {
+      "clip_ratio/high_max": 0.002606173169624526,
+      "clip_ratio/high_mean": 0.0011081011762144044,
+      "clip_ratio/low_mean": 0.0007196222613856662,
+      "clip_ratio/low_min": 1.628028076083865e-05,
+      "clip_ratio/region_mean": 0.0018277234412380494,
+      "epoch": 9.177259475218658,
+      "grad_norm": 0.13819114863872528,
+      "learning_rate": 1e-06,
+      "loss": -0.0547,
+      "step": 883
+    },
+    {
+      "clip_ratio/high_max": 0.002367043045524042,
+      "clip_ratio/high_mean": 0.0008472119334328454,
+      "clip_ratio/low_mean": 0.000717189993338252,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015644018567400053,
+      "epoch": 9.186588921282798,
+      "grad_norm": 0.13511182367801666,
+      "learning_rate": 1e-06,
+      "loss": -0.0331,
+      "step": 884
+    },
+    {
+      "clip_ratio/high_max": 0.0023582188223372214,
+      "clip_ratio/high_mean": 0.0009955569930752972,
+      "clip_ratio/low_mean": 0.000746332599192101,
+      "clip_ratio/low_min": 1.848292231443338e-05,
+      "clip_ratio/region_mean": 0.0017418895695300307,
+      "epoch": 9.19591836734694,
+      "grad_norm": 0.14253586530685425,
+      "learning_rate": 1e-06,
+      "loss": 0.0025,
+      "step": 885
+    },
+    {
+      "clip_ratio/high_max": 0.0023443693935405463,
+      "clip_ratio/high_mean": 0.0009998419973271666,
+      "clip_ratio/low_mean": 0.0008602960460848408,
+      "clip_ratio/low_min": 4.051099676871672e-05,
+      "clip_ratio/region_mean": 0.0018601380397740286,
+      "epoch": 9.205247813411079,
+      "grad_norm": 0.1792212277650833,
+      "learning_rate": 1e-06,
+      "loss": 0.0204,
+      "step": 886
+    },
+    {
+      "clip_ratio/high_max": 0.002691985217097681,
+      "clip_ratio/high_mean": 0.0010963495406031143,
+      "clip_ratio/low_mean": 0.0007340300808209577,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018303796096006408,
+      "epoch": 9.214577259475218,
+      "grad_norm": 0.12959793210029602,
+      "learning_rate": 1e-06,
+      "loss": -0.0782,
+      "step": 887
+    },
+    {
+      "clip_ratio/high_max": 0.0027285710093565285,
+      "clip_ratio/high_mean": 0.0010925341503025265,
+      "clip_ratio/low_mean": 0.0007487652110285126,
+      "clip_ratio/low_min": 4.35825659224065e-05,
+      "clip_ratio/region_mean": 0.0018412994031677954,
+      "epoch": 9.223906705539358,
+      "grad_norm": 0.12535342574119568,
+      "learning_rate": 1e-06,
+      "loss": -0.0286,
+      "step": 888
+    },
+    {
+      "clip_ratio/high_max": 0.0022187597714946605,
+      "clip_ratio/high_mean": 0.0009067620048881508,
+      "clip_ratio/low_mean": 0.000779069689087919,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001685831717622932,
+      "epoch": 9.2332361516035,
+      "grad_norm": 0.14810459315776825,
+      "learning_rate": 1e-06,
+      "loss": -0.0147,
+      "step": 889
+    },
+    {
+      "clip_ratio/high_max": 0.0024139151282724924,
+      "clip_ratio/high_mean": 0.0010855920045287348,
+      "clip_ratio/low_mean": 0.0008292405182146467,
+      "clip_ratio/low_min": 2.0226536435075104e-05,
+      "clip_ratio/region_mean": 0.001914832493639551,
+      "epoch": 9.242565597667639,
+      "grad_norm": 1.5517913103103638,
+      "learning_rate": 1e-06,
+      "loss": -0.0294,
+      "step": 890
+    },
+    {
+      "clip_ratio/high_max": 0.0022826083513791673,
+      "clip_ratio/high_mean": 0.0009988567380787572,
+      "clip_ratio/low_mean": 0.0007746484843664803,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017735052315401845,
+      "epoch": 9.251895043731778,
+      "grad_norm": 0.1243315041065216,
+      "learning_rate": 1e-06,
+      "loss": -0.0161,
+      "step": 891
+    },
+    {
+      "clip_ratio/high_max": 0.0024114770130836405,
+      "clip_ratio/high_mean": 0.0010205379367107525,
+      "clip_ratio/low_mean": 0.000787219390986138,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018077573186019436,
+      "epoch": 9.261224489795918,
+      "grad_norm": 0.21037563681602478,
+      "learning_rate": 1e-06,
+      "loss": -0.0489,
+      "step": 892
+    },
+    {
+      "clip_ratio/high_max": 0.0032194397499551997,
+      "clip_ratio/high_mean": 0.0011879017220053356,
+      "clip_ratio/low_mean": 0.0008789769035502104,
+      "clip_ratio/low_min": 1.4595982975151855e-05,
+      "clip_ratio/region_mean": 0.0020668786310125142,
+      "epoch": 9.270553935860057,
+      "grad_norm": 0.1283675879240036,
+      "learning_rate": 1e-06,
+      "loss": -0.0359,
+      "step": 893
+    },
+    {
+      "clip_ratio/high_max": 0.002518921763112303,
+      "clip_ratio/high_mean": 0.0010827663318195846,
+      "clip_ratio/low_mean": 0.0008738407523196656,
+      "clip_ratio/low_min": 7.028763320704456e-05,
+      "clip_ratio/region_mean": 0.0019566070841392502,
+      "epoch": 9.279883381924199,
+      "grad_norm": 0.14181050658226013,
+      "learning_rate": 1e-06,
+      "loss": -0.0145,
+      "step": 894
+    },
+    {
+      "clip_ratio/high_max": 0.002490412203769665,
+      "clip_ratio/high_mean": 0.0010082446679007262,
+      "clip_ratio/low_mean": 0.0009095173118112143,
+      "clip_ratio/low_min": 6.218976886884775e-05,
+      "clip_ratio/region_mean": 0.0019177619578840677,
+      "epoch": 9.289212827988338,
+      "grad_norm": 0.14314906299114227,
+      "learning_rate": 1e-06,
+      "loss": -0.0007,
+      "step": 895
+    },
+    {
+      "clip_ratio/high_max": 0.002657593671756331,
+      "clip_ratio/high_mean": 0.0010601921894703992,
+      "clip_ratio/low_mean": 0.0007704217950958991,
+      "clip_ratio/low_min": 1.8328446458326653e-05,
+      "clip_ratio/region_mean": 0.0018306139463675208,
+      "epoch": 9.298542274052478,
+      "grad_norm": 0.14605265855789185,
+      "learning_rate": 1e-06,
+      "loss": -0.0295,
+      "step": 896
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0738699776785714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4082.0,
+      "completions/mean_length": 802.8367919921875,
+      "completions/mean_terminated_length": 540.16748046875,
+      "completions/min_length": 4.0,
+      "completions/min_terminated_length": 4.0,
+      "epoch": 9.307871720116617,
+      "grad_norm": 0.13294248282909393,
+      "learning_rate": 1e-06,
+      "loss": -0.0295,
+      "num_tokens": 526703124.0,
+      "reward": 0.6496233344078064,
+      "reward_std": 0.15484841167926788,
+      "rewards/simpleverify_reward/mean": 0.6496233344078064,
+      "rewards/simpleverify_reward/std": 0.47709622979164124,
+      "step": 897
+    },
+    {
+      "clip_ratio/high_max": 0.0018489369431335945,
+      "clip_ratio/high_mean": 0.0007859781935621868,
+      "clip_ratio/low_mean": 0.000501667747812462,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012876459477411117,
+      "epoch": 9.317201166180759,
+      "grad_norm": 0.15338732302188873,
+      "learning_rate": 1e-06,
+      "loss": -0.0355,
+      "step": 898
+    },
+    {
+      "clip_ratio/high_max": 0.0020353350846562535,
+      "clip_ratio/high_mean": 0.0007973251358635025,
+      "clip_ratio/low_mean": 0.000407775437906821,
+      "clip_ratio/low_min": 1.2590652659127954e-05,
+      "clip_ratio/region_mean": 0.0012051006087858696,
+      "epoch": 9.326530612244898,
+      "grad_norm": 0.14864759147167206,
+      "learning_rate": 1e-06,
+      "loss": -0.0295,
+      "step": 899
+    },
+    {
+      "clip_ratio/high_max": 0.002195424778619781,
+      "clip_ratio/high_mean": 0.0006893596782902023,
+      "clip_ratio/low_mean": 0.00047398182232427644,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011633415342657827,
+      "epoch": 9.335860058309038,
+      "grad_norm": 0.1383877396583557,
+      "learning_rate": 1e-06,
+      "loss": -0.0123,
+      "step": 900
+    },
+    {
+      "clip_ratio/high_max": 0.0021147289808141068,
+      "clip_ratio/high_mean": 0.0008393442094529746,
+      "clip_ratio/low_mean": 0.0005201814681186079,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013595257078122813,
+      "epoch": 9.345189504373177,
+      "grad_norm": 0.15192995965480804,
+      "learning_rate": 1e-06,
+      "loss": -0.0428,
+      "step": 901
+    },
+    {
+      "clip_ratio/high_max": 0.002063903808448231,
+      "clip_ratio/high_mean": 0.0007696333414060064,
+      "clip_ratio/low_mean": 0.000701427194144344,
+      "clip_ratio/low_min": 2.8400838345987722e-05,
+      "clip_ratio/region_mean": 0.001471060520998435,
+      "epoch": 9.354518950437317,
+      "grad_norm": 4.191668510437012,
+      "learning_rate": 1e-06,
+      "loss": -0.0049,
+      "step": 902
+    },
+    {
+      "clip_ratio/high_max": 0.0025516079913359135,
+      "clip_ratio/high_mean": 0.0009315781180703198,
+      "clip_ratio/low_mean": 0.0005230867236605263,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014546648308169097,
+      "epoch": 9.363848396501458,
+      "grad_norm": 0.13582298159599304,
+      "learning_rate": 1e-06,
+      "loss": -0.0609,
+      "step": 903
+    },
+    {
+      "clip_ratio/high_max": 0.001924297415826004,
+      "clip_ratio/high_mean": 0.0007267159489856567,
+      "clip_ratio/low_mean": 0.000714097479431075,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014408134011318907,
+      "epoch": 9.373177842565598,
+      "grad_norm": 0.17259620130062103,
+      "learning_rate": 1e-06,
+      "loss": 0.0305,
+      "step": 904
+    },
+    {
+      "clip_ratio/high_max": 0.002297422422998352,
+      "clip_ratio/high_mean": 0.0010069530871987808,
+      "clip_ratio/low_mean": 0.0005643024314849754,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015712555396021344,
+      "epoch": 9.382507288629737,
+      "grad_norm": 0.13771282136440277,
+      "learning_rate": 1e-06,
+      "loss": -0.02,
+      "step": 905
+    },
+    {
+      "clip_ratio/high_max": 0.0020101481750316452,
+      "clip_ratio/high_mean": 0.0009243096374120796,
+      "clip_ratio/low_mean": 0.0007376313924396527,
+      "clip_ratio/low_min": 4.073871787113603e-05,
+      "clip_ratio/region_mean": 0.001661941030761227,
+      "epoch": 9.391836734693877,
+      "grad_norm": 0.15268857777118683,
+      "learning_rate": 1e-06,
+      "loss": -0.0167,
+      "step": 906
+    },
+    {
+      "clip_ratio/high_max": 0.0024281362602778245,
+      "clip_ratio/high_mean": 0.000886371630258509,
+      "clip_ratio/low_mean": 0.0006705998439429095,
+      "clip_ratio/low_min": 1.3196790860092733e-05,
+      "clip_ratio/region_mean": 0.0015569714887533337,
+      "epoch": 9.401166180758018,
+      "grad_norm": 0.14203545451164246,
+      "learning_rate": 1e-06,
+      "loss": -0.0279,
+      "step": 907
+    },
+    {
+      "clip_ratio/high_max": 0.002260651883261744,
+      "clip_ratio/high_mean": 0.0009041009507200215,
+      "clip_ratio/low_mean": 0.0008411090529989451,
+      "clip_ratio/low_min": 2.9925784474471584e-05,
+      "clip_ratio/region_mean": 0.001745210014632903,
+      "epoch": 9.410495626822158,
+      "grad_norm": 0.14670445024967194,
+      "learning_rate": 1e-06,
+      "loss": -0.0201,
+      "step": 908
+    },
+    {
+      "clip_ratio/high_max": 0.0028523376568045933,
+      "clip_ratio/high_mean": 0.0010644993308233097,
+      "clip_ratio/low_mean": 0.0006402868766599568,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017047862493200228,
+      "epoch": 9.419825072886297,
+      "grad_norm": 0.32806357741355896,
+      "learning_rate": 1e-06,
+      "loss": -0.0364,
+      "step": 909
+    },
+    {
+      "clip_ratio/high_max": 0.0020576808783516753,
+      "clip_ratio/high_mean": 0.0008409076972384355,
+      "clip_ratio/low_mean": 0.0005672940469594323,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014082017260079738,
+      "epoch": 9.429154518950437,
+      "grad_norm": 0.13929541409015656,
+      "learning_rate": 1e-06,
+      "loss": -0.0455,
+      "step": 910
+    },
+    {
+      "clip_ratio/high_max": 0.002748642309597926,
+      "clip_ratio/high_mean": 0.0010775229129649233,
+      "clip_ratio/low_mean": 0.0007187987357610837,
+      "clip_ratio/low_min": 2.8682881747954525e-05,
+      "clip_ratio/region_mean": 0.00179632160870824,
+      "epoch": 9.438483965014576,
+      "grad_norm": 0.14641906321048737,
+      "learning_rate": 1e-06,
+      "loss": -0.0461,
+      "step": 911
+    },
+    {
+      "clip_ratio/high_max": 0.00220919926505303,
+      "clip_ratio/high_mean": 0.0008464168568025343,
+      "clip_ratio/low_mean": 0.0008198327122954652,
+      "clip_ratio/low_min": 4.1326868085889146e-05,
+      "clip_ratio/region_mean": 0.001666249580011936,
+      "epoch": 9.447813411078718,
+      "grad_norm": 0.13365840911865234,
+      "learning_rate": 1e-06,
+      "loss": -0.0027,
+      "step": 912
+    },
+    {
+      "clip_ratio/high_max": 0.0023387129294860642,
+      "clip_ratio/high_mean": 0.0008236713329097256,
+      "clip_ratio/low_mean": 0.0007478946163246292,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015715659501438495,
+      "epoch": 9.457142857142857,
+      "grad_norm": 0.1388624608516693,
+      "learning_rate": 1e-06,
+      "loss": -0.0206,
+      "step": 913
+    },
+    {
+      "clip_ratio/high_max": 0.002195760105678346,
+      "clip_ratio/high_mean": 0.0009363750432385132,
+      "clip_ratio/low_mean": 0.000804728825642087,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017411038934369572,
+      "epoch": 9.466472303206997,
+      "grad_norm": 0.37228792905807495,
+      "learning_rate": 1e-06,
+      "loss": 0.0032,
+      "step": 914
+    },
+    {
+      "clip_ratio/high_max": 0.0021250151621643454,
+      "clip_ratio/high_mean": 0.0009652398130128859,
+      "clip_ratio/low_mean": 0.0007873637578086345,
+      "clip_ratio/low_min": 2.9377285500231665e-05,
+      "clip_ratio/region_mean": 0.0017526035226183012,
+      "epoch": 9.475801749271136,
+      "grad_norm": 0.1489114910364151,
+      "learning_rate": 1e-06,
+      "loss": -0.0593,
+      "step": 915
+    },
+    {
+      "clip_ratio/high_max": 0.002142500175978057,
+      "clip_ratio/high_mean": 0.000912011008040281,
+      "clip_ratio/low_mean": 0.0008255670236394508,
+      "clip_ratio/low_min": 9.25621934584342e-05,
+      "clip_ratio/region_mean": 0.0017375780080328695,
+      "epoch": 9.485131195335278,
+      "grad_norm": 0.12724773585796356,
+      "learning_rate": 1e-06,
+      "loss": -0.0462,
+      "step": 916
+    },
+    {
+      "clip_ratio/high_max": 0.002542140573495999,
+      "clip_ratio/high_mean": 0.0009185514627461089,
+      "clip_ratio/low_mean": 0.0008871549835021142,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018057064444292337,
+      "epoch": 9.494460641399417,
+      "grad_norm": 0.14874504506587982,
+      "learning_rate": 1e-06,
+      "loss": 0.0173,
+      "step": 917
+    },
+    {
+      "clip_ratio/high_max": 0.0026165570234297775,
+      "clip_ratio/high_mean": 0.0009514420962659642,
+      "clip_ratio/low_mean": 0.0009032052030306659,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018546472638263367,
+      "epoch": 9.503790087463557,
+      "grad_norm": 0.26071786880493164,
+      "learning_rate": 1e-06,
+      "loss": -0.0016,
+      "step": 918
+    },
+    {
+      "clip_ratio/high_max": 0.002320240644621663,
+      "clip_ratio/high_mean": 0.000963028876867611,
+      "clip_ratio/low_mean": 0.0008858062192302896,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018488350833649747,
+      "epoch": 9.513119533527696,
+      "grad_norm": 0.1486770063638687,
+      "learning_rate": 1e-06,
+      "loss": -0.0353,
+      "step": 919
+    },
+    {
+      "clip_ratio/high_max": 0.0023801244315109216,
+      "clip_ratio/high_mean": 0.0009178674863505876,
+      "clip_ratio/low_mean": 0.0009687192214187235,
+      "clip_ratio/low_min": 5.618819341179915e-05,
+      "clip_ratio/region_mean": 0.0018865866877604276,
+      "epoch": 9.522448979591836,
+      "grad_norm": 0.1972588300704956,
+      "learning_rate": 1e-06,
+      "loss": -0.0036,
+      "step": 920
+    },
+    {
+      "clip_ratio/high_max": 0.0024615703368908726,
+      "clip_ratio/high_mean": 0.0010931789802270941,
+      "clip_ratio/low_mean": 0.0009637643106543692,
+      "clip_ratio/low_min": 4.1408433389733545e-05,
+      "clip_ratio/region_mean": 0.002056943289062474,
+      "epoch": 9.531778425655977,
+      "grad_norm": 0.15567514300346375,
+      "learning_rate": 1e-06,
+      "loss": -0.0805,
+      "step": 921
+    },
+    {
+      "clip_ratio/high_max": 0.0024112346836773213,
+      "clip_ratio/high_mean": 0.0008924721605580999,
+      "clip_ratio/low_mean": 0.000902465867511637,
+      "clip_ratio/low_min": 4.443024590727873e-05,
+      "clip_ratio/region_mean": 0.0017949380198842846,
+      "epoch": 9.541107871720117,
+      "grad_norm": 0.1386972814798355,
+      "learning_rate": 1e-06,
+      "loss": -0.0034,
+      "step": 922
+    },
+    {
+      "clip_ratio/high_max": 0.0026421247457619756,
+      "clip_ratio/high_mean": 0.001152587039541686,
+      "clip_ratio/low_mean": 0.0009233891105395742,
+      "clip_ratio/low_min": 1.9869654352078214e-05,
+      "clip_ratio/region_mean": 0.0020759761391673237,
+      "epoch": 9.550437317784256,
+      "grad_norm": 0.1365169733762741,
+      "learning_rate": 1e-06,
+      "loss": -0.0408,
+      "step": 923
+    },
+    {
+      "clip_ratio/high_max": 0.0023661662999074906,
+      "clip_ratio/high_mean": 0.0009867211047094315,
+      "clip_ratio/low_mean": 0.0011236081500101136,
+      "clip_ratio/low_min": 3.420908615225926e-05,
+      "clip_ratio/region_mean": 0.002110329245624598,
+      "epoch": 9.559766763848396,
+      "grad_norm": 0.14760929346084595,
+      "learning_rate": 1e-06,
+      "loss": -0.0201,
+      "step": 924
+    },
+    {
+      "clip_ratio/high_max": 0.002754248365818057,
+      "clip_ratio/high_mean": 0.0012121745976401144,
+      "clip_ratio/low_mean": 0.0009409423691977281,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0021531169550144114,
+      "epoch": 9.569096209912537,
+      "grad_norm": 0.14264555275440216,
+      "learning_rate": 1e-06,
+      "loss": -0.0666,
+      "step": 925
+    },
+    {
+      "clip_ratio/high_max": 0.003279394790297374,
+      "clip_ratio/high_mean": 0.001171923639049055,
+      "clip_ratio/low_mean": 0.0010396044708613772,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0022115280662546866,
+      "epoch": 9.578425655976677,
+      "grad_norm": 0.1504513919353485,
+      "learning_rate": 1e-06,
+      "loss": 0.0008,
+      "step": 926
+    },
+    {
+      "clip_ratio/high_max": 0.002248778815555852,
+      "clip_ratio/high_mean": 0.0009007656153698917,
+      "clip_ratio/low_mean": 0.0009944847370206844,
+      "clip_ratio/low_min": 4.9996930101769976e-05,
+      "clip_ratio/region_mean": 0.001895250337838661,
+      "epoch": 9.587755102040816,
+      "grad_norm": 0.14505545794963837,
+      "learning_rate": 1e-06,
+      "loss": 0.0217,
+      "step": 927
+    },
+    {
+      "clip_ratio/high_max": 0.002946940003312193,
+      "clip_ratio/high_mean": 0.0011926191473321524,
+      "clip_ratio/low_mean": 0.0010505889367777854,
+      "clip_ratio/low_min": 7.204565736174118e-05,
+      "clip_ratio/region_mean": 0.0022432081241277047,
+      "epoch": 9.597084548104956,
+      "grad_norm": 0.13531829416751862,
+      "learning_rate": 1e-06,
+      "loss": -0.0584,
+      "step": 928
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0738699776785714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4075.0,
+      "completions/mean_length": 805.7715454101562,
+      "completions/mean_terminated_length": 543.3363037109375,
+      "completions/min_length": 5.0,
+      "completions/min_terminated_length": 5.0,
+      "epoch": 9.606413994169095,
+      "grad_norm": 0.14112375676631927,
+      "learning_rate": 1e-06,
+      "loss": -0.0645,
+      "num_tokens": 543967325.0,
+      "reward": 0.6425432562828064,
+      "reward_std": 0.15433722734451294,
+      "rewards/simpleverify_reward/mean": 0.6425432562828064,
+      "rewards/simpleverify_reward/std": 0.4792592525482178,
+      "step": 929
+    },
+    {
+      "clip_ratio/high_max": 0.0020987102761864662,
+      "clip_ratio/high_mean": 0.0007994007464731112,
+      "clip_ratio/low_mean": 0.00048655450063961325,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012859552480222192,
+      "epoch": 9.615743440233237,
+      "grad_norm": 0.15093941986560822,
+      "learning_rate": 1e-06,
+      "loss": -0.0253,
+      "step": 930
+    },
+    {
+      "clip_ratio/high_max": 0.0018653494771569967,
+      "clip_ratio/high_mean": 0.0007498073318856768,
+      "clip_ratio/low_mean": 0.00040532318053010385,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001155130523329717,
+      "epoch": 9.625072886297376,
+      "grad_norm": 0.14469389617443085,
+      "learning_rate": 1e-06,
+      "loss": -0.0605,
+      "step": 931
+    },
+    {
+      "clip_ratio/high_max": 0.0017922082042787224,
+      "clip_ratio/high_mean": 0.0006247796482057311,
+      "clip_ratio/low_mean": 0.00044246219249544083,
+      "clip_ratio/low_min": 1.3423539712675847e-05,
+      "clip_ratio/region_mean": 0.0010672418247850146,
+      "epoch": 9.634402332361516,
+      "grad_norm": 0.13388541340827942,
+      "learning_rate": 1e-06,
+      "loss": 0.0017,
+      "step": 932
+    },
+    {
+      "clip_ratio/high_max": 0.0019075371528742835,
+      "clip_ratio/high_mean": 0.0007945345150801586,
+      "clip_ratio/low_mean": 0.0005729380218326696,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00136747251599445,
+      "epoch": 9.643731778425655,
+      "grad_norm": 0.13685829937458038,
+      "learning_rate": 1e-06,
+      "loss": -0.0106,
+      "step": 933
+    },
+    {
+      "clip_ratio/high_max": 0.0017625001273700036,
+      "clip_ratio/high_mean": 0.0006951841442059958,
+      "clip_ratio/low_mean": 0.0005532228292395303,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012484070102800615,
+      "epoch": 9.653061224489797,
+      "grad_norm": 0.13377708196640015,
+      "learning_rate": 1e-06,
+      "loss": -0.0358,
+      "step": 934
+    },
+    {
+      "clip_ratio/high_max": 0.0017062682309187949,
+      "clip_ratio/high_mean": 0.0006445129001804162,
+      "clip_ratio/low_mean": 0.0006428167789636063,
+      "clip_ratio/low_min": 3.075030690524727e-05,
+      "clip_ratio/region_mean": 0.0012873296909674536,
+      "epoch": 9.662390670553936,
+      "grad_norm": 0.1544046550989151,
+      "learning_rate": 1e-06,
+      "loss": 0.0065,
+      "step": 935
+    },
+    {
+      "clip_ratio/high_max": 0.0021471931104315445,
+      "clip_ratio/high_mean": 0.0008416242453677114,
+      "clip_ratio/low_mean": 0.0005645988157993997,
+      "clip_ratio/low_min": 1.0792608918563928e-05,
+      "clip_ratio/region_mean": 0.0014062230366107542,
+      "epoch": 9.671720116618076,
+      "grad_norm": 0.1447170525789261,
+      "learning_rate": 1e-06,
+      "loss": 0.0034,
+      "step": 936
+    },
+    {
+      "clip_ratio/high_max": 0.002280427162986598,
+      "clip_ratio/high_mean": 0.0007694852911299677,
+      "clip_ratio/low_mean": 0.0006271884849411435,
+      "clip_ratio/low_min": 4.1026994949788786e-05,
+      "clip_ratio/region_mean": 0.0013966737569717225,
+      "epoch": 9.681049562682215,
+      "grad_norm": 0.14356733858585358,
+      "learning_rate": 1e-06,
+      "loss": -0.0507,
+      "step": 937
+    },
+    {
+      "clip_ratio/high_max": 0.0023910022500786,
+      "clip_ratio/high_mean": 0.0008914276631912799,
+      "clip_ratio/low_mean": 0.0006133094175311271,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001504737083450891,
+      "epoch": 9.690379008746355,
+      "grad_norm": 0.13298338651657104,
+      "learning_rate": 1e-06,
+      "loss": -0.0472,
+      "step": 938
+    },
+    {
+      "clip_ratio/high_max": 0.0021206866440479644,
+      "clip_ratio/high_mean": 0.000876473819516832,
+      "clip_ratio/low_mean": 0.0006826361877756426,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015591100382152945,
+      "epoch": 9.699708454810496,
+      "grad_norm": 0.14761479198932648,
+      "learning_rate": 1e-06,
+      "loss": -0.0333,
+      "step": 939
+    },
+    {
+      "clip_ratio/high_max": 0.0022532403490913566,
+      "clip_ratio/high_mean": 0.0008637648279545829,
+      "clip_ratio/low_mean": 0.0007729321687293123,
+      "clip_ratio/low_min": 3.0113225875538774e-05,
+      "clip_ratio/region_mean": 0.0016366969648515806,
+      "epoch": 9.709037900874636,
+      "grad_norm": 0.14637258648872375,
+      "learning_rate": 1e-06,
+      "loss": -0.0069,
+      "step": 940
+    },
+    {
+      "clip_ratio/high_max": 0.0019963298182119615,
+      "clip_ratio/high_mean": 0.0007406541462842142,
+      "clip_ratio/low_mean": 0.0006486438278443529,
+      "clip_ratio/low_min": 1.6120711734401993e-05,
+      "clip_ratio/region_mean": 0.0013892979332013056,
+      "epoch": 9.718367346938775,
+      "grad_norm": 0.1345418095588684,
+      "learning_rate": 1e-06,
+      "loss": -0.0168,
+      "step": 941
+    },
+    {
+      "clip_ratio/high_max": 0.0027508915663929656,
+      "clip_ratio/high_mean": 0.0010285658572684042,
+      "clip_ratio/low_mean": 0.0005490175335580716,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015775834253872745,
+      "epoch": 9.727696793002915,
+      "grad_norm": 0.12532925605773926,
+      "learning_rate": 1e-06,
+      "loss": -0.0575,
+      "step": 942
+    },
+    {
+      "clip_ratio/high_max": 0.002239793619082775,
+      "clip_ratio/high_mean": 0.0008860702728270553,
+      "clip_ratio/low_mean": 0.0006159189852041891,
+      "clip_ratio/low_min": 1.7322616258752532e-05,
+      "clip_ratio/region_mean": 0.001501989238022361,
+      "epoch": 9.737026239067056,
+      "grad_norm": 0.12550191581249237,
+      "learning_rate": 1e-06,
+      "loss": -0.0662,
+      "step": 943
+    },
+    {
+      "clip_ratio/high_max": 0.002336008237762144,
+      "clip_ratio/high_mean": 0.0008461864526907448,
+      "clip_ratio/low_mean": 0.0008429012323176721,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016890876649995334,
+      "epoch": 9.746355685131196,
+      "grad_norm": 0.14688219130039215,
+      "learning_rate": 1e-06,
+      "loss": 0.0095,
+      "step": 944
+    },
+    {
+      "clip_ratio/high_max": 0.0022440484863182064,
+      "clip_ratio/high_mean": 0.0008293414812214905,
+      "clip_ratio/low_mean": 0.0007935643543532933,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016229058164753951,
+      "epoch": 9.755685131195335,
+      "grad_norm": 0.1360652595758438,
+      "learning_rate": 1e-06,
+      "loss": -0.0094,
+      "step": 945
+    },
+    {
+      "clip_ratio/high_max": 0.0023962190389283933,
+      "clip_ratio/high_mean": 0.0009036373121489305,
+      "clip_ratio/low_mean": 0.0007466155984729994,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016502528815180995,
+      "epoch": 9.765014577259475,
+      "grad_norm": 0.14179359376430511,
+      "learning_rate": 1e-06,
+      "loss": -0.048,
+      "step": 946
+    },
+    {
+      "clip_ratio/high_max": 0.0026654254470486194,
+      "clip_ratio/high_mean": 0.0010938354153040564,
+      "clip_ratio/low_mean": 0.0008020839327400608,
+      "clip_ratio/low_min": 4.934125900035724e-05,
+      "clip_ratio/region_mean": 0.0018959193766932003,
+      "epoch": 9.774344023323614,
+      "grad_norm": 0.1481010466814041,
+      "learning_rate": 1e-06,
+      "loss": -0.0376,
+      "step": 947
+    },
+    {
+      "clip_ratio/high_max": 0.0025731913483468816,
+      "clip_ratio/high_mean": 0.0009571427381160902,
+      "clip_ratio/low_mean": 0.0008804390326986322,
+      "clip_ratio/low_min": 4.955558506480884e-05,
+      "clip_ratio/region_mean": 0.0018375817744527012,
+      "epoch": 9.783673469387756,
+      "grad_norm": 0.17433428764343262,
+      "learning_rate": 1e-06,
+      "loss": 0.0002,
+      "step": 948
+    },
+    {
+      "clip_ratio/high_max": 0.0022778754209866747,
+      "clip_ratio/high_mean": 0.0009427083605260123,
+      "clip_ratio/low_mean": 0.0006592823738174047,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016019907197915018,
+      "epoch": 9.793002915451895,
+      "grad_norm": 0.11593896895647049,
+      "learning_rate": 1e-06,
+      "loss": -0.079,
+      "step": 949
+    },
+    {
+      "clip_ratio/high_max": 0.002450108717312105,
+      "clip_ratio/high_mean": 0.000996564293927804,
+      "clip_ratio/low_mean": 0.0009397516132594319,
+      "clip_ratio/low_min": 3.1743858016852755e-05,
+      "clip_ratio/region_mean": 0.0019363159808563069,
+      "epoch": 9.802332361516035,
+      "grad_norm": 0.1490112692117691,
+      "learning_rate": 1e-06,
+      "loss": -0.0053,
+      "step": 950
+    },
+    {
+      "clip_ratio/high_max": 0.002661286336660851,
+      "clip_ratio/high_mean": 0.0010696067693061195,
+      "clip_ratio/low_mean": 0.0009575898402545135,
+      "clip_ratio/low_min": 2.5826446290011518e-05,
+      "clip_ratio/region_mean": 0.0020271966495784,
+      "epoch": 9.811661807580174,
+      "grad_norm": 0.14619016647338867,
+      "learning_rate": 1e-06,
+      "loss": -0.0569,
+      "step": 951
+    },
+    {
+      "clip_ratio/high_max": 0.002604166518722195,
+      "clip_ratio/high_mean": 0.001022584103338886,
+      "clip_ratio/low_mean": 0.000799032362010621,
+      "clip_ratio/low_min": 1.5574383724015206e-05,
+      "clip_ratio/region_mean": 0.0018216164826299064,
+      "epoch": 9.820991253644316,
+      "grad_norm": 0.14257757365703583,
+      "learning_rate": 1e-06,
+      "loss": -0.0578,
+      "step": 952
+    },
+    {
+      "clip_ratio/high_max": 0.0025133486633421853,
+      "clip_ratio/high_mean": 0.0009504987392574549,
+      "clip_ratio/low_mean": 0.0008947350024754996,
+      "clip_ratio/low_min": 4.656071541830897e-05,
+      "clip_ratio/region_mean": 0.0018452337681083009,
+      "epoch": 9.830320699708455,
+      "grad_norm": 0.13759483397006989,
+      "learning_rate": 1e-06,
+      "loss": -0.0167,
+      "step": 953
+    },
+    {
+      "clip_ratio/high_max": 0.0026442376256454736,
+      "clip_ratio/high_mean": 0.0010522497577767354,
+      "clip_ratio/low_mean": 0.0007457789815816795,
+      "clip_ratio/low_min": 4.582609744829824e-05,
+      "clip_ratio/region_mean": 0.0017980287229875103,
+      "epoch": 9.839650145772595,
+      "grad_norm": 0.149187371134758,
+      "learning_rate": 1e-06,
+      "loss": -0.055,
+      "step": 954
+    },
+    {
+      "clip_ratio/high_max": 0.0023868798598414287,
+      "clip_ratio/high_mean": 0.0009610945671738591,
+      "clip_ratio/low_mean": 0.000800088853793568,
+      "clip_ratio/low_min": 3.902185198967345e-05,
+      "clip_ratio/region_mean": 0.00176118341187248,
+      "epoch": 9.848979591836734,
+      "grad_norm": 0.13883022964000702,
+      "learning_rate": 1e-06,
+      "loss": -0.0438,
+      "step": 955
+    },
+    {
+      "clip_ratio/high_max": 0.0019565389593481086,
+      "clip_ratio/high_mean": 0.000855520960612921,
+      "clip_ratio/low_mean": 0.000916341752599692,
+      "clip_ratio/low_min": 1.6783029423095286e-05,
+      "clip_ratio/region_mean": 0.0017718627204885706,
+      "epoch": 9.858309037900874,
+      "grad_norm": 0.1580863744020462,
+      "learning_rate": 1e-06,
+      "loss": 0.0115,
+      "step": 956
+    },
+    {
+      "clip_ratio/high_max": 0.0021067335655970965,
+      "clip_ratio/high_mean": 0.000857573374560161,
+      "clip_ratio/low_mean": 0.0008684557051310549,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001726029055134859,
+      "epoch": 9.867638483965015,
+      "grad_norm": 0.14364555478096008,
+      "learning_rate": 1e-06,
+      "loss": -0.0197,
+      "step": 957
+    },
+    {
+      "clip_ratio/high_max": 0.002229661498859059,
+      "clip_ratio/high_mean": 0.0009492823483014945,
+      "clip_ratio/low_mean": 0.0008248334534073365,
+      "clip_ratio/low_min": 6.832440703874454e-05,
+      "clip_ratio/region_mean": 0.0017741157826094422,
+      "epoch": 9.876967930029155,
+      "grad_norm": 0.13506869971752167,
+      "learning_rate": 1e-06,
+      "loss": -0.0274,
+      "step": 958
+    },
+    {
+      "clip_ratio/high_max": 0.0022754226156393997,
+      "clip_ratio/high_mean": 0.0010122565799974836,
+      "clip_ratio/low_mean": 0.000683598967953003,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016958555206656456,
+      "epoch": 9.886297376093294,
+      "grad_norm": 0.14167319238185883,
+      "learning_rate": 1e-06,
+      "loss": -0.0589,
+      "step": 959
+    },
+    {
+      "clip_ratio/high_max": 0.002559137537900824,
+      "clip_ratio/high_mean": 0.001125937244069064,
+      "clip_ratio/low_mean": 0.0008741367091715802,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020000739677925594,
+      "epoch": 9.895626822157434,
+      "grad_norm": 0.13148806989192963,
+      "learning_rate": 1e-06,
+      "loss": -0.0495,
+      "step": 960
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0765904017857143,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4091.0,
+      "completions/mean_length": 815.6360473632812,
+      "completions/mean_terminated_length": 543.5526733398438,
+      "completions/min_length": 25.0,
+      "completions/min_terminated_length": 25.0,
+      "epoch": 10.00932944606414,
+      "grad_norm": 0.1500326693058014,
+      "learning_rate": 1e-06,
+      "loss": -0.0043,
+      "num_tokens": 561264962.0,
+      "reward": 0.6439732313156128,
+      "reward_std": 0.15593308210372925,
+      "rewards/simpleverify_reward/mean": 0.6439732313156128,
+      "rewards/simpleverify_reward/std": 0.47883158922195435,
+      "step": 961
+    },
+    {
+      "clip_ratio/high_max": 0.001909110196720576,
+      "clip_ratio/high_mean": 0.0006390837424987694,
+      "clip_ratio/low_mean": 0.00045998847508599283,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010990722221322358,
+      "epoch": 10.018658892128279,
+      "grad_norm": 0.12861394882202148,
+      "learning_rate": 1e-06,
+      "loss": -0.0153,
+      "step": 962
+    },
+    {
+      "clip_ratio/high_max": 0.001726972492178902,
+      "clip_ratio/high_mean": 0.0006959185730011086,
+      "clip_ratio/low_mean": 0.0004754106967084226,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00117132927334751,
+      "epoch": 10.02798833819242,
+      "grad_norm": 0.14925359189510345,
+      "learning_rate": 1e-06,
+      "loss": -0.0361,
+      "step": 963
+    },
+    {
+      "clip_ratio/high_max": 0.0020835115428781137,
+      "clip_ratio/high_mean": 0.0007047521248750854,
+      "clip_ratio/low_mean": 0.00050725807705021,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012120101964683272,
+      "epoch": 10.03731778425656,
+      "grad_norm": 0.1678958386182785,
+      "learning_rate": 1e-06,
+      "loss": -0.0414,
+      "step": 964
+    },
+    {
+      "clip_ratio/high_max": 0.0018111032077285927,
+      "clip_ratio/high_mean": 0.0007327352122956654,
+      "clip_ratio/low_mean": 0.0005728224514314206,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013055576564511284,
+      "epoch": 10.0466472303207,
+      "grad_norm": 0.17154543101787567,
+      "learning_rate": 1e-06,
+      "loss": -0.0272,
+      "step": 965
+    },
+    {
+      "clip_ratio/high_max": 0.0022328296727209818,
+      "clip_ratio/high_mean": 0.0007948934726300649,
+      "clip_ratio/low_mean": 0.0005207437361605116,
+      "clip_ratio/low_min": 1.6356974811060354e-05,
+      "clip_ratio/region_mean": 0.001315637222432997,
+      "epoch": 10.055976676384839,
+      "grad_norm": 0.14976218342781067,
+      "learning_rate": 1e-06,
+      "loss": -0.0237,
+      "step": 966
+    },
+    {
+      "clip_ratio/high_max": 0.0021570081589743495,
+      "clip_ratio/high_mean": 0.0008220616673497716,
+      "clip_ratio/low_mean": 0.0005323067780409474,
+      "clip_ratio/low_min": 1.504573901911499e-05,
+      "clip_ratio/region_mean": 0.0013543684217438567,
+      "epoch": 10.06530612244898,
+      "grad_norm": 0.1393507421016693,
+      "learning_rate": 1e-06,
+      "loss": -0.0162,
+      "step": 967
+    },
+    {
+      "clip_ratio/high_max": 0.0016665338625898585,
+      "clip_ratio/high_mean": 0.0006571479025296867,
+      "clip_ratio/low_mean": 0.0006993826282268856,
+      "clip_ratio/low_min": 3.510578608256765e-05,
+      "clip_ratio/region_mean": 0.0013565305198426358,
+      "epoch": 10.07463556851312,
+      "grad_norm": 0.14176590740680695,
+      "learning_rate": 1e-06,
+      "loss": -0.0061,
+      "step": 968
+    },
+    {
+      "clip_ratio/high_max": 0.0020020450137963053,
+      "clip_ratio/high_mean": 0.0007357875847446849,
+      "clip_ratio/low_mean": 0.0006004850783938309,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013362726676859893,
+      "epoch": 10.08396501457726,
+      "grad_norm": 0.13857552409172058,
+      "learning_rate": 1e-06,
+      "loss": 0.0088,
+      "step": 969
+    },
+    {
+      "clip_ratio/high_max": 0.0020954376086592674,
+      "clip_ratio/high_mean": 0.0008055659618548816,
+      "clip_ratio/low_mean": 0.0006883723835926503,
+      "clip_ratio/low_min": 4.625426026905188e-05,
+      "clip_ratio/region_mean": 0.0014939383763703518,
+      "epoch": 10.093294460641399,
+      "grad_norm": 0.13805639743804932,
+      "learning_rate": 1e-06,
+      "loss": -0.0118,
+      "step": 970
+    },
+    {
+      "clip_ratio/high_max": 0.0020044581760885194,
+      "clip_ratio/high_mean": 0.0008033949143282371,
+      "clip_ratio/low_mean": 0.0006440322326852765,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014474271447397768,
+      "epoch": 10.102623906705539,
+      "grad_norm": 0.15415208041667938,
+      "learning_rate": 1e-06,
+      "loss": -0.0237,
+      "step": 971
+    },
+    {
+      "clip_ratio/high_max": 0.0021812322447658516,
+      "clip_ratio/high_mean": 0.0008289018569485052,
+      "clip_ratio/low_mean": 0.00048655639693606645,
+      "clip_ratio/low_min": 1.7189218851854093e-05,
+      "clip_ratio/region_mean": 0.0013154582702554762,
+      "epoch": 10.11195335276968,
+      "grad_norm": 0.14797736704349518,
+      "learning_rate": 1e-06,
+      "loss": -0.0567,
+      "step": 972
+    },
+    {
+      "clip_ratio/high_max": 0.0020825552601309028,
+      "clip_ratio/high_mean": 0.0008909320931707043,
+      "clip_ratio/low_mean": 0.0007273344635905232,
+      "clip_ratio/low_min": 1.7610595023143105e-05,
+      "clip_ratio/region_mean": 0.0016182665895030368,
+      "epoch": 10.12128279883382,
+      "grad_norm": 0.15112736821174622,
+      "learning_rate": 1e-06,
+      "loss": -0.047,
+      "step": 973
+    },
+    {
+      "clip_ratio/high_max": 0.00220344615081558,
+      "clip_ratio/high_mean": 0.0009626862138247816,
+      "clip_ratio/low_mean": 0.0007337629031098913,
+      "clip_ratio/low_min": 1.0976466910506133e-05,
+      "clip_ratio/region_mean": 0.001696449086011853,
+      "epoch": 10.130612244897959,
+      "grad_norm": 0.1311543732881546,
+      "learning_rate": 1e-06,
+      "loss": -0.0352,
+      "step": 974
+    },
+    {
+      "clip_ratio/high_max": 0.002281056142237503,
+      "clip_ratio/high_mean": 0.0008180264849215746,
+      "clip_ratio/low_mean": 0.0008300371300720144,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001648063614993589,
+      "epoch": 10.139941690962099,
+      "grad_norm": 0.15632712841033936,
+      "learning_rate": 1e-06,
+      "loss": 0.0079,
+      "step": 975
+    },
+    {
+      "clip_ratio/high_max": 0.0021112567228556145,
+      "clip_ratio/high_mean": 0.0008205298727261834,
+      "clip_ratio/low_mean": 0.0007860472078391467,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016065770796558354,
+      "epoch": 10.14927113702624,
+      "grad_norm": 0.1610146015882492,
+      "learning_rate": 1e-06,
+      "loss": 0.0208,
+      "step": 976
+    },
+    {
+      "clip_ratio/high_max": 0.002420800185063854,
+      "clip_ratio/high_mean": 0.0010541762094362639,
+      "clip_ratio/low_mean": 0.000777995390308206,
+      "clip_ratio/low_min": 3.944130912714172e-05,
+      "clip_ratio/region_mean": 0.0018321715797355864,
+      "epoch": 10.15860058309038,
+      "grad_norm": 0.15158908069133759,
+      "learning_rate": 1e-06,
+      "loss": -0.0482,
+      "step": 977
+    },
+    {
+      "clip_ratio/high_max": 0.0025289334371336736,
+      "clip_ratio/high_mean": 0.001032315978591214,
+      "clip_ratio/low_mean": 0.0007892725134297507,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018215884956589434,
+      "epoch": 10.167930029154519,
+      "grad_norm": 0.1517321765422821,
+      "learning_rate": 1e-06,
+      "loss": -0.0449,
+      "step": 978
+    },
+    {
+      "clip_ratio/high_max": 0.00251186663081171,
+      "clip_ratio/high_mean": 0.0009472684323554859,
+      "clip_ratio/low_mean": 0.0010198395466431975,
+      "clip_ratio/low_min": 6.164200749481097e-05,
+      "clip_ratio/region_mean": 0.001967107964446768,
+      "epoch": 10.177259475218658,
+      "grad_norm": 0.17969931662082672,
+      "learning_rate": 1e-06,
+      "loss": -0.023,
+      "step": 979
+    },
+    {
+      "clip_ratio/high_max": 0.0024426607706118375,
+      "clip_ratio/high_mean": 0.0010387922484369483,
+      "clip_ratio/low_mean": 0.0008045482863963116,
+      "clip_ratio/low_min": 1.3954007954453118e-05,
+      "clip_ratio/region_mean": 0.0018433404984534718,
+      "epoch": 10.186588921282798,
+      "grad_norm": 0.13014066219329834,
+      "learning_rate": 1e-06,
+      "loss": -0.036,
+      "step": 980
+    },
+    {
+      "clip_ratio/high_max": 0.0023669466099818237,
+      "clip_ratio/high_mean": 0.0009498952094872948,
+      "clip_ratio/low_mean": 0.0007533242551289732,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017032194300554693,
+      "epoch": 10.19591836734694,
+      "grad_norm": 0.140021413564682,
+      "learning_rate": 1e-06,
+      "loss": -0.0382,
+      "step": 981
+    },
+    {
+      "clip_ratio/high_max": 0.0024754825062700547,
+      "clip_ratio/high_mean": 0.0010548727586865425,
+      "clip_ratio/low_mean": 0.000770688675402198,
+      "clip_ratio/low_min": 4.954419819114264e-05,
+      "clip_ratio/region_mean": 0.0018255614413646981,
+      "epoch": 10.205247813411079,
+      "grad_norm": 0.16048754751682281,
+      "learning_rate": 1e-06,
+      "loss": -0.0356,
+      "step": 982
+    },
+    {
+      "clip_ratio/high_max": 0.0026544872162048705,
+      "clip_ratio/high_mean": 0.0011467996628198307,
+      "clip_ratio/low_mean": 0.0007798313927196432,
+      "clip_ratio/low_min": 4.544352850643918e-05,
+      "clip_ratio/region_mean": 0.0019266310628154315,
+      "epoch": 10.214577259475218,
+      "grad_norm": 0.14217230677604675,
+      "learning_rate": 1e-06,
+      "loss": -0.103,
+      "step": 983
+    },
+    {
+      "clip_ratio/high_max": 0.002223150913778227,
+      "clip_ratio/high_mean": 0.0009643569683248643,
+      "clip_ratio/low_mean": 0.0008032557434489718,
+      "clip_ratio/low_min": 6.634566125285346e-05,
+      "clip_ratio/region_mean": 0.0017676126954029314,
+      "epoch": 10.223906705539358,
+      "grad_norm": 0.14624422788619995,
+      "learning_rate": 1e-06,
+      "loss": -0.0517,
+      "step": 984
+    },
+    {
+      "clip_ratio/high_max": 0.0025765645259525627,
+      "clip_ratio/high_mean": 0.0010526941732678097,
+      "clip_ratio/low_mean": 0.0008813243184704334,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001934018473548349,
+      "epoch": 10.2332361516035,
+      "grad_norm": 0.14733164012432098,
+      "learning_rate": 1e-06,
+      "loss": -0.0382,
+      "step": 985
+    },
+    {
+      "clip_ratio/high_max": 0.0025515191155136563,
+      "clip_ratio/high_mean": 0.0010626111306919483,
+      "clip_ratio/low_mean": 0.0009340102224086877,
+      "clip_ratio/low_min": 4.5101674913894385e-05,
+      "clip_ratio/region_mean": 0.0019966213731095195,
+      "epoch": 10.242565597667639,
+      "grad_norm": 0.146484836935997,
+      "learning_rate": 1e-06,
+      "loss": -0.0013,
+      "step": 986
+    },
+    {
+      "clip_ratio/high_max": 0.002395473995420616,
+      "clip_ratio/high_mean": 0.0009680404637038009,
+      "clip_ratio/low_mean": 0.0007183443294707104,
+      "clip_ratio/low_min": 1.2942638022650499e-05,
+      "clip_ratio/region_mean": 0.0016863847995409742,
+      "epoch": 10.251895043731778,
+      "grad_norm": 0.14619509875774384,
+      "learning_rate": 1e-06,
+      "loss": -0.067,
+      "step": 987
+    },
+    {
+      "clip_ratio/high_max": 0.0019169974184478633,
+      "clip_ratio/high_mean": 0.0008501538814016385,
+      "clip_ratio/low_mean": 0.0009479271257077926,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001798080986191053,
+      "epoch": 10.261224489795918,
+      "grad_norm": 0.13349026441574097,
+      "learning_rate": 1e-06,
+      "loss": -0.0003,
+      "step": 988
+    },
+    {
+      "clip_ratio/high_max": 0.002456573558447417,
+      "clip_ratio/high_mean": 0.0011083521458203904,
+      "clip_ratio/low_mean": 0.0006360037540389385,
+      "clip_ratio/low_min": 1.4082920642977115e-05,
+      "clip_ratio/region_mean": 0.0017443558826926164,
+      "epoch": 10.270553935860057,
+      "grad_norm": 0.1264098435640335,
+      "learning_rate": 1e-06,
+      "loss": -0.1013,
+      "step": 989
+    },
+    {
+      "clip_ratio/high_max": 0.002665226122189779,
+      "clip_ratio/high_mean": 0.001071601793228183,
+      "clip_ratio/low_mean": 0.001032287153066136,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0021038888953626156,
+      "epoch": 10.279883381924199,
+      "grad_norm": 0.15136873722076416,
+      "learning_rate": 1e-06,
+      "loss": 0.0035,
+      "step": 990
+    },
+    {
+      "clip_ratio/high_max": 0.002894303062930703,
+      "clip_ratio/high_mean": 0.001205394135467941,
+      "clip_ratio/low_mean": 0.0009282167047786061,
+      "clip_ratio/low_min": 3.288608422735706e-05,
+      "clip_ratio/region_mean": 0.002133610840246547,
+      "epoch": 10.289212827988338,
+      "grad_norm": 0.1624498814344406,
+      "learning_rate": 1e-06,
+      "loss": -0.0426,
+      "step": 991
+    },
+    {
+      "clip_ratio/high_max": 0.002337928242923226,
+      "clip_ratio/high_mean": 0.0010890516423387453,
+      "clip_ratio/low_mean": 0.0009289082154282369,
+      "clip_ratio/low_min": 5.011562643630896e-05,
+      "clip_ratio/region_mean": 0.0020179598650429398,
+      "epoch": 10.298542274052478,
+      "grad_norm": 0.13214270770549774,
+      "learning_rate": 1e-06,
+      "loss": -0.0424,
+      "step": 992
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.074951171875,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4039.0,
+      "completions/mean_length": 799.0311889648438,
+      "completions/mean_terminated_length": 531.8975219726562,
+      "completions/min_length": 102.0,
+      "completions/min_terminated_length": 102.0,
+      "epoch": 10.307871720116617,
+      "grad_norm": 0.13746856153011322,
+      "learning_rate": 1e-06,
+      "loss": -0.029,
+      "num_tokens": 578202624.0,
+      "reward": 0.6584821939468384,
+      "reward_std": 0.14528833329677582,
+      "rewards/simpleverify_reward/mean": 0.6584821343421936,
+      "rewards/simpleverify_reward/std": 0.47422701120376587,
+      "step": 993
+    },
+    {
+      "clip_ratio/high_max": 0.0015481885375265847,
+      "clip_ratio/high_mean": 0.0006157521645491215,
+      "clip_ratio/low_mean": 0.00044636771599471103,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010621198707667645,
+      "epoch": 10.317201166180759,
+      "grad_norm": 0.14652684330940247,
+      "learning_rate": 1e-06,
+      "loss": -0.0046,
+      "step": 994
+    },
+    {
+      "clip_ratio/high_max": 0.0014798309166508261,
+      "clip_ratio/high_mean": 0.0005467086066346383,
+      "clip_ratio/low_mean": 0.0004052429840157856,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0009519515879219398,
+      "epoch": 10.326530612244898,
+      "grad_norm": 0.13823182880878448,
+      "learning_rate": 1e-06,
+      "loss": -0.03,
+      "step": 995
+    },
+    {
+      "clip_ratio/high_max": 0.001953846891410649,
+      "clip_ratio/high_mean": 0.0008146686814143322,
+      "clip_ratio/low_mean": 0.0004966830947523704,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001311351777985692,
+      "epoch": 10.335860058309038,
+      "grad_norm": 0.16025929152965546,
+      "learning_rate": 1e-06,
+      "loss": -0.0394,
+      "step": 996
+    },
+    {
+      "clip_ratio/high_max": 0.0020486487228481565,
+      "clip_ratio/high_mean": 0.0006944371525605675,
+      "clip_ratio/low_mean": 0.0004157131415922777,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001110150309614255,
+      "epoch": 10.345189504373177,
+      "grad_norm": 0.17115995287895203,
+      "learning_rate": 1e-06,
+      "loss": -0.0344,
+      "step": 997
+    },
+    {
+      "clip_ratio/high_max": 0.0016736180223233532,
+      "clip_ratio/high_mean": 0.0005873534410056891,
+      "clip_ratio/low_mean": 0.000505985150084598,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010933385747193824,
+      "epoch": 10.354518950437317,
+      "grad_norm": 0.15793941915035248,
+      "learning_rate": 1e-06,
+      "loss": 0.0112,
+      "step": 998
+    },
+    {
+      "clip_ratio/high_max": 0.0020567545798257925,
+      "clip_ratio/high_mean": 0.0008095356388366781,
+      "clip_ratio/low_mean": 0.00044864863411930855,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012581843002408277,
+      "epoch": 10.363848396501458,
+      "grad_norm": 0.13997121155261993,
+      "learning_rate": 1e-06,
+      "loss": -0.0425,
+      "step": 999
+    },
+    {
+      "clip_ratio/high_max": 0.001971263758605346,
+      "clip_ratio/high_mean": 0.0007790296240273165,
+      "clip_ratio/low_mean": 0.0005122356722040422,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001291265331019531,
+      "epoch": 10.373177842565598,
+      "grad_norm": 0.13614171743392944,
+      "learning_rate": 1e-06,
+      "loss": -0.0523,
+      "step": 1000
+    },
+    {
+      "clip_ratio/high_max": 0.0017270641328650527,
+      "clip_ratio/high_mean": 0.0006994562040745222,
+      "clip_ratio/low_mean": 0.0005206879040997592,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012201440931676188,
+      "epoch": 10.382507288629737,
+      "grad_norm": 0.1247684508562088,
+      "learning_rate": 1e-06,
+      "loss": -0.0098,
+      "step": 1001
+    },
+    {
+      "clip_ratio/high_max": 0.0020272995679988526,
+      "clip_ratio/high_mean": 0.0008106346303975442,
+      "clip_ratio/low_mean": 0.0007284987614184502,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015391333799925633,
+      "epoch": 10.391836734693877,
+      "grad_norm": 0.1570490151643753,
+      "learning_rate": 1e-06,
+      "loss": -0.0212,
+      "step": 1002
+    },
+    {
+      "clip_ratio/high_max": 0.002310659623617539,
+      "clip_ratio/high_mean": 0.000947707203522441,
+      "clip_ratio/low_mean": 0.0005226368625699251,
+      "clip_ratio/low_min": 1.7085838408092968e-05,
+      "clip_ratio/region_mean": 0.001470344082918018,
+      "epoch": 10.401166180758018,
+      "grad_norm": 0.13579373061656952,
+      "learning_rate": 1e-06,
+      "loss": -0.0647,
+      "step": 1003
+    },
+    {
+      "clip_ratio/high_max": 0.002140488439181354,
+      "clip_ratio/high_mean": 0.0007807352449162863,
+      "clip_ratio/low_mean": 0.0006728586513418122,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014535938717017416,
+      "epoch": 10.410495626822158,
+      "grad_norm": 0.16249175369739532,
+      "learning_rate": 1e-06,
+      "loss": -0.0038,
+      "step": 1004
+    },
+    {
+      "clip_ratio/high_max": 0.0019618572914623655,
+      "clip_ratio/high_mean": 0.0007476367372873938,
+      "clip_ratio/low_mean": 0.0006206943739925919,
+      "clip_ratio/low_min": 2.3062730178935453e-05,
+      "clip_ratio/region_mean": 0.001368331093544839,
+      "epoch": 10.419825072886297,
+      "grad_norm": 0.13393568992614746,
+      "learning_rate": 1e-06,
+      "loss": -0.0501,
+      "step": 1005
+    },
+    {
+      "clip_ratio/high_max": 0.002194969914853573,
+      "clip_ratio/high_mean": 0.0008562507773604011,
+      "clip_ratio/low_mean": 0.0007611364399053855,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016173872318177018,
+      "epoch": 10.429154518950437,
+      "grad_norm": 0.1739315390586853,
+      "learning_rate": 1e-06,
+      "loss": -0.0046,
+      "step": 1006
+    },
+    {
+      "clip_ratio/high_max": 0.002169353683711961,
+      "clip_ratio/high_mean": 0.0008425281757808989,
+      "clip_ratio/low_mean": 0.0008274836281998432,
+      "clip_ratio/low_min": 1.9531249563442543e-05,
+      "clip_ratio/region_mean": 0.001670011792157311,
+      "epoch": 10.438483965014576,
+      "grad_norm": 0.13166306912899017,
+      "learning_rate": 1e-06,
+      "loss": 0.0061,
+      "step": 1007
+    },
+    {
+      "clip_ratio/high_max": 0.0024623638528282754,
+      "clip_ratio/high_mean": 0.0008831531267787796,
+      "clip_ratio/low_mean": 0.0006474850888480432,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001530638248368632,
+      "epoch": 10.447813411078718,
+      "grad_norm": 0.14744344353675842,
+      "learning_rate": 1e-06,
+      "loss": -0.045,
+      "step": 1008
+    },
+    {
+      "clip_ratio/high_max": 0.0020738662824442144,
+      "clip_ratio/high_mean": 0.0008392149939027149,
+      "clip_ratio/low_mean": 0.0006633309924382047,
+      "clip_ratio/low_min": 3.963221388403326e-05,
+      "clip_ratio/region_mean": 0.001502545925177401,
+      "epoch": 10.457142857142857,
+      "grad_norm": 0.13794247806072235,
+      "learning_rate": 1e-06,
+      "loss": -0.0291,
+      "step": 1009
+    },
+    {
+      "clip_ratio/high_max": 0.0019332764422870241,
+      "clip_ratio/high_mean": 0.000819532713649096,
+      "clip_ratio/low_mean": 0.0006383089366863715,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014578416448784992,
+      "epoch": 10.466472303206997,
+      "grad_norm": 0.1313249170780182,
+      "learning_rate": 1e-06,
+      "loss": -0.028,
+      "step": 1010
+    },
+    {
+      "clip_ratio/high_max": 0.0024970816666609608,
+      "clip_ratio/high_mean": 0.0010146628519578371,
+      "clip_ratio/low_mean": 0.0006212481612237752,
+      "clip_ratio/low_min": 1.6425756257376634e-05,
+      "clip_ratio/region_mean": 0.001635911063203821,
+      "epoch": 10.475801749271136,
+      "grad_norm": 0.15788981318473816,
+      "learning_rate": 1e-06,
+      "loss": -0.0315,
+      "step": 1011
+    },
+    {
+      "clip_ratio/high_max": 0.001872238477517385,
+      "clip_ratio/high_mean": 0.000791782209489611,
+      "clip_ratio/low_mean": 0.0007423156257573282,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015340978534368332,
+      "epoch": 10.485131195335278,
+      "grad_norm": 0.12612873315811157,
+      "learning_rate": 1e-06,
+      "loss": -0.0201,
+      "step": 1012
+    },
+    {
+      "clip_ratio/high_max": 0.002122684250934981,
+      "clip_ratio/high_mean": 0.0009315368479292374,
+      "clip_ratio/low_mean": 0.0006933887671038974,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00162492562958505,
+      "epoch": 10.494460641399417,
+      "grad_norm": 0.13648264110088348,
+      "learning_rate": 1e-06,
+      "loss": -0.0476,
+      "step": 1013
+    },
+    {
+      "clip_ratio/high_max": 0.0026277231590938754,
+      "clip_ratio/high_mean": 0.0010553726569924038,
+      "clip_ratio/low_mean": 0.0007376682660833467,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017930409012478776,
+      "epoch": 10.503790087463557,
+      "grad_norm": 0.15639662742614746,
+      "learning_rate": 1e-06,
+      "loss": -0.0353,
+      "step": 1014
+    },
+    {
+      "clip_ratio/high_max": 0.0022894211288075894,
+      "clip_ratio/high_mean": 0.0008977182860689936,
+      "clip_ratio/low_mean": 0.0006727097893417522,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015704280922363978,
+      "epoch": 10.513119533527696,
+      "grad_norm": 0.1324109584093094,
+      "learning_rate": 1e-06,
+      "loss": -0.047,
+      "step": 1015
+    },
+    {
+      "clip_ratio/high_max": 0.0024130656456691213,
+      "clip_ratio/high_mean": 0.0010579552144918125,
+      "clip_ratio/low_mean": 0.0007388108697341522,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017967660933209118,
+      "epoch": 10.522448979591836,
+      "grad_norm": 0.1491599828004837,
+      "learning_rate": 1e-06,
+      "loss": -0.0614,
+      "step": 1016
+    },
+    {
+      "clip_ratio/high_max": 0.002259333457914181,
+      "clip_ratio/high_mean": 0.0008822530953693786,
+      "clip_ratio/low_mean": 0.0005619981329800794,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014442512001551222,
+      "epoch": 10.531778425655977,
+      "grad_norm": 0.12334480881690979,
+      "learning_rate": 1e-06,
+      "loss": -0.0539,
+      "step": 1017
+    },
+    {
+      "clip_ratio/high_max": 0.002269917298690416,
+      "clip_ratio/high_mean": 0.0009123131312662736,
+      "clip_ratio/low_mean": 0.0007771314358251402,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001689444536168594,
+      "epoch": 10.541107871720117,
+      "grad_norm": 0.13667048513889313,
+      "learning_rate": 1e-06,
+      "loss": -0.0554,
+      "step": 1018
+    },
+    {
+      "clip_ratio/high_max": 0.0021292364071996417,
+      "clip_ratio/high_mean": 0.0008424350307905115,
+      "clip_ratio/low_mean": 0.0006537908955124294,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014962259156163782,
+      "epoch": 10.550437317784256,
+      "grad_norm": 0.12534432113170624,
+      "learning_rate": 1e-06,
+      "loss": -0.0401,
+      "step": 1019
+    },
+    {
+      "clip_ratio/high_max": 0.002169070368836401,
+      "clip_ratio/high_mean": 0.0008813484819256701,
+      "clip_ratio/low_mean": 0.0007044645008136285,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015858130136621185,
+      "epoch": 10.559766763848396,
+      "grad_norm": 0.15574602782726288,
+      "learning_rate": 1e-06,
+      "loss": -0.0144,
+      "step": 1020
+    },
+    {
+      "clip_ratio/high_max": 0.002250806392112281,
+      "clip_ratio/high_mean": 0.0009773394449439365,
+      "clip_ratio/low_mean": 0.0006898111887494451,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016671506164129823,
+      "epoch": 10.569096209912537,
+      "grad_norm": 0.15688715875148773,
+      "learning_rate": 1e-06,
+      "loss": -0.0297,
+      "step": 1021
+    },
+    {
+      "clip_ratio/high_max": 0.002465681689500343,
+      "clip_ratio/high_mean": 0.0009920224474626593,
+      "clip_ratio/low_mean": 0.0007774367732054088,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017694592170300893,
+      "epoch": 10.578425655976677,
+      "grad_norm": 0.13506655395030975,
+      "learning_rate": 1e-06,
+      "loss": -0.0508,
+      "step": 1022
+    },
+    {
+      "clip_ratio/high_max": 0.0027156087235198356,
+      "clip_ratio/high_mean": 0.001017479526126408,
+      "clip_ratio/low_mean": 0.0008608054013166111,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001878284921986051,
+      "epoch": 10.587755102040816,
+      "grad_norm": 0.14973346889019012,
+      "learning_rate": 1e-06,
+      "loss": -0.0209,
+      "step": 1023
+    },
+    {
+      "clip_ratio/high_max": 0.002389958310232032,
+      "clip_ratio/high_mean": 0.0010566878281679237,
+      "clip_ratio/low_mean": 0.0008665779268994811,
+      "clip_ratio/low_min": 4.5609340304508805e-05,
+      "clip_ratio/region_mean": 0.0019232657468819525,
+      "epoch": 10.597084548104956,
+      "grad_norm": 11.152640342712402,
+      "learning_rate": 1e-06,
+      "loss": -0.0647,
+      "step": 1024
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0752650669642857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4062.0,
+      "completions/mean_length": 803.6248168945312,
+      "completions/mean_terminated_length": 535.6552124023438,
+      "completions/min_length": 98.0,
+      "completions/min_terminated_length": 98.0,
+      "epoch": 10.606413994169095,
+      "grad_norm": 0.1410028487443924,
+      "learning_rate": 1e-06,
+      "loss": 0.0005,
+      "num_tokens": 595257418.0,
+      "reward": 0.6659110188484192,
+      "reward_std": 0.15039822459220886,
+      "rewards/simpleverify_reward/mean": 0.6659110188484192,
+      "rewards/simpleverify_reward/std": 0.4716792404651642,
+      "step": 1025
+    },
+    {
+      "clip_ratio/high_max": 0.002021494052314665,
+      "clip_ratio/high_mean": 0.0007912164637673413,
+      "clip_ratio/low_mean": 0.00037528156758526166,
+      "clip_ratio/low_min": 1.2195121598779224e-05,
+      "clip_ratio/region_mean": 0.0011664980265777558,
+      "epoch": 10.615743440233237,
+      "grad_norm": 0.14194558560848236,
+      "learning_rate": 1e-06,
+      "loss": -0.0791,
+      "step": 1026
+    },
+    {
+      "clip_ratio/high_max": 0.0018872605214710347,
+      "clip_ratio/high_mean": 0.0007259628055180656,
+      "clip_ratio/low_mean": 0.00042551893329800805,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011514817488205153,
+      "epoch": 10.625072886297376,
+      "grad_norm": 0.14285390079021454,
+      "learning_rate": 1e-06,
+      "loss": -0.0176,
+      "step": 1027
+    },
+    {
+      "clip_ratio/high_max": 0.0016201781836571172,
+      "clip_ratio/high_mean": 0.0006363311986206099,
+      "clip_ratio/low_mean": 0.00041314868667541305,
+      "clip_ratio/low_min": 1.901429823192302e-05,
+      "clip_ratio/region_mean": 0.0010494798443687614,
+      "epoch": 10.634402332361516,
+      "grad_norm": 0.13606666028499603,
+      "learning_rate": 1e-06,
+      "loss": -0.0602,
+      "step": 1028
+    },
+    {
+      "clip_ratio/high_max": 0.002066865847154986,
+      "clip_ratio/high_mean": 0.0008451949615846388,
+      "clip_ratio/low_mean": 0.00043474988342495635,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012799448377336375,
+      "epoch": 10.643731778425655,
+      "grad_norm": 0.13159887492656708,
+      "learning_rate": 1e-06,
+      "loss": -0.0591,
+      "step": 1029
+    },
+    {
+      "clip_ratio/high_max": 0.002269327269459609,
+      "clip_ratio/high_mean": 0.0009070521882676985,
+      "clip_ratio/low_mean": 0.00044856136355519993,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013556135527323931,
+      "epoch": 10.653061224489797,
+      "grad_norm": 0.14049817621707916,
+      "learning_rate": 1e-06,
+      "loss": -0.0614,
+      "step": 1030
+    },
+    {
+      "clip_ratio/high_max": 0.0023682578103034757,
+      "clip_ratio/high_mean": 0.0008562794646422844,
+      "clip_ratio/low_mean": 0.0005246619630270288,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013809413976559881,
+      "epoch": 10.662390670553936,
+      "grad_norm": 0.15171289443969727,
+      "learning_rate": 1e-06,
+      "loss": 0.012,
+      "step": 1031
+    },
+    {
+      "clip_ratio/high_max": 0.0021879767955397256,
+      "clip_ratio/high_mean": 0.0007816203960828716,
+      "clip_ratio/low_mean": 0.0005043730348006648,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012859934104199056,
+      "epoch": 10.671720116618076,
+      "grad_norm": 0.15205611288547516,
+      "learning_rate": 1e-06,
+      "loss": -0.0093,
+      "step": 1032
+    },
+    {
+      "clip_ratio/high_max": 0.0019532801998138893,
+      "clip_ratio/high_mean": 0.0007974514246598119,
+      "clip_ratio/low_mean": 0.0005037566470491583,
+      "clip_ratio/low_min": 1.664447336224839e-05,
+      "clip_ratio/region_mean": 0.0013012080526095815,
+      "epoch": 10.681049562682215,
+      "grad_norm": 15584.228515625,
+      "learning_rate": 1e-06,
+      "loss": 0.5,
+      "step": 1033
+    },
+    {
+      "clip_ratio/high_max": 0.0018114284139301162,
+      "clip_ratio/high_mean": 0.0007672809588257223,
+      "clip_ratio/low_mean": 0.0005547672235479695,
+      "clip_ratio/low_min": 1.0955302059301175e-05,
+      "clip_ratio/region_mean": 0.0013220481705502607,
+      "epoch": 10.690379008746355,
+      "grad_norm": 0.171482652425766,
+      "learning_rate": 1e-06,
+      "loss": -0.0325,
+      "step": 1034
+    },
+    {
+      "clip_ratio/high_max": 0.0020000563999929,
+      "clip_ratio/high_mean": 0.0007531167448178167,
+      "clip_ratio/low_mean": 0.0004810703903785907,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001234187151567312,
+      "epoch": 10.699708454810496,
+      "grad_norm": 0.2236158698797226,
+      "learning_rate": 1e-06,
+      "loss": -0.0238,
+      "step": 1035
+    },
+    {
+      "clip_ratio/high_max": 0.0020849624852417037,
+      "clip_ratio/high_mean": 0.0008802694610494655,
+      "clip_ratio/low_mean": 0.0006394156494025083,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015196851218206575,
+      "epoch": 10.709037900874636,
+      "grad_norm": 0.5721369981765747,
+      "learning_rate": 1e-06,
+      "loss": -0.0393,
+      "step": 1036
+    },
+    {
+      "clip_ratio/high_max": 0.0017929532914422452,
+      "clip_ratio/high_mean": 0.0008235112454713089,
+      "clip_ratio/low_mean": 0.0007058635137582314,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015293747892428655,
+      "epoch": 10.718367346938775,
+      "grad_norm": 0.14972050487995148,
+      "learning_rate": 1e-06,
+      "loss": 0.0004,
+      "step": 1037
+    },
+    {
+      "clip_ratio/high_max": 0.002275098788231844,
+      "clip_ratio/high_mean": 0.00086568585538771,
+      "clip_ratio/low_mean": 0.0006933855574970949,
+      "clip_ratio/low_min": 4.1519044316373765e-05,
+      "clip_ratio/region_mean": 0.0015590714174322784,
+      "epoch": 10.727696793002915,
+      "grad_norm": 0.14840513467788696,
+      "learning_rate": 1e-06,
+      "loss": -0.0305,
+      "step": 1038
+    },
+    {
+      "clip_ratio/high_max": 0.002317817572475178,
+      "clip_ratio/high_mean": 0.0008742880145291565,
+      "clip_ratio/low_mean": 0.00068940183155064,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015636898024240509,
+      "epoch": 10.737026239067056,
+      "grad_norm": 0.8079626560211182,
+      "learning_rate": 1e-06,
+      "loss": 0.0016,
+      "step": 1039
+    },
+    {
+      "clip_ratio/high_max": 0.0022558583441423252,
+      "clip_ratio/high_mean": 0.0009273416599171469,
+      "clip_ratio/low_mean": 0.0007737451942375628,
+      "clip_ratio/low_min": 1.4507892046822235e-05,
+      "clip_ratio/region_mean": 0.0017010868796205614,
+      "epoch": 10.746355685131196,
+      "grad_norm": 0.14360742270946503,
+      "learning_rate": 1e-06,
+      "loss": -0.036,
+      "step": 1040
+    },
+    {
+      "clip_ratio/high_max": 0.0023380135316983797,
+      "clip_ratio/high_mean": 0.0010714017262216657,
+      "clip_ratio/low_mean": 0.0007289969998964807,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018003987133852206,
+      "epoch": 10.755685131195335,
+      "grad_norm": 1.0572714805603027,
+      "learning_rate": 1e-06,
+      "loss": -0.0315,
+      "step": 1041
+    },
+    {
+      "clip_ratio/high_max": 0.0025086066671065055,
+      "clip_ratio/high_mean": 0.0009457175438001286,
+      "clip_ratio/low_mean": 0.0006730296145178727,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016187471919693053,
+      "epoch": 10.765014577259475,
+      "grad_norm": 0.13252829015254974,
+      "learning_rate": 1e-06,
+      "loss": -0.0452,
+      "step": 1042
+    },
+    {
+      "clip_ratio/high_max": 0.002531970283598639,
+      "clip_ratio/high_mean": 0.0009256004741473589,
+      "clip_ratio/low_mean": 0.0007404327625408769,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016660332257742994,
+      "epoch": 10.774344023323614,
+      "grad_norm": 0.1442527323961258,
+      "learning_rate": 1e-06,
+      "loss": -0.0367,
+      "step": 1043
+    },
+    {
+      "clip_ratio/high_max": 0.0022089687918196432,
+      "clip_ratio/high_mean": 0.0007793087006575661,
+      "clip_ratio/low_mean": 0.0007593696791445836,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015386783597932663,
+      "epoch": 10.783673469387756,
+      "grad_norm": 0.12284589558839798,
+      "learning_rate": 1e-06,
+      "loss": -0.0211,
+      "step": 1044
+    },
+    {
+      "clip_ratio/high_max": 0.0022242231389100198,
+      "clip_ratio/high_mean": 0.0010228058781649452,
+      "clip_ratio/low_mean": 0.0008191006645574817,
+      "clip_ratio/low_min": 2.860411950678099e-05,
+      "clip_ratio/region_mean": 0.001841906581830699,
+      "epoch": 10.793002915451895,
+      "grad_norm": 0.15630225837230682,
+      "learning_rate": 1e-06,
+      "loss": -0.0672,
+      "step": 1045
+    },
+    {
+      "clip_ratio/high_max": 0.002444708348775748,
+      "clip_ratio/high_mean": 0.0009511769749224186,
+      "clip_ratio/low_mean": 0.00080776714366948,
+      "clip_ratio/low_min": 1.2603347386175301e-05,
+      "clip_ratio/region_mean": 0.0017589441340533085,
+      "epoch": 10.802332361516035,
+      "grad_norm": 0.16826385259628296,
+      "learning_rate": 1e-06,
+      "loss": -0.004,
+      "step": 1046
+    },
+    {
+      "clip_ratio/high_max": 0.0023710838868282735,
+      "clip_ratio/high_mean": 0.0009455571434955345,
+      "clip_ratio/low_mean": 0.0007226421157611185,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016681992783560418,
+      "epoch": 10.811661807580174,
+      "grad_norm": 0.16420753300189972,
+      "learning_rate": 1e-06,
+      "loss": -0.0201,
+      "step": 1047
+    },
+    {
+      "clip_ratio/high_max": 0.002061777740891557,
+      "clip_ratio/high_mean": 0.0008383155145565979,
+      "clip_ratio/low_mean": 0.0008746818220970454,
+      "clip_ratio/low_min": 5.1062092097708955e-05,
+      "clip_ratio/region_mean": 0.001712997331196675,
+      "epoch": 10.820991253644316,
+      "grad_norm": 0.1311642974615097,
+      "learning_rate": 1e-06,
+      "loss": 0.0009,
+      "step": 1048
+    },
+    {
+      "clip_ratio/high_max": 0.0023175532405730337,
+      "clip_ratio/high_mean": 0.0009104331784328679,
+      "clip_ratio/low_mean": 0.0007680237549720914,
+      "clip_ratio/low_min": 1.806358341127634e-05,
+      "clip_ratio/region_mean": 0.00167845691612456,
+      "epoch": 10.830320699708455,
+      "grad_norm": 0.14354373514652252,
+      "learning_rate": 1e-06,
+      "loss": -0.0456,
+      "step": 1049
+    },
+    {
+      "clip_ratio/high_max": 0.0026579302721074782,
+      "clip_ratio/high_mean": 0.0010823156262631528,
+      "clip_ratio/low_mean": 0.0008512995691489778,
+      "clip_ratio/low_min": 3.4530385164543986e-05,
+      "clip_ratio/region_mean": 0.0019336151381139643,
+      "epoch": 10.839650145772595,
+      "grad_norm": 0.1458643078804016,
+      "learning_rate": 1e-06,
+      "loss": -0.0411,
+      "step": 1050
+    },
+    {
+      "clip_ratio/high_max": 0.002385486623097677,
+      "clip_ratio/high_mean": 0.0009239543069270439,
+      "clip_ratio/low_mean": 0.0006246837492653867,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015486380470974836,
+      "epoch": 10.848979591836734,
+      "grad_norm": 0.12350092083215714,
+      "learning_rate": 1e-06,
+      "loss": -0.0407,
+      "step": 1051
+    },
+    {
+      "clip_ratio/high_max": 0.002507519064238295,
+      "clip_ratio/high_mean": 0.001014399287669221,
+      "clip_ratio/low_mean": 0.0009685164277470903,
+      "clip_ratio/low_min": 2.2030313630239107e-05,
+      "clip_ratio/region_mean": 0.0019829157354251947,
+      "epoch": 10.858309037900874,
+      "grad_norm": 0.1365729421377182,
+      "learning_rate": 1e-06,
+      "loss": -0.0484,
+      "step": 1052
+    },
+    {
+      "clip_ratio/high_max": 0.0026707619763328694,
+      "clip_ratio/high_mean": 0.0010016462292696815,
+      "clip_ratio/low_mean": 0.0008648984112369362,
+      "clip_ratio/low_min": 2.8014343115501106e-05,
+      "clip_ratio/region_mean": 0.0018665446332306601,
+      "epoch": 10.867638483965015,
+      "grad_norm": 0.17057707905769348,
+      "learning_rate": 1e-06,
+      "loss": -0.0409,
+      "step": 1053
+    },
+    {
+      "clip_ratio/high_max": 0.002549129589169752,
+      "clip_ratio/high_mean": 0.0010482151537871687,
+      "clip_ratio/low_mean": 0.0009107267833314836,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019589419753174298,
+      "epoch": 10.876967930029155,
+      "grad_norm": 0.32911109924316406,
+      "learning_rate": 1e-06,
+      "loss": -0.0245,
+      "step": 1054
+    },
+    {
+      "clip_ratio/high_max": 0.002313146091182716,
+      "clip_ratio/high_mean": 0.000970059651081101,
+      "clip_ratio/low_mean": 0.0009378099002788076,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019078695331700146,
+      "epoch": 10.886297376093294,
+      "grad_norm": 0.14788420498371124,
+      "learning_rate": 1e-06,
+      "loss": -0.0046,
+      "step": 1055
+    },
+    {
+      "clip_ratio/high_max": 0.002571078708569985,
+      "clip_ratio/high_mean": 0.00102688998413214,
+      "clip_ratio/low_mean": 0.0008060240352278925,
+      "clip_ratio/low_min": 1.539029835839756e-05,
+      "clip_ratio/region_mean": 0.0018329140439163893,
+      "epoch": 10.895626822157434,
+      "grad_norm": 0.14282891154289246,
+      "learning_rate": 1e-06,
+      "loss": -0.0094,
+      "step": 1056
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0776715959821429,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4023.0,
+      "completions/mean_length": 813.9260864257812,
+      "completions/mean_terminated_length": 537.5343627929688,
+      "completions/min_length": 96.0,
+      "completions/min_terminated_length": 96.0,
+      "epoch": 11.00932944606414,
+      "grad_norm": 0.15120987594127655,
+      "learning_rate": 1e-06,
+      "loss": -0.0189,
+      "num_tokens": 612352642.0,
+      "reward": 0.6597726345062256,
+      "reward_std": 0.1406591534614563,
+      "rewards/simpleverify_reward/mean": 0.6597725749015808,
+      "rewards/simpleverify_reward/std": 0.47379374504089355,
+      "step": 1057
+    },
+    {
+      "clip_ratio/high_max": 0.0015920113619358744,
+      "clip_ratio/high_mean": 0.0006614641097257845,
+      "clip_ratio/low_mean": 0.0003684297939798853,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010298939087078907,
+      "epoch": 11.018658892128279,
+      "grad_norm": 0.12487295269966125,
+      "learning_rate": 1e-06,
+      "loss": -0.042,
+      "step": 1058
+    },
+    {
+      "clip_ratio/high_max": 0.0018454749588272534,
+      "clip_ratio/high_mean": 0.0007487962429877371,
+      "clip_ratio/low_mean": 0.00045943752138555283,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012082337161700707,
+      "epoch": 11.02798833819242,
+      "grad_norm": 0.1372573971748352,
+      "learning_rate": 1e-06,
+      "loss": -0.0206,
+      "step": 1059
+    },
+    {
+      "clip_ratio/high_max": 0.0018122699639206985,
+      "clip_ratio/high_mean": 0.0006949476301087998,
+      "clip_ratio/low_mean": 0.0004402407466841396,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011351884058967698,
+      "epoch": 11.03731778425656,
+      "grad_norm": 0.12935875356197357,
+      "learning_rate": 1e-06,
+      "loss": -0.0111,
+      "step": 1060
+    },
+    {
+      "clip_ratio/high_max": 0.0018673782069527078,
+      "clip_ratio/high_mean": 0.0007298047512449557,
+      "clip_ratio/low_mean": 0.00034867336216848344,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010784781370603014,
+      "epoch": 11.0466472303207,
+      "grad_norm": 0.1439608484506607,
+      "learning_rate": 1e-06,
+      "loss": -0.1065,
+      "step": 1061
+    },
+    {
+      "clip_ratio/high_max": 0.0016545736907573882,
+      "clip_ratio/high_mean": 0.0006580228446182446,
+      "clip_ratio/low_mean": 0.0004363252883194946,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010943481265712762,
+      "epoch": 11.055976676384839,
+      "grad_norm": 0.14292700588703156,
+      "learning_rate": 1e-06,
+      "loss": -0.0589,
+      "step": 1062
+    },
+    {
+      "clip_ratio/high_max": 0.002028540722676553,
+      "clip_ratio/high_mean": 0.000700796017099492,
+      "clip_ratio/low_mean": 0.0006795988219892024,
+      "clip_ratio/low_min": 2.517019493097905e-05,
+      "clip_ratio/region_mean": 0.001380394820444053,
+      "epoch": 11.06530612244898,
+      "grad_norm": 1.630363941192627,
+      "learning_rate": 1e-06,
+      "loss": 0.0159,
+      "step": 1063
+    },
+    {
+      "clip_ratio/high_max": 0.002253053469758015,
+      "clip_ratio/high_mean": 0.0008082645563263213,
+      "clip_ratio/low_mean": 0.0005325971733327606,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013408617378445342,
+      "epoch": 11.07463556851312,
+      "grad_norm": 0.15001767873764038,
+      "learning_rate": 1e-06,
+      "loss": -0.0407,
+      "step": 1064
+    },
+    {
+      "clip_ratio/high_max": 0.0017601002182345837,
+      "clip_ratio/high_mean": 0.0006535378470289288,
+      "clip_ratio/low_mean": 0.0005012225556129124,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011547603953658836,
+      "epoch": 11.08396501457726,
+      "grad_norm": 0.12722033262252808,
+      "learning_rate": 1e-06,
+      "loss": -0.018,
+      "step": 1065
+    },
+    {
+      "clip_ratio/high_max": 0.0016888787613424938,
+      "clip_ratio/high_mean": 0.000663585405163758,
+      "clip_ratio/low_mean": 0.0005574824645009357,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001221067886945093,
+      "epoch": 11.093294460641399,
+      "grad_norm": 0.1354394406080246,
+      "learning_rate": 1e-06,
+      "loss": -0.0403,
+      "step": 1066
+    },
+    {
+      "clip_ratio/high_max": 0.002266250623506494,
+      "clip_ratio/high_mean": 0.0008081148334895261,
+      "clip_ratio/low_mean": 0.0005596310902546975,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001367745946481591,
+      "epoch": 11.102623906705539,
+      "grad_norm": 0.13672518730163574,
+      "learning_rate": 1e-06,
+      "loss": -0.0069,
+      "step": 1067
+    },
+    {
+      "clip_ratio/high_max": 0.0021165791622479446,
+      "clip_ratio/high_mean": 0.0008931939355534269,
+      "clip_ratio/low_mean": 0.0007323976897168905,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016255916525551584,
+      "epoch": 11.11195335276968,
+      "grad_norm": 0.16483911871910095,
+      "learning_rate": 1e-06,
+      "loss": -0.0311,
+      "step": 1068
+    },
+    {
+      "clip_ratio/high_max": 0.001983118178031873,
+      "clip_ratio/high_mean": 0.0007438092798111029,
+      "clip_ratio/low_mean": 0.0008516333882653271,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015954426853568293,
+      "epoch": 11.12128279883382,
+      "grad_norm": 0.1664646863937378,
+      "learning_rate": 1e-06,
+      "loss": -0.0162,
+      "step": 1069
+    },
+    {
+      "clip_ratio/high_max": 0.0021535297055379488,
+      "clip_ratio/high_mean": 0.0008280100446427241,
+      "clip_ratio/low_mean": 0.0005900636169826612,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00141807364707347,
+      "epoch": 11.130612244897959,
+      "grad_norm": 0.195279061794281,
+      "learning_rate": 1e-06,
+      "loss": -0.0162,
+      "step": 1070
+    },
+    {
+      "clip_ratio/high_max": 0.0019865077701979317,
+      "clip_ratio/high_mean": 0.0008226514473790303,
+      "clip_ratio/low_mean": 0.0006189503556015552,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014416018129850272,
+      "epoch": 11.139941690962099,
+      "grad_norm": 0.15905873477458954,
+      "learning_rate": 1e-06,
+      "loss": -0.0485,
+      "step": 1071
+    },
+    {
+      "clip_ratio/high_max": 0.00243154483177932,
+      "clip_ratio/high_mean": 0.0009928193921950879,
+      "clip_ratio/low_mean": 0.0006957122373023594,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016885316363186575,
+      "epoch": 11.14927113702624,
+      "grad_norm": 0.13798582553863525,
+      "learning_rate": 1e-06,
+      "loss": -0.0599,
+      "step": 1072
+    },
+    {
+      "clip_ratio/high_max": 0.0019647090084617957,
+      "clip_ratio/high_mean": 0.000768944821174955,
+      "clip_ratio/low_mean": 0.0008300694935314823,
+      "clip_ratio/low_min": 3.128128082607873e-05,
+      "clip_ratio/region_mean": 0.0015990143037925009,
+      "epoch": 11.15860058309038,
+      "grad_norm": 0.16089555621147156,
+      "learning_rate": 1e-06,
+      "loss": -0.0308,
+      "step": 1073
+    },
+    {
+      "clip_ratio/high_max": 0.002555462058808189,
+      "clip_ratio/high_mean": 0.0009445642044738634,
+      "clip_ratio/low_mean": 0.0009199578753396054,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018645220552571118,
+      "epoch": 11.167930029154519,
+      "grad_norm": 0.15208464860916138,
+      "learning_rate": 1e-06,
+      "loss": -0.0245,
+      "step": 1074
+    },
+    {
+      "clip_ratio/high_max": 0.002173075459722895,
+      "clip_ratio/high_mean": 0.0008532288757123752,
+      "clip_ratio/low_mean": 0.001072563537491078,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019257924432167783,
+      "epoch": 11.177259475218658,
+      "grad_norm": 0.13196150958538055,
+      "learning_rate": 1e-06,
+      "loss": -0.0531,
+      "step": 1075
+    },
+    {
+      "clip_ratio/high_max": 0.0022783958047511987,
+      "clip_ratio/high_mean": 0.0008463830054097343,
+      "clip_ratio/low_mean": 0.0007740050641587004,
+      "clip_ratio/low_min": 5.8647666264732834e-05,
+      "clip_ratio/region_mean": 0.0016203880804823712,
+      "epoch": 11.186588921282798,
+      "grad_norm": 0.15812310576438904,
+      "learning_rate": 1e-06,
+      "loss": -0.0209,
+      "step": 1076
+    },
+    {
+      "clip_ratio/high_max": 0.0025219095041393302,
+      "clip_ratio/high_mean": 0.0008976641238405136,
+      "clip_ratio/low_mean": 0.0009242955366062233,
+      "clip_ratio/low_min": 7.02002398611512e-05,
+      "clip_ratio/region_mean": 0.0018219596859125886,
+      "epoch": 11.19591836734694,
+      "grad_norm": 0.14058281481266022,
+      "learning_rate": 1e-06,
+      "loss": -0.0031,
+      "step": 1077
+    },
+    {
+      "clip_ratio/high_max": 0.0021862751091248356,
+      "clip_ratio/high_mean": 0.0009059881931534619,
+      "clip_ratio/low_mean": 0.0009726934404170606,
+      "clip_ratio/low_min": 1.5151515071920585e-05,
+      "clip_ratio/region_mean": 0.0018786816399369854,
+      "epoch": 11.205247813411079,
+      "grad_norm": 0.13323284685611725,
+      "learning_rate": 1e-06,
+      "loss": -0.0022,
+      "step": 1078
+    },
+    {
+      "clip_ratio/high_max": 0.002862885819922667,
+      "clip_ratio/high_mean": 0.0011893186674569733,
+      "clip_ratio/low_mean": 0.0008400276219617808,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020293462694098707,
+      "epoch": 11.214577259475218,
+      "grad_norm": 0.12973734736442566,
+      "learning_rate": 1e-06,
+      "loss": -0.0907,
+      "step": 1079
+    },
+    {
+      "clip_ratio/high_max": 0.002883542016206775,
+      "clip_ratio/high_mean": 0.0010056324135803152,
+      "clip_ratio/low_mean": 0.0008221942205182131,
+      "clip_ratio/low_min": 1.2067966963513754e-05,
+      "clip_ratio/region_mean": 0.0018278266215929762,
+      "epoch": 11.223906705539358,
+      "grad_norm": 0.1280481070280075,
+      "learning_rate": 1e-06,
+      "loss": -0.0327,
+      "step": 1080
+    },
+    {
+      "clip_ratio/high_max": 0.0023884848778834566,
+      "clip_ratio/high_mean": 0.0009425718526472338,
+      "clip_ratio/low_mean": 0.0008561756294511724,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017987474566325545,
+      "epoch": 11.2332361516035,
+      "grad_norm": 0.16232584416866302,
+      "learning_rate": 1e-06,
+      "loss": -0.0307,
+      "step": 1081
+    },
+    {
+      "clip_ratio/high_max": 0.0026021743105957285,
+      "clip_ratio/high_mean": 0.001054947157172137,
+      "clip_ratio/low_mean": 0.0008475461308989907,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019024933026230428,
+      "epoch": 11.242565597667639,
+      "grad_norm": 0.1246662512421608,
+      "learning_rate": 1e-06,
+      "loss": -0.0312,
+      "step": 1082
+    },
+    {
+      "clip_ratio/high_max": 0.0024336452333955094,
+      "clip_ratio/high_mean": 0.0009560207709000679,
+      "clip_ratio/low_mean": 0.0007528945643571205,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017089152934204321,
+      "epoch": 11.251895043731778,
+      "grad_norm": 0.16376838088035583,
+      "learning_rate": 1e-06,
+      "loss": -0.046,
+      "step": 1083
+    },
+    {
+      "clip_ratio/high_max": 0.002522799048165325,
+      "clip_ratio/high_mean": 0.001010906449664617,
+      "clip_ratio/low_mean": 0.0008124366049742093,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018233430819236673,
+      "epoch": 11.261224489795918,
+      "grad_norm": 0.13568460941314697,
+      "learning_rate": 1e-06,
+      "loss": -0.0469,
+      "step": 1084
+    },
+    {
+      "clip_ratio/high_max": 0.002387663222179981,
+      "clip_ratio/high_mean": 0.0009682664513093187,
+      "clip_ratio/low_mean": 0.0007788152188368258,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017470816892455332,
+      "epoch": 11.270553935860057,
+      "grad_norm": 0.12385565042495728,
+      "learning_rate": 1e-06,
+      "loss": -0.0633,
+      "step": 1085
+    },
+    {
+      "clip_ratio/high_max": 0.0021982938997098245,
+      "clip_ratio/high_mean": 0.0009584870022081304,
+      "clip_ratio/low_mean": 0.0010174376839131583,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019759247079491615,
+      "epoch": 11.279883381924199,
+      "grad_norm": 0.14474475383758545,
+      "learning_rate": 1e-06,
+      "loss": -0.0035,
+      "step": 1086
+    },
+    {
+      "clip_ratio/high_max": 0.002302657645486761,
+      "clip_ratio/high_mean": 0.0010387126931163948,
+      "clip_ratio/low_mean": 0.0007650050119991647,
+      "clip_ratio/low_min": 3.943217598134652e-05,
+      "clip_ratio/region_mean": 0.0018037176887446549,
+      "epoch": 11.289212827988338,
+      "grad_norm": 0.13189628720283508,
+      "learning_rate": 1e-06,
+      "loss": -0.0493,
+      "step": 1087
+    },
+    {
+      "clip_ratio/high_max": 0.0030264778033597395,
+      "clip_ratio/high_mean": 0.0011500599903229158,
+      "clip_ratio/low_mean": 0.0010938255545624997,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0022438855885411613,
+      "epoch": 11.298542274052478,
+      "grad_norm": 0.14285577833652496,
+      "learning_rate": 1e-06,
+      "loss": -0.0025,
+      "step": 1088
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0746023995535714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4005.0,
+      "completions/mean_length": 801.2931518554688,
+      "completions/mean_terminated_length": 535.6849975585938,
+      "completions/min_length": 93.0,
+      "completions/min_terminated_length": 93.0,
+      "epoch": 11.307871720116617,
+      "grad_norm": 0.1754704862833023,
+      "learning_rate": 1e-06,
+      "loss": -0.0253,
+      "num_tokens": 629390973.0,
+      "reward": 0.6665388345718384,
+      "reward_std": 0.14645832777023315,
+      "rewards/simpleverify_reward/mean": 0.6665387749671936,
+      "rewards/simpleverify_reward/std": 0.47145789861679077,
+      "step": 1089
+    },
+    {
+      "clip_ratio/high_max": 0.0017820471184677444,
+      "clip_ratio/high_mean": 0.0006932171554581146,
+      "clip_ratio/low_mean": 0.00031046748927110457,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010036846579168923,
+      "epoch": 11.317201166180759,
+      "grad_norm": 0.1414964497089386,
+      "learning_rate": 1e-06,
+      "loss": -0.0402,
+      "step": 1090
+    },
+    {
+      "clip_ratio/high_max": 0.001742104286677204,
+      "clip_ratio/high_mean": 0.0007483205354219535,
+      "clip_ratio/low_mean": 0.00037099295786902076,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011193134923814796,
+      "epoch": 11.326530612244898,
+      "grad_norm": 0.14169956743717194,
+      "learning_rate": 1e-06,
+      "loss": -0.0568,
+      "step": 1091
+    },
+    {
+      "clip_ratio/high_max": 0.0015461074726772495,
+      "clip_ratio/high_mean": 0.0006735009446856566,
+      "clip_ratio/low_mean": 0.0004402636313898256,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011137645651615458,
+      "epoch": 11.335860058309038,
+      "grad_norm": 0.11127534508705139,
+      "learning_rate": 1e-06,
+      "loss": -0.0299,
+      "step": 1092
+    },
+    {
+      "clip_ratio/high_max": 0.0017622153463889845,
+      "clip_ratio/high_mean": 0.000721482259905315,
+      "clip_ratio/low_mean": 0.00044460590356720786,
+      "clip_ratio/low_min": 1.5042118320707232e-05,
+      "clip_ratio/region_mean": 0.001166088150057476,
+      "epoch": 11.345189504373177,
+      "grad_norm": 0.14358475804328918,
+      "learning_rate": 1e-06,
+      "loss": -0.0506,
+      "step": 1093
+    },
+    {
+      "clip_ratio/high_max": 0.002066727141937008,
+      "clip_ratio/high_mean": 0.0008654190587549238,
+      "clip_ratio/low_mean": 0.000463999939711357,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013294189920998178,
+      "epoch": 11.354518950437317,
+      "grad_norm": 0.1574108600616455,
+      "learning_rate": 1e-06,
+      "loss": -0.0665,
+      "step": 1094
+    },
+    {
+      "clip_ratio/high_max": 0.0022268158791121095,
+      "clip_ratio/high_mean": 0.0008985387466964312,
+      "clip_ratio/low_mean": 0.0005623355675652419,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014608743585995398,
+      "epoch": 11.363848396501458,
+      "grad_norm": 0.14443717896938324,
+      "learning_rate": 1e-06,
+      "loss": -0.079,
+      "step": 1095
+    },
+    {
+      "clip_ratio/high_max": 0.0020587054823408835,
+      "clip_ratio/high_mean": 0.0007668820617254823,
+      "clip_ratio/low_mean": 0.0005758043498644838,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001342686424322892,
+      "epoch": 11.373177842565598,
+      "grad_norm": 0.1323336809873581,
+      "learning_rate": 1e-06,
+      "loss": -0.0268,
+      "step": 1096
+    },
+    {
+      "clip_ratio/high_max": 0.0019139542564516887,
+      "clip_ratio/high_mean": 0.0007136751073630876,
+      "clip_ratio/low_mean": 0.0005419626295406488,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001255637744179694,
+      "epoch": 11.382507288629737,
+      "grad_norm": 0.16975180804729462,
+      "learning_rate": 1e-06,
+      "loss": -0.0369,
+      "step": 1097
+    },
+    {
+      "clip_ratio/high_max": 0.0016596432360529434,
+      "clip_ratio/high_mean": 0.0006236433364392724,
+      "clip_ratio/low_mean": 0.0006460134773078607,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012696568155661225,
+      "epoch": 11.391836734693877,
+      "grad_norm": 0.14871634542942047,
+      "learning_rate": 1e-06,
+      "loss": 0.0164,
+      "step": 1098
+    },
+    {
+      "clip_ratio/high_max": 0.0024096415436360985,
+      "clip_ratio/high_mean": 0.0008074062152445549,
+      "clip_ratio/low_mean": 0.0005820828559990332,
+      "clip_ratio/low_min": 1.2254901776032057e-05,
+      "clip_ratio/region_mean": 0.0013894890835217666,
+      "epoch": 11.401166180758018,
+      "grad_norm": 0.1535893678665161,
+      "learning_rate": 1e-06,
+      "loss": -0.0605,
+      "step": 1099
+    },
+    {
+      "clip_ratio/high_max": 0.0019717682516784407,
+      "clip_ratio/high_mean": 0.0007068977811286459,
+      "clip_ratio/low_mean": 0.0006789030367144733,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001385800806019688,
+      "epoch": 11.410495626822158,
+      "grad_norm": 0.15019330382347107,
+      "learning_rate": 1e-06,
+      "loss": -0.0229,
+      "step": 1100
+    },
+    {
+      "clip_ratio/high_max": 0.002162405078706797,
+      "clip_ratio/high_mean": 0.0009733231090649497,
+      "clip_ratio/low_mean": 0.0006966889504838036,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016700120613677427,
+      "epoch": 11.419825072886297,
+      "grad_norm": 0.14072251319885254,
+      "learning_rate": 1e-06,
+      "loss": -0.0583,
+      "step": 1101
+    },
+    {
+      "clip_ratio/high_max": 0.0024383579584537074,
+      "clip_ratio/high_mean": 0.0009564513356963289,
+      "clip_ratio/low_mean": 0.0007808191749063553,
+      "clip_ratio/low_min": 2.9531922336900607e-05,
+      "clip_ratio/region_mean": 0.0017372705551679246,
+      "epoch": 11.429154518950437,
+      "grad_norm": 0.15458250045776367,
+      "learning_rate": 1e-06,
+      "loss": -0.0182,
+      "step": 1102
+    },
+    {
+      "clip_ratio/high_max": 0.0020752936543431133,
+      "clip_ratio/high_mean": 0.0009176840958389221,
+      "clip_ratio/low_mean": 0.0007021674146017176,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016198515222640708,
+      "epoch": 11.438483965014576,
+      "grad_norm": 0.13857978582382202,
+      "learning_rate": 1e-06,
+      "loss": -0.0262,
+      "step": 1103
+    },
+    {
+      "clip_ratio/high_max": 0.0022316328904707916,
+      "clip_ratio/high_mean": 0.0009267121322409366,
+      "clip_ratio/low_mean": 0.0006105196152930148,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001537231779366266,
+      "epoch": 11.447813411078718,
+      "grad_norm": 0.12467081099748611,
+      "learning_rate": 1e-06,
+      "loss": -0.0285,
+      "step": 1104
+    },
+    {
+      "clip_ratio/high_max": 0.00216190998617094,
+      "clip_ratio/high_mean": 0.0008915891085052863,
+      "clip_ratio/low_mean": 0.0007676189543417422,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016592080792179331,
+      "epoch": 11.457142857142857,
+      "grad_norm": 0.14669372141361237,
+      "learning_rate": 1e-06,
+      "loss": -0.0145,
+      "step": 1105
+    },
+    {
+      "clip_ratio/high_max": 0.0025387482783116866,
+      "clip_ratio/high_mean": 0.0009284275101890671,
+      "clip_ratio/low_mean": 0.0005932793610554654,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015217068685160484,
+      "epoch": 11.466472303206997,
+      "grad_norm": 0.124748095870018,
+      "learning_rate": 1e-06,
+      "loss": -0.0241,
+      "step": 1106
+    },
+    {
+      "clip_ratio/high_max": 0.002182352251111297,
+      "clip_ratio/high_mean": 0.0008965330180217279,
+      "clip_ratio/low_mean": 0.0006304568923951592,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015269898794940673,
+      "epoch": 11.475801749271136,
+      "grad_norm": 0.14877058565616608,
+      "learning_rate": 1e-06,
+      "loss": -0.0382,
+      "step": 1107
+    },
+    {
+      "clip_ratio/high_max": 0.0022348828715621494,
+      "clip_ratio/high_mean": 0.0009596692616469227,
+      "clip_ratio/low_mean": 0.0006714849441777915,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016311542167386506,
+      "epoch": 11.485131195335278,
+      "grad_norm": 0.22986245155334473,
+      "learning_rate": 1e-06,
+      "loss": -0.0471,
+      "step": 1108
+    },
+    {
+      "clip_ratio/high_max": 0.0021594388381345198,
+      "clip_ratio/high_mean": 0.0008019785054784734,
+      "clip_ratio/low_mean": 0.0007085141814968665,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015104926933418028,
+      "epoch": 11.494460641399417,
+      "grad_norm": 0.15406882762908936,
+      "learning_rate": 1e-06,
+      "loss": -0.0038,
+      "step": 1109
+    },
+    {
+      "clip_ratio/high_max": 0.002414992093690671,
+      "clip_ratio/high_mean": 0.0009492818662693026,
+      "clip_ratio/low_mean": 0.0006458067891799146,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015950886445352808,
+      "epoch": 11.503790087463557,
+      "grad_norm": 0.16856832802295685,
+      "learning_rate": 1e-06,
+      "loss": -0.0496,
+      "step": 1110
+    },
+    {
+      "clip_ratio/high_max": 0.002206098590249894,
+      "clip_ratio/high_mean": 0.0007776361526339315,
+      "clip_ratio/low_mean": 0.0007788886541675311,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015565247886115685,
+      "epoch": 11.513119533527696,
+      "grad_norm": 0.1853359490633011,
+      "learning_rate": 1e-06,
+      "loss": 0.0125,
+      "step": 1111
+    },
+    {
+      "clip_ratio/high_max": 0.002284768743265886,
+      "clip_ratio/high_mean": 0.0009445244395465124,
+      "clip_ratio/low_mean": 0.0008526741839887109,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017971985798794776,
+      "epoch": 11.522448979591836,
+      "grad_norm": 0.16563715040683746,
+      "learning_rate": 1e-06,
+      "loss": -0.0036,
+      "step": 1112
+    },
+    {
+      "clip_ratio/high_max": 0.0026219166393275373,
+      "clip_ratio/high_mean": 0.0010779583171824925,
+      "clip_ratio/low_mean": 0.0007722936206846498,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018502519378671423,
+      "epoch": 11.531778425655977,
+      "grad_norm": 0.15771019458770752,
+      "learning_rate": 1e-06,
+      "loss": -0.0478,
+      "step": 1113
+    },
+    {
+      "clip_ratio/high_max": 0.00226580751768779,
+      "clip_ratio/high_mean": 0.0009537130390526727,
+      "clip_ratio/low_mean": 0.0008044518890528707,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017581649153726175,
+      "epoch": 11.541107871720117,
+      "grad_norm": 0.14264440536499023,
+      "learning_rate": 1e-06,
+      "loss": -0.0349,
+      "step": 1114
+    },
+    {
+      "clip_ratio/high_max": 0.0023334657889790833,
+      "clip_ratio/high_mean": 0.0008980763013823889,
+      "clip_ratio/low_mean": 0.0006712785125273513,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015693548193667084,
+      "epoch": 11.550437317784256,
+      "grad_norm": 0.15568388998508453,
+      "learning_rate": 1e-06,
+      "loss": -0.0144,
+      "step": 1115
+    },
+    {
+      "clip_ratio/high_max": 0.0024960231021395884,
+      "clip_ratio/high_mean": 0.001090276135073509,
+      "clip_ratio/low_mean": 0.0008367642330995295,
+      "clip_ratio/low_min": 2.8674108762061223e-05,
+      "clip_ratio/region_mean": 0.0019270403863629326,
+      "epoch": 11.559766763848396,
+      "grad_norm": 0.2161525934934616,
+      "learning_rate": 1e-06,
+      "loss": -0.0224,
+      "step": 1116
+    },
+    {
+      "clip_ratio/high_max": 0.002084986503177788,
+      "clip_ratio/high_mean": 0.0008541378192603588,
+      "clip_ratio/low_mean": 0.0006564938057636027,
+      "clip_ratio/low_min": 1.7346656022709794e-05,
+      "clip_ratio/region_mean": 0.001510631660494255,
+      "epoch": 11.569096209912537,
+      "grad_norm": 0.14218589663505554,
+      "learning_rate": 1e-06,
+      "loss": -0.0287,
+      "step": 1117
+    },
+    {
+      "clip_ratio/high_max": 0.002428565036098007,
+      "clip_ratio/high_mean": 0.0008952251610025996,
+      "clip_ratio/low_mean": 0.0007816510260454379,
+      "clip_ratio/low_min": 8.259017522505019e-05,
+      "clip_ratio/region_mean": 0.0016768761743151117,
+      "epoch": 11.578425655976677,
+      "grad_norm": 0.1381969451904297,
+      "learning_rate": 1e-06,
+      "loss": -0.0185,
+      "step": 1118
+    },
+    {
+      "clip_ratio/high_max": 0.0025781829353945795,
+      "clip_ratio/high_mean": 0.001114851766033098,
+      "clip_ratio/low_mean": 0.0008869078947100206,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020017596398247406,
+      "epoch": 11.587755102040816,
+      "grad_norm": 0.14114214479923248,
+      "learning_rate": 1e-06,
+      "loss": -0.0125,
+      "step": 1119
+    },
+    {
+      "clip_ratio/high_max": 0.00310697757959133,
+      "clip_ratio/high_mean": 0.0012362512752588373,
+      "clip_ratio/low_mean": 0.0008617528910690453,
+      "clip_ratio/low_min": 4.026196984341368e-05,
+      "clip_ratio/region_mean": 0.002098004195431713,
+      "epoch": 11.597084548104956,
+      "grad_norm": 0.14526280760765076,
+      "learning_rate": 1e-06,
+      "loss": -0.0502,
+      "step": 1120
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.082763671875,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4056.0,
+      "completions/mean_length": 828.3167114257812,
+      "completions/mean_terminated_length": 533.468505859375,
+      "completions/min_length": 5.0,
+      "completions/min_terminated_length": 5.0,
+      "epoch": 11.606413994169095,
+      "grad_norm": 0.1565808355808258,
+      "learning_rate": 1e-06,
+      "loss": -0.0216,
+      "num_tokens": 646304013.0,
+      "reward": 0.6595633625984192,
+      "reward_std": 0.1461329460144043,
+      "rewards/simpleverify_reward/mean": 0.6595633625984192,
+      "rewards/simpleverify_reward/std": 0.47386428713798523,
+      "step": 1121
+    },
+    {
+      "clip_ratio/high_max": 0.0017801259782572743,
+      "clip_ratio/high_mean": 0.0006902775126036431,
+      "clip_ratio/low_mean": 0.0003831740370969783,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010734515453805216,
+      "epoch": 11.615743440233237,
+      "grad_norm": 0.16050131618976593,
+      "learning_rate": 1e-06,
+      "loss": -0.0236,
+      "step": 1122
+    },
+    {
+      "clip_ratio/high_max": 0.002041045550868148,
+      "clip_ratio/high_mean": 0.0007911560469437973,
+      "clip_ratio/low_mean": 0.00032368482652600505,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011148408520966768,
+      "epoch": 11.625072886297376,
+      "grad_norm": 0.13591721653938293,
+      "learning_rate": 1e-06,
+      "loss": -0.049,
+      "step": 1123
+    },
+    {
+      "clip_ratio/high_max": 0.0019790680162259378,
+      "clip_ratio/high_mean": 0.0006888173520565033,
+      "clip_ratio/low_mean": 0.00039165548423625296,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010804728335642722,
+      "epoch": 11.634402332361516,
+      "grad_norm": 0.1681247055530548,
+      "learning_rate": 1e-06,
+      "loss": 0.0051,
+      "step": 1124
+    },
+    {
+      "clip_ratio/high_max": 0.0018455690842529293,
+      "clip_ratio/high_mean": 0.0006823032963438891,
+      "clip_ratio/low_mean": 0.0005648048909279169,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012471081790863536,
+      "epoch": 11.643731778425655,
+      "grad_norm": 0.13590890169143677,
+      "learning_rate": 1e-06,
+      "loss": -0.0385,
+      "step": 1125
+    },
+    {
+      "clip_ratio/high_max": 0.0015292885655071586,
+      "clip_ratio/high_mean": 0.0006225187535164878,
+      "clip_ratio/low_mean": 0.00045897044947196264,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010814892229973339,
+      "epoch": 11.653061224489797,
+      "grad_norm": 0.22801119089126587,
+      "learning_rate": 1e-06,
+      "loss": -0.017,
+      "step": 1126
+    },
+    {
+      "clip_ratio/high_max": 0.0018514734074415173,
+      "clip_ratio/high_mean": 0.0007027422670944361,
+      "clip_ratio/low_mean": 0.0005222668733040337,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012250091349415015,
+      "epoch": 11.662390670553936,
+      "grad_norm": 0.19190406799316406,
+      "learning_rate": 1e-06,
+      "loss": -0.0325,
+      "step": 1127
+    },
+    {
+      "clip_ratio/high_max": 0.002189035360061098,
+      "clip_ratio/high_mean": 0.0007538662812294206,
+      "clip_ratio/low_mean": 0.0005416949161372031,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012955612219229806,
+      "epoch": 11.671720116618076,
+      "grad_norm": 0.1414567530155182,
+      "learning_rate": 1e-06,
+      "loss": -0.0404,
+      "step": 1128
+    },
+    {
+      "clip_ratio/high_max": 0.0018500980258977506,
+      "clip_ratio/high_mean": 0.0008095502416836098,
+      "clip_ratio/low_mean": 0.0006172982866701204,
+      "clip_ratio/low_min": 1.4266149264585692e-05,
+      "clip_ratio/region_mean": 0.0014268485283537302,
+      "epoch": 11.681049562682215,
+      "grad_norm": 0.15000969171524048,
+      "learning_rate": 1e-06,
+      "loss": -0.022,
+      "step": 1129
+    },
+    {
+      "clip_ratio/high_max": 0.001987406205444131,
+      "clip_ratio/high_mean": 0.0008222470642067492,
+      "clip_ratio/low_mean": 0.000526983537383785,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013492306097759865,
+      "epoch": 11.690379008746355,
+      "grad_norm": 0.13672062754631042,
+      "learning_rate": 1e-06,
+      "loss": -0.0539,
+      "step": 1130
+    },
+    {
+      "clip_ratio/high_max": 0.0023744198042550124,
+      "clip_ratio/high_mean": 0.000983795503998408,
+      "clip_ratio/low_mean": 0.0005859847988176625,
+      "clip_ratio/low_min": 7.875242954469286e-05,
+      "clip_ratio/region_mean": 0.0015697802664362825,
+      "epoch": 11.699708454810496,
+      "grad_norm": 0.14747218787670135,
+      "learning_rate": 1e-06,
+      "loss": -0.0589,
+      "step": 1131
+    },
+    {
+      "clip_ratio/high_max": 0.0022438299274654128,
+      "clip_ratio/high_mean": 0.0008859738409228157,
+      "clip_ratio/low_mean": 0.000632581080935779,
+      "clip_ratio/low_min": 1.681011235632468e-05,
+      "clip_ratio/region_mean": 0.001518554927315563,
+      "epoch": 11.709037900874636,
+      "grad_norm": 0.16229085624217987,
+      "learning_rate": 1e-06,
+      "loss": -0.017,
+      "step": 1132
+    },
+    {
+      "clip_ratio/high_max": 0.0023319523097597994,
+      "clip_ratio/high_mean": 0.0008855273881636094,
+      "clip_ratio/low_mean": 0.0007731926016276702,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001658720000705216,
+      "epoch": 11.718367346938775,
+      "grad_norm": 0.15086589753627777,
+      "learning_rate": 1e-06,
+      "loss": 0.0011,
+      "step": 1133
+    },
+    {
+      "clip_ratio/high_max": 0.0021720065706176683,
+      "clip_ratio/high_mean": 0.0008946373563958332,
+      "clip_ratio/low_mean": 0.0007507508071284974,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016453881762572564,
+      "epoch": 11.727696793002915,
+      "grad_norm": 0.14836829900741577,
+      "learning_rate": 1e-06,
+      "loss": -0.0335,
+      "step": 1134
+    },
+    {
+      "clip_ratio/high_max": 0.002065588501864113,
+      "clip_ratio/high_mean": 0.0008268410601885989,
+      "clip_ratio/low_mean": 0.0007444371713063447,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001571278175106272,
+      "epoch": 11.737026239067056,
+      "grad_norm": 0.14221884310245514,
+      "learning_rate": 1e-06,
+      "loss": -0.0492,
+      "step": 1135
+    },
+    {
+      "clip_ratio/high_max": 0.0023058720107655972,
+      "clip_ratio/high_mean": 0.0008397777110076277,
+      "clip_ratio/low_mean": 0.000686158112330304,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015259358442563098,
+      "epoch": 11.746355685131196,
+      "grad_norm": 0.153856560587883,
+      "learning_rate": 1e-06,
+      "loss": -0.0087,
+      "step": 1136
+    },
+    {
+      "clip_ratio/high_max": 0.002365320349781541,
+      "clip_ratio/high_mean": 0.0009605860432202462,
+      "clip_ratio/low_mean": 0.0005241689832473639,
+      "clip_ratio/low_min": 1.842027631937526e-05,
+      "clip_ratio/region_mean": 0.001484755066485377,
+      "epoch": 11.755685131195335,
+      "grad_norm": 0.15080800652503967,
+      "learning_rate": 1e-06,
+      "loss": -0.0715,
+      "step": 1137
+    },
+    {
+      "clip_ratio/high_max": 0.0025613146572140977,
+      "clip_ratio/high_mean": 0.0009624600697861752,
+      "clip_ratio/low_mean": 0.0006789249946450582,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016413850753451698,
+      "epoch": 11.765014577259475,
+      "grad_norm": 0.1395125538110733,
+      "learning_rate": 1e-06,
+      "loss": -0.0699,
+      "step": 1138
+    },
+    {
+      "clip_ratio/high_max": 0.0024246389730251394,
+      "clip_ratio/high_mean": 0.0008917832983570406,
+      "clip_ratio/low_mean": 0.0009530246443318902,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018448079281370156,
+      "epoch": 11.774344023323614,
+      "grad_norm": 0.17317137122154236,
+      "learning_rate": 1e-06,
+      "loss": -0.0064,
+      "step": 1139
+    },
+    {
+      "clip_ratio/high_max": 0.002195367793319747,
+      "clip_ratio/high_mean": 0.0009457959349674638,
+      "clip_ratio/low_mean": 0.0007271658605532139,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016729618364479393,
+      "epoch": 11.783673469387756,
+      "grad_norm": 0.1614680141210556,
+      "learning_rate": 1e-06,
+      "loss": -0.0669,
+      "step": 1140
+    },
+    {
+      "clip_ratio/high_max": 0.0023895534322946332,
+      "clip_ratio/high_mean": 0.0009715906599012669,
+      "clip_ratio/low_mean": 0.0007251465067383833,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016967371921055019,
+      "epoch": 11.793002915451895,
+      "grad_norm": 0.13667620718479156,
+      "learning_rate": 1e-06,
+      "loss": -0.0345,
+      "step": 1141
+    },
+    {
+      "clip_ratio/high_max": 0.002259471220895648,
+      "clip_ratio/high_mean": 0.0009282710616389522,
+      "clip_ratio/low_mean": 0.0007498116001443123,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001678082684520632,
+      "epoch": 11.802332361516035,
+      "grad_norm": 0.264766126871109,
+      "learning_rate": 1e-06,
+      "loss": -0.0423,
+      "step": 1142
+    },
+    {
+      "clip_ratio/high_max": 0.0025943254731828347,
+      "clip_ratio/high_mean": 0.0009999378216889454,
+      "clip_ratio/low_mean": 0.0006869605222163955,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001686898362095235,
+      "epoch": 11.811661807580174,
+      "grad_norm": 0.15261907875537872,
+      "learning_rate": 1e-06,
+      "loss": -0.0198,
+      "step": 1143
+    },
+    {
+      "clip_ratio/high_max": 0.0025551571816322394,
+      "clip_ratio/high_mean": 0.001012136519420892,
+      "clip_ratio/low_mean": 0.0007965297954797279,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001808666311262641,
+      "epoch": 11.820991253644316,
+      "grad_norm": 0.14053599536418915,
+      "learning_rate": 1e-06,
+      "loss": -0.0309,
+      "step": 1144
+    },
+    {
+      "clip_ratio/high_max": 0.002615540954138851,
+      "clip_ratio/high_mean": 0.0012030341440549819,
+      "clip_ratio/low_mean": 0.0006149866539999493,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018180207880504895,
+      "epoch": 11.830320699708455,
+      "grad_norm": 0.12848827242851257,
+      "learning_rate": 1e-06,
+      "loss": -0.0709,
+      "step": 1145
+    },
+    {
+      "clip_ratio/high_max": 0.0027119321821373887,
+      "clip_ratio/high_mean": 0.0010350280590500915,
+      "clip_ratio/low_mean": 0.0008117324632621603,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018467605113983154,
+      "epoch": 11.839650145772595,
+      "grad_norm": 0.1603071093559265,
+      "learning_rate": 1e-06,
+      "loss": -0.0536,
+      "step": 1146
+    },
+    {
+      "clip_ratio/high_max": 0.002456844762491528,
+      "clip_ratio/high_mean": 0.001064599880919559,
+      "clip_ratio/low_mean": 0.0007531777409894858,
+      "clip_ratio/low_min": 3.7324574805097654e-05,
+      "clip_ratio/region_mean": 0.0018177776291850023,
+      "epoch": 11.848979591836734,
+      "grad_norm": 0.16494296491146088,
+      "learning_rate": 1e-06,
+      "loss": -0.0546,
+      "step": 1147
+    },
+    {
+      "clip_ratio/high_max": 0.002657020988408476,
+      "clip_ratio/high_mean": 0.0008777734765317291,
+      "clip_ratio/low_mean": 0.0008918813109630719,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001769654823874589,
+      "epoch": 11.858309037900874,
+      "grad_norm": 0.14670419692993164,
+      "learning_rate": 1e-06,
+      "loss": -0.0113,
+      "step": 1148
+    },
+    {
+      "clip_ratio/high_max": 0.002420163080387283,
+      "clip_ratio/high_mean": 0.0010264310185448267,
+      "clip_ratio/low_mean": 0.0007823877294867998,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018088187789544463,
+      "epoch": 11.867638483965015,
+      "grad_norm": 0.18839552998542786,
+      "learning_rate": 1e-06,
+      "loss": -0.0494,
+      "step": 1149
+    },
+    {
+      "clip_ratio/high_max": 0.0022179387451615185,
+      "clip_ratio/high_mean": 0.0009804271467146464,
+      "clip_ratio/low_mean": 0.0006909294452270842,
+      "clip_ratio/low_min": 4.6728971938136965e-05,
+      "clip_ratio/region_mean": 0.0016713566074031405,
+      "epoch": 11.876967930029155,
+      "grad_norm": 0.15449899435043335,
+      "learning_rate": 1e-06,
+      "loss": -0.0545,
+      "step": 1150
+    },
+    {
+      "clip_ratio/high_max": 0.002441216573060956,
+      "clip_ratio/high_mean": 0.0009098972404899541,
+      "clip_ratio/low_mean": 0.0007784011968396953,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016882984309631865,
+      "epoch": 11.886297376093294,
+      "grad_norm": 0.15630044043064117,
+      "learning_rate": 1e-06,
+      "loss": -0.0353,
+      "step": 1151
+    },
+    {
+      "clip_ratio/high_max": 0.00278202784829773,
+      "clip_ratio/high_mean": 0.0010964530520141125,
+      "clip_ratio/low_mean": 0.0007686192102482892,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018650722195161507,
+      "epoch": 11.895626822157434,
+      "grad_norm": 0.13340958952903748,
+      "learning_rate": 1e-06,
+      "loss": -0.0368,
+      "step": 1152
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0818568638392857,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4024.0,
+      "completions/mean_length": 820.6602172851562,
+      "completions/mean_terminated_length": 528.64794921875,
+      "completions/min_length": 104.0,
+      "completions/min_terminated_length": 104.0,
+      "epoch": 12.00932944606414,
+      "grad_norm": 0.17377983033657074,
+      "learning_rate": 1e-06,
+      "loss": -0.0428,
+      "num_tokens": 663036318.0,
+      "reward": 0.6762346625328064,
+      "reward_std": 0.1436760425567627,
+      "rewards/simpleverify_reward/mean": 0.6762346625328064,
+      "rewards/simpleverify_reward/std": 0.46791985630989075,
+      "step": 1153
+    },
+    {
+      "clip_ratio/high_max": 0.001661277285165852,
+      "clip_ratio/high_mean": 0.0006638219110755017,
+      "clip_ratio/low_mean": 0.0003342748977956944,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0009980968297895743,
+      "epoch": 12.018658892128279,
+      "grad_norm": 0.16826342046260834,
+      "learning_rate": 1e-06,
+      "loss": -0.0394,
+      "step": 1154
+    },
+    {
+      "clip_ratio/high_max": 0.0015568419294140767,
+      "clip_ratio/high_mean": 0.0006034252019162523,
+      "clip_ratio/low_mean": 0.00037461021292983787,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.000978035403022659,
+      "epoch": 12.02798833819242,
+      "grad_norm": 0.16304998099803925,
+      "learning_rate": 1e-06,
+      "loss": -0.0207,
+      "step": 1155
+    },
+    {
+      "clip_ratio/high_max": 0.0016580984483880457,
+      "clip_ratio/high_mean": 0.0005983781202303362,
+      "clip_ratio/low_mean": 0.000489492094857269,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010878702087211423,
+      "epoch": 12.03731778425656,
+      "grad_norm": 0.1583404690027237,
+      "learning_rate": 1e-06,
+      "loss": -0.0031,
+      "step": 1156
+    },
+    {
+      "clip_ratio/high_max": 0.002228031638878747,
+      "clip_ratio/high_mean": 0.0007559251525890431,
+      "clip_ratio/low_mean": 0.000435610113072471,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011915352697542403,
+      "epoch": 12.0466472303207,
+      "grad_norm": 0.14719834923744202,
+      "learning_rate": 1e-06,
+      "loss": -0.0341,
+      "step": 1157
+    },
+    {
+      "clip_ratio/high_max": 0.0021915329198236577,
+      "clip_ratio/high_mean": 0.0008532390220352681,
+      "clip_ratio/low_mean": 0.00038506327337017865,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012383022949506994,
+      "epoch": 12.055976676384839,
+      "grad_norm": 0.14217408001422882,
+      "learning_rate": 1e-06,
+      "loss": -0.0671,
+      "step": 1158
+    },
+    {
+      "clip_ratio/high_max": 0.0021985833955113776,
+      "clip_ratio/high_mean": 0.0007887626852607355,
+      "clip_ratio/low_mean": 0.0004897638227703283,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012785264952981379,
+      "epoch": 12.06530612244898,
+      "grad_norm": 0.15787319839000702,
+      "learning_rate": 1e-06,
+      "loss": -0.0307,
+      "step": 1159
+    },
+    {
+      "clip_ratio/high_max": 0.0020062327530467883,
+      "clip_ratio/high_mean": 0.000756065099267289,
+      "clip_ratio/low_mean": 0.0005255451924313093,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012816103153454605,
+      "epoch": 12.07463556851312,
+      "grad_norm": 0.12930822372436523,
+      "learning_rate": 1e-06,
+      "loss": -0.0063,
+      "step": 1160
+    },
+    {
+      "clip_ratio/high_max": 0.0022514497832162306,
+      "clip_ratio/high_mean": 0.0009500038177066017,
+      "clip_ratio/low_mean": 0.0006511448782475782,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016011486950446852,
+      "epoch": 12.08396501457726,
+      "grad_norm": 0.16800308227539062,
+      "learning_rate": 1e-06,
+      "loss": -0.0156,
+      "step": 1161
+    },
+    {
+      "clip_ratio/high_max": 0.002100137160596205,
+      "clip_ratio/high_mean": 0.0007769330804876518,
+      "clip_ratio/low_mean": 0.0006053432002772752,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013822762921336107,
+      "epoch": 12.093294460641399,
+      "grad_norm": 0.1570649892091751,
+      "learning_rate": 1e-06,
+      "loss": -0.0555,
+      "step": 1162
+    },
+    {
+      "clip_ratio/high_max": 0.0023757684321026318,
+      "clip_ratio/high_mean": 0.0009196359078487149,
+      "clip_ratio/low_mean": 0.0005699716239178088,
+      "clip_ratio/low_min": 1.540927041787654e-05,
+      "clip_ratio/region_mean": 0.001489607551775407,
+      "epoch": 12.102623906705539,
+      "grad_norm": 0.17271654307842255,
+      "learning_rate": 1e-06,
+      "loss": -0.0836,
+      "step": 1163
+    },
+    {
+      "clip_ratio/high_max": 0.002208869773312472,
+      "clip_ratio/high_mean": 0.0009188851108774543,
+      "clip_ratio/low_mean": 0.000640220639979816,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001559105785418069,
+      "epoch": 12.11195335276968,
+      "grad_norm": 0.1625974178314209,
+      "learning_rate": 1e-06,
+      "loss": -0.0492,
+      "step": 1164
+    },
+    {
+      "clip_ratio/high_max": 0.0024372977586608613,
+      "clip_ratio/high_mean": 0.0008482209918838635,
+      "clip_ratio/low_mean": 0.000770468699556659,
+      "clip_ratio/low_min": 5.015483475290239e-05,
+      "clip_ratio/region_mean": 0.0016186896900762804,
+      "epoch": 12.12128279883382,
+      "grad_norm": 0.17503564059734344,
+      "learning_rate": 1e-06,
+      "loss": -0.0359,
+      "step": 1165
+    },
+    {
+      "clip_ratio/high_max": 0.0022809507172496524,
+      "clip_ratio/high_mean": 0.0008475133727188222,
+      "clip_ratio/low_mean": 0.0006917627615621313,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015392761379189324,
+      "epoch": 12.130612244897959,
+      "grad_norm": 0.5186414122581482,
+      "learning_rate": 1e-06,
+      "loss": -0.0343,
+      "step": 1166
+    },
+    {
+      "clip_ratio/high_max": 0.0028250148170627654,
+      "clip_ratio/high_mean": 0.0010407908230263274,
+      "clip_ratio/low_mean": 0.0008246911302194349,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001865481972345151,
+      "epoch": 12.139941690962099,
+      "grad_norm": 0.15463808178901672,
+      "learning_rate": 1e-06,
+      "loss": -0.0562,
+      "step": 1167
+    },
+    {
+      "clip_ratio/high_max": 0.002911760821007192,
+      "clip_ratio/high_mean": 0.001097201085940469,
+      "clip_ratio/low_mean": 0.0005832727783854352,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016804738334030844,
+      "epoch": 12.14927113702624,
+      "grad_norm": 0.1505255252122879,
+      "learning_rate": 1e-06,
+      "loss": -0.0431,
+      "step": 1168
+    },
+    {
+      "clip_ratio/high_max": 0.0018844895166694187,
+      "clip_ratio/high_mean": 0.0007062927579681855,
+      "clip_ratio/low_mean": 0.0006535016027555685,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013597943834611215,
+      "epoch": 12.15860058309038,
+      "grad_norm": 0.14496773481369019,
+      "learning_rate": 1e-06,
+      "loss": -0.0017,
+      "step": 1169
+    },
+    {
+      "clip_ratio/high_max": 0.002417742849502247,
+      "clip_ratio/high_mean": 0.0009744290146045387,
+      "clip_ratio/low_mean": 0.0007717087100900244,
+      "clip_ratio/low_min": 9.492709978076164e-06,
+      "clip_ratio/region_mean": 0.0017461377283325419,
+      "epoch": 12.167930029154519,
+      "grad_norm": 0.15995532274246216,
+      "learning_rate": 1e-06,
+      "loss": -0.0366,
+      "step": 1170
+    },
+    {
+      "clip_ratio/high_max": 0.0022656575602013618,
+      "clip_ratio/high_mean": 0.0009435345400561346,
+      "clip_ratio/low_mean": 0.0006132847665867303,
+      "clip_ratio/low_min": 2.340385617571883e-05,
+      "clip_ratio/region_mean": 0.0015568192975479178,
+      "epoch": 12.177259475218658,
+      "grad_norm": 0.13755114376544952,
+      "learning_rate": 1e-06,
+      "loss": -0.0658,
+      "step": 1171
+    },
+    {
+      "clip_ratio/high_max": 0.0025760534699657,
+      "clip_ratio/high_mean": 0.001054451164236525,
+      "clip_ratio/low_mean": 0.0006196158155944431,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016740669816499576,
+      "epoch": 12.186588921282798,
+      "grad_norm": 0.1431213766336441,
+      "learning_rate": 1e-06,
+      "loss": -0.0631,
+      "step": 1172
+    },
+    {
+      "clip_ratio/high_max": 0.001973423048184486,
+      "clip_ratio/high_mean": 0.000847454575705342,
+      "clip_ratio/low_mean": 0.0008472516219626414,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016947062213148456,
+      "epoch": 12.19591836734694,
+      "grad_norm": 0.16347478330135345,
+      "learning_rate": 1e-06,
+      "loss": -0.0129,
+      "step": 1173
+    },
+    {
+      "clip_ratio/high_max": 0.002401221830950817,
+      "clip_ratio/high_mean": 0.0009602002337487647,
+      "clip_ratio/low_mean": 0.0008040178308874601,
+      "clip_ratio/low_min": 6.460615986725315e-05,
+      "clip_ratio/region_mean": 0.001764218024618458,
+      "epoch": 12.205247813411079,
+      "grad_norm": 0.15968972444534302,
+      "learning_rate": 1e-06,
+      "loss": -0.0076,
+      "step": 1174
+    },
+    {
+      "clip_ratio/high_max": 0.0023099683894542977,
+      "clip_ratio/high_mean": 0.001005184072710108,
+      "clip_ratio/low_mean": 0.000844478718136088,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001849662781751249,
+      "epoch": 12.214577259475218,
+      "grad_norm": 0.17376315593719482,
+      "learning_rate": 1e-06,
+      "loss": 0.0113,
+      "step": 1175
+    },
+    {
+      "clip_ratio/high_max": 0.0026273516996297985,
+      "clip_ratio/high_mean": 0.0010249346269119997,
+      "clip_ratio/low_mean": 0.0006908012546773534,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017157358888653107,
+      "epoch": 12.223906705539358,
+      "grad_norm": 0.1493109166622162,
+      "learning_rate": 1e-06,
+      "loss": -0.0525,
+      "step": 1176
+    },
+    {
+      "clip_ratio/high_max": 0.002960449979582336,
+      "clip_ratio/high_mean": 0.0012713336218439508,
+      "clip_ratio/low_mean": 0.0007620664873684291,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020334001383162104,
+      "epoch": 12.2332361516035,
+      "grad_norm": 0.1553051620721817,
+      "learning_rate": 1e-06,
+      "loss": -0.0428,
+      "step": 1177
+    },
+    {
+      "clip_ratio/high_max": 0.0022085066048020963,
+      "clip_ratio/high_mean": 0.0009750107274157926,
+      "clip_ratio/low_mean": 0.0006929903797754378,
+      "clip_ratio/low_min": 1.6988311472232454e-05,
+      "clip_ratio/region_mean": 0.0016680011394782923,
+      "epoch": 12.242565597667639,
+      "grad_norm": 0.46785208582878113,
+      "learning_rate": 1e-06,
+      "loss": -0.0545,
+      "step": 1178
+    },
+    {
+      "clip_ratio/high_max": 0.0026444776012795046,
+      "clip_ratio/high_mean": 0.001111933815991506,
+      "clip_ratio/low_mean": 0.0006248688005143777,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001736802645609714,
+      "epoch": 12.251895043731778,
+      "grad_norm": 0.17787323892116547,
+      "learning_rate": 1e-06,
+      "loss": -0.0546,
+      "step": 1179
+    },
+    {
+      "clip_ratio/high_max": 0.00251927028875798,
+      "clip_ratio/high_mean": 0.0011117884787381627,
+      "clip_ratio/low_mean": 0.0006955171284062089,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001807305590773467,
+      "epoch": 12.261224489795918,
+      "grad_norm": 0.13165973126888275,
+      "learning_rate": 1e-06,
+      "loss": -0.0375,
+      "step": 1180
+    },
+    {
+      "clip_ratio/high_max": 0.0024573969349148683,
+      "clip_ratio/high_mean": 0.0010340923709009076,
+      "clip_ratio/low_mean": 0.000628108978162345,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016622013572487049,
+      "epoch": 12.270553935860057,
+      "grad_norm": 0.12693646550178528,
+      "learning_rate": 1e-06,
+      "loss": -0.0748,
+      "step": 1181
+    },
+    {
+      "clip_ratio/high_max": 0.002866101589461323,
+      "clip_ratio/high_mean": 0.0011937303715967573,
+      "clip_ratio/low_mean": 0.0007009933542576618,
+      "clip_ratio/low_min": 1.3261192179925274e-05,
+      "clip_ratio/region_mean": 0.0018947236749227159,
+      "epoch": 12.279883381924199,
+      "grad_norm": 0.14077836275100708,
+      "learning_rate": 1e-06,
+      "loss": -0.0689,
+      "step": 1182
+    },
+    {
+      "clip_ratio/high_max": 0.0024632771528558806,
+      "clip_ratio/high_mean": 0.0010278113368258346,
+      "clip_ratio/low_mean": 0.0008552577764930902,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018830691114999354,
+      "epoch": 12.289212827988338,
+      "grad_norm": 0.14951001107692719,
+      "learning_rate": 1e-06,
+      "loss": -0.0035,
+      "step": 1183
+    },
+    {
+      "clip_ratio/high_max": 0.002585261157946661,
+      "clip_ratio/high_mean": 0.0009649988878663862,
+      "clip_ratio/low_mean": 0.0009446480489714304,
+      "clip_ratio/low_min": 4.653760333894752e-05,
+      "clip_ratio/region_mean": 0.0019096469259238802,
+      "epoch": 12.298542274052478,
+      "grad_norm": 0.15220119059085846,
+      "learning_rate": 1e-06,
+      "loss": 0.0197,
+      "step": 1184
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0818219866071429,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4070.0,
+      "completions/mean_length": 829.8917846679688,
+      "completions/mean_terminated_length": 538.837646484375,
+      "completions/min_length": 98.0,
+      "completions/min_terminated_length": 98.0,
+      "epoch": 12.307871720116617,
+      "grad_norm": 0.14995774626731873,
+      "learning_rate": 1e-06,
+      "loss": -0.0209,
+      "num_tokens": 680086743.0,
+      "reward": 0.665597140789032,
+      "reward_std": 0.14333374798297882,
+      "rewards/simpleverify_reward/mean": 0.6655970811843872,
+      "rewards/simpleverify_reward/std": 0.47178953886032104,
+      "step": 1185
+    },
+    {
+      "clip_ratio/high_max": 0.0015716331981820986,
+      "clip_ratio/high_mean": 0.0006381597031577257,
+      "clip_ratio/low_mean": 0.0002720080365179456,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0009101677278522402,
+      "epoch": 12.317201166180759,
+      "grad_norm": 0.2688741683959961,
+      "learning_rate": 1e-06,
+      "loss": -0.0211,
+      "step": 1186
+    },
+    {
+      "clip_ratio/high_max": 0.0018787017543218099,
+      "clip_ratio/high_mean": 0.0006574818698936724,
+      "clip_ratio/low_mean": 0.0003735257723747054,
+      "clip_ratio/low_min": 2.355823562538717e-05,
+      "clip_ratio/region_mean": 0.0010310076304449467,
+      "epoch": 12.326530612244898,
+      "grad_norm": 0.15530653297901154,
+      "learning_rate": 1e-06,
+      "loss": -0.0312,
+      "step": 1187
+    },
+    {
+      "clip_ratio/high_max": 0.0018491997216187883,
+      "clip_ratio/high_mean": 0.0006808987827753299,
+      "clip_ratio/low_mean": 0.00038289290523607633,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010637916948326165,
+      "epoch": 12.335860058309038,
+      "grad_norm": 0.16759537160396576,
+      "learning_rate": 1e-06,
+      "loss": -0.0273,
+      "step": 1188
+    },
+    {
+      "clip_ratio/high_max": 0.0017889955852297135,
+      "clip_ratio/high_mean": 0.0006620636977459071,
+      "clip_ratio/low_mean": 0.0004419795286594308,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011040432254958432,
+      "epoch": 12.345189504373177,
+      "grad_norm": 0.28052854537963867,
+      "learning_rate": 1e-06,
+      "loss": -0.0422,
+      "step": 1189
+    },
+    {
+      "clip_ratio/high_max": 0.0015503696704399772,
+      "clip_ratio/high_mean": 0.0006633316825173097,
+      "clip_ratio/low_mean": 0.0004858939437326626,
+      "clip_ratio/low_min": 2.6003745006164536e-05,
+      "clip_ratio/region_mean": 0.001149225608969573,
+      "epoch": 12.354518950437317,
+      "grad_norm": 0.12861286103725433,
+      "learning_rate": 1e-06,
+      "loss": -0.0341,
+      "step": 1190
+    },
+    {
+      "clip_ratio/high_max": 0.0019206314973416738,
+      "clip_ratio/high_mean": 0.0007159586466514156,
+      "clip_ratio/low_mean": 0.0006350382118398556,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001350996859400766,
+      "epoch": 12.363848396501458,
+      "grad_norm": 0.16097664833068848,
+      "learning_rate": 1e-06,
+      "loss": -0.0219,
+      "step": 1191
+    },
+    {
+      "clip_ratio/high_max": 0.0021182185373618267,
+      "clip_ratio/high_mean": 0.0008341489028680371,
+      "clip_ratio/low_mean": 0.0005425302297226153,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013766791016678326,
+      "epoch": 12.373177842565598,
+      "grad_norm": 0.16182248294353485,
+      "learning_rate": 1e-06,
+      "loss": -0.0193,
+      "step": 1192
+    },
+    {
+      "clip_ratio/high_max": 0.0021218025794951245,
+      "clip_ratio/high_mean": 0.0008499425330228405,
+      "clip_ratio/low_mean": 0.0005363503687476623,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001386292922688881,
+      "epoch": 12.382507288629737,
+      "grad_norm": 2.356157064437866,
+      "learning_rate": 1e-06,
+      "loss": -0.0614,
+      "step": 1193
+    },
+    {
+      "clip_ratio/high_max": 0.002036281057371525,
+      "clip_ratio/high_mean": 0.0007653305492567597,
+      "clip_ratio/low_mean": 0.0006597809488084749,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014251114953367505,
+      "epoch": 12.391836734693877,
+      "grad_norm": 0.15206362307071686,
+      "learning_rate": 1e-06,
+      "loss": -0.0191,
+      "step": 1194
+    },
+    {
+      "clip_ratio/high_max": 0.0015204007358988747,
+      "clip_ratio/high_mean": 0.0006446774850701331,
+      "clip_ratio/low_mean": 0.0005821640047543042,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012268414902791847,
+      "epoch": 12.401166180758018,
+      "grad_norm": 0.13985669612884521,
+      "learning_rate": 1e-06,
+      "loss": -0.0132,
+      "step": 1195
+    },
+    {
+      "clip_ratio/high_max": 0.0018950283629237674,
+      "clip_ratio/high_mean": 0.0007218277751235291,
+      "clip_ratio/low_mean": 0.0007843420871722628,
+      "clip_ratio/low_min": 4.07298794016242e-05,
+      "clip_ratio/region_mean": 0.0015061698541103397,
+      "epoch": 12.410495626822158,
+      "grad_norm": 111.48577880859375,
+      "learning_rate": 1e-06,
+      "loss": 0.1556,
+      "step": 1196
+    },
+    {
+      "clip_ratio/high_max": 0.0020736753212986514,
+      "clip_ratio/high_mean": 0.0008079645431280369,
+      "clip_ratio/low_mean": 0.000687326496517926,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014952910496504046,
+      "epoch": 12.419825072886297,
+      "grad_norm": 0.2254544049501419,
+      "learning_rate": 1e-06,
+      "loss": -0.026,
+      "step": 1197
+    },
+    {
+      "clip_ratio/high_max": 0.0022566876796190627,
+      "clip_ratio/high_mean": 0.0009089765117096249,
+      "clip_ratio/low_mean": 0.0008627541565147112,
+      "clip_ratio/low_min": 1.8195050870417617e-05,
+      "clip_ratio/region_mean": 0.001771730683685746,
+      "epoch": 12.429154518950437,
+      "grad_norm": 0.17400749027729034,
+      "learning_rate": 1e-06,
+      "loss": -0.0263,
+      "step": 1198
+    },
+    {
+      "clip_ratio/high_max": 0.0023348169997916557,
+      "clip_ratio/high_mean": 0.0010773108697321732,
+      "clip_ratio/low_mean": 0.0008090580035968742,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018863688965211622,
+      "epoch": 12.438483965014576,
+      "grad_norm": 0.1585964560508728,
+      "learning_rate": 1e-06,
+      "loss": -0.0759,
+      "step": 1199
+    },
+    {
+      "clip_ratio/high_max": 0.0022800183614890557,
+      "clip_ratio/high_mean": 0.0010704686665121699,
+      "clip_ratio/low_mean": 0.0008826940847939113,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019531627694959752,
+      "epoch": 12.447813411078718,
+      "grad_norm": 0.14367038011550903,
+      "learning_rate": 1e-06,
+      "loss": -0.0437,
+      "step": 1200
+    },
+    {
+      "clip_ratio/high_max": 0.002560265827924013,
+      "clip_ratio/high_mean": 0.0010434905234433245,
+      "clip_ratio/low_mean": 0.0007443507874995703,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001787841334589757,
+      "epoch": 12.457142857142857,
+      "grad_norm": 0.16841132938861847,
+      "learning_rate": 1e-06,
+      "loss": -0.0625,
+      "step": 1201
+    },
+    {
+      "clip_ratio/high_max": 0.002361693448619917,
+      "clip_ratio/high_mean": 0.0010176355517614866,
+      "clip_ratio/low_mean": 0.0008699332593096187,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018875688620028086,
+      "epoch": 12.466472303206997,
+      "grad_norm": 0.16177020967006683,
+      "learning_rate": 1e-06,
+      "loss": -0.0414,
+      "step": 1202
+    },
+    {
+      "clip_ratio/high_max": 0.002028581624472281,
+      "clip_ratio/high_mean": 0.0007894065965956543,
+      "clip_ratio/low_mean": 0.0008450373916275566,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016344439718523063,
+      "epoch": 12.475801749271136,
+      "grad_norm": 0.1632426232099533,
+      "learning_rate": 1e-06,
+      "loss": 0.0107,
+      "step": 1203
+    },
+    {
+      "clip_ratio/high_max": 0.0023649368304177187,
+      "clip_ratio/high_mean": 0.0009850279566308018,
+      "clip_ratio/low_mean": 0.0009320356111857109,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019170635860064067,
+      "epoch": 12.485131195335278,
+      "grad_norm": 0.2862667441368103,
+      "learning_rate": 1e-06,
+      "loss": -0.0131,
+      "step": 1204
+    },
+    {
+      "clip_ratio/high_max": 0.0025797787675401196,
+      "clip_ratio/high_mean": 0.0010242903772450518,
+      "clip_ratio/low_mean": 0.0009106253055506386,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019349157155374996,
+      "epoch": 12.494460641399417,
+      "grad_norm": 0.142133429646492,
+      "learning_rate": 1e-06,
+      "loss": -0.0779,
+      "step": 1205
+    },
+    {
+      "clip_ratio/high_max": 0.0021059031423646957,
+      "clip_ratio/high_mean": 0.0008474146397929871,
+      "clip_ratio/low_mean": 0.0008995021053124219,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017469167687522713,
+      "epoch": 12.503790087463557,
+      "grad_norm": 0.12770113348960876,
+      "learning_rate": 1e-06,
+      "loss": -0.0133,
+      "step": 1206
+    },
+    {
+      "clip_ratio/high_max": 0.0025423521728953347,
+      "clip_ratio/high_mean": 0.0010155218787986087,
+      "clip_ratio/low_mean": 0.0009215239351760829,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019370458103367127,
+      "epoch": 12.513119533527696,
+      "grad_norm": 0.16851669549942017,
+      "learning_rate": 1e-06,
+      "loss": -0.0235,
+      "step": 1207
+    },
+    {
+      "clip_ratio/high_max": 0.002701407494896557,
+      "clip_ratio/high_mean": 0.0010981410996464547,
+      "clip_ratio/low_mean": 0.0008389247943796363,
+      "clip_ratio/low_min": 3.055487832170911e-05,
+      "clip_ratio/region_mean": 0.0019370659210835584,
+      "epoch": 12.522448979591836,
+      "grad_norm": 0.14179156720638275,
+      "learning_rate": 1e-06,
+      "loss": -0.0639,
+      "step": 1208
+    },
+    {
+      "clip_ratio/high_max": 0.0026506004578550346,
+      "clip_ratio/high_mean": 0.0009729711891850457,
+      "clip_ratio/low_mean": 0.0008822552590572741,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018552264373283833,
+      "epoch": 12.531778425655977,
+      "grad_norm": 0.1276615858078003,
+      "learning_rate": 1e-06,
+      "loss": -0.036,
+      "step": 1209
+    },
+    {
+      "clip_ratio/high_max": 0.0028841284147347324,
+      "clip_ratio/high_mean": 0.0011803064953710418,
+      "clip_ratio/low_mean": 0.001103073386275355,
+      "clip_ratio/low_min": 0.00011602670292631956,
+      "clip_ratio/region_mean": 0.00228337987937266,
+      "epoch": 12.541107871720117,
+      "grad_norm": 0.1623101532459259,
+      "learning_rate": 1e-06,
+      "loss": -0.0556,
+      "step": 1210
+    },
+    {
+      "clip_ratio/high_max": 0.003024261233804282,
+      "clip_ratio/high_mean": 0.0010838951675395947,
+      "clip_ratio/low_mean": 0.0008472310910292435,
+      "clip_ratio/low_min": 1.6792047972558066e-05,
+      "clip_ratio/region_mean": 0.0019311262149130926,
+      "epoch": 12.550437317784256,
+      "grad_norm": 0.14724132418632507,
+      "learning_rate": 1e-06,
+      "loss": -0.0407,
+      "step": 1211
+    },
+    {
+      "clip_ratio/high_max": 0.002793861669488251,
+      "clip_ratio/high_mean": 0.001170790350442985,
+      "clip_ratio/low_mean": 0.0009176432140520774,
+      "clip_ratio/low_min": 2.0424837202881463e-05,
+      "clip_ratio/region_mean": 0.002088433553581126,
+      "epoch": 12.559766763848396,
+      "grad_norm": 0.13823272287845612,
+      "learning_rate": 1e-06,
+      "loss": -0.0507,
+      "step": 1212
+    },
+    {
+      "clip_ratio/high_max": 0.0022934300650376827,
+      "clip_ratio/high_mean": 0.0010273688276356552,
+      "clip_ratio/low_mean": 0.0008572914548494737,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018846603270503692,
+      "epoch": 12.569096209912537,
+      "grad_norm": 0.1401015967130661,
+      "learning_rate": 1e-06,
+      "loss": -0.0253,
+      "step": 1213
+    },
+    {
+      "clip_ratio/high_max": 0.002863459063519258,
+      "clip_ratio/high_mean": 0.0011268034413660644,
+      "clip_ratio/low_mean": 0.0010628795262164203,
+      "clip_ratio/low_min": 1.4070238648855593e-05,
+      "clip_ratio/region_mean": 0.002189683000324294,
+      "epoch": 12.578425655976677,
+      "grad_norm": 0.15729497373104095,
+      "learning_rate": 1e-06,
+      "loss": -0.0453,
+      "step": 1214
+    },
+    {
+      "clip_ratio/high_max": 0.0030355383714777417,
+      "clip_ratio/high_mean": 0.001130250846472336,
+      "clip_ratio/low_mean": 0.0009242892556358129,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.002054540134849958,
+      "epoch": 12.587755102040816,
+      "grad_norm": 0.17753230035305023,
+      "learning_rate": 1e-06,
+      "loss": -0.0494,
+      "step": 1215
+    },
+    {
+      "clip_ratio/high_max": 0.002656020864378661,
+      "clip_ratio/high_mean": 0.0008885436200216645,
+      "clip_ratio/low_mean": 0.0010008937297243392,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018894373206421733,
+      "epoch": 12.597084548104956,
+      "grad_norm": 0.1408652812242508,
+      "learning_rate": 1e-06,
+      "loss": -0.0311,
+      "step": 1216
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0905412946428571,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4061.0,
+      "completions/mean_length": 855.383544921875,
+      "completions/mean_terminated_length": 532.7634887695312,
+      "completions/min_length": 6.0,
+      "completions/min_terminated_length": 6.0,
+      "epoch": 12.606413994169095,
+      "grad_norm": 0.16870707273483276,
+      "learning_rate": 1e-06,
+      "loss": -0.0646,
+      "num_tokens": 696894708.0,
+      "reward": 0.6483328938484192,
+      "reward_std": 0.1489199846982956,
+      "rewards/simpleverify_reward/mean": 0.6483328938484192,
+      "rewards/simpleverify_reward/std": 0.4774990379810333,
+      "step": 1217
+    },
+    {
+      "clip_ratio/high_max": 0.0017404887694283389,
+      "clip_ratio/high_mean": 0.0006783103326597484,
+      "clip_ratio/low_mean": 0.00045006274649495026,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011283730818831827,
+      "epoch": 12.615743440233237,
+      "grad_norm": 0.16225512325763702,
+      "learning_rate": 1e-06,
+      "loss": -0.0542,
+      "step": 1218
+    },
+    {
+      "clip_ratio/high_max": 0.0022510680719278753,
+      "clip_ratio/high_mean": 0.0008329924803547328,
+      "clip_ratio/low_mean": 0.00038368742434613523,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012166798915131949,
+      "epoch": 12.625072886297376,
+      "grad_norm": 0.1448415368795395,
+      "learning_rate": 1e-06,
+      "loss": -0.0461,
+      "step": 1219
+    },
+    {
+      "clip_ratio/high_max": 0.0021501267547137104,
+      "clip_ratio/high_mean": 0.0008077281981968554,
+      "clip_ratio/low_mean": 0.0003889538829753292,
+      "clip_ratio/low_min": 1.5508685464737937e-05,
+      "clip_ratio/region_mean": 0.001196682067529764,
+      "epoch": 12.634402332361516,
+      "grad_norm": 0.15959766507148743,
+      "learning_rate": 1e-06,
+      "loss": -0.0422,
+      "step": 1220
+    },
+    {
+      "clip_ratio/high_max": 0.0018303566830581985,
+      "clip_ratio/high_mean": 0.0008155402720149141,
+      "clip_ratio/low_mean": 0.0004060740284330677,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012216143004479818,
+      "epoch": 12.643731778425655,
+      "grad_norm": 0.18018299341201782,
+      "learning_rate": 1e-06,
+      "loss": -0.06,
+      "step": 1221
+    },
+    {
+      "clip_ratio/high_max": 0.0019239308567193802,
+      "clip_ratio/high_mean": 0.0007672081537748454,
+      "clip_ratio/low_mean": 0.0005562840024140314,
+      "clip_ratio/low_min": 3.0171373509801924e-05,
+      "clip_ratio/region_mean": 0.0013234921425464563,
+      "epoch": 12.653061224489797,
+      "grad_norm": 0.16592402756214142,
+      "learning_rate": 1e-06,
+      "loss": -0.0161,
+      "step": 1222
+    },
+    {
+      "clip_ratio/high_max": 0.0021908340640948154,
+      "clip_ratio/high_mean": 0.0007934398290672107,
+      "clip_ratio/low_mean": 0.0005543048810068285,
+      "clip_ratio/low_min": 1.4657598512712866e-05,
+      "clip_ratio/region_mean": 0.0013477446991601028,
+      "epoch": 12.662390670553936,
+      "grad_norm": 0.1567879468202591,
+      "learning_rate": 1e-06,
+      "loss": -0.0149,
+      "step": 1223
+    },
+    {
+      "clip_ratio/high_max": 0.001866518781753257,
+      "clip_ratio/high_mean": 0.0007898500016381149,
+      "clip_ratio/low_mean": 0.0005112250491947634,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013010750262765214,
+      "epoch": 12.671720116618076,
+      "grad_norm": 0.13546861708164215,
+      "learning_rate": 1e-06,
+      "loss": -0.0214,
+      "step": 1224
+    },
+    {
+      "clip_ratio/high_max": 0.0021611627216771012,
+      "clip_ratio/high_mean": 0.0008454649478153442,
+      "clip_ratio/low_mean": 0.0004844476770813344,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013299125894263852,
+      "epoch": 12.681049562682215,
+      "grad_norm": 0.1423157900571823,
+      "learning_rate": 1e-06,
+      "loss": -0.0984,
+      "step": 1225
+    },
+    {
+      "clip_ratio/high_max": 0.002142385797924362,
+      "clip_ratio/high_mean": 0.0008808996535663027,
+      "clip_ratio/low_mean": 0.0005733223561037448,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014542220123985317,
+      "epoch": 12.690379008746355,
+      "grad_norm": 0.13812783360481262,
+      "learning_rate": 1e-06,
+      "loss": -0.0585,
+      "step": 1226
+    },
+    {
+      "clip_ratio/high_max": 0.002020554711634759,
+      "clip_ratio/high_mean": 0.0008732901296752971,
+      "clip_ratio/low_mean": 0.000523690925547271,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013969810424896423,
+      "epoch": 12.699708454810496,
+      "grad_norm": 0.24717958271503448,
+      "learning_rate": 1e-06,
+      "loss": -0.0644,
+      "step": 1227
+    },
+    {
+      "clip_ratio/high_max": 0.002180870473239338,
+      "clip_ratio/high_mean": 0.0008601616191299399,
+      "clip_ratio/low_mean": 0.0006040267426214996,
+      "clip_ratio/low_min": 1.4328289580589626e-05,
+      "clip_ratio/region_mean": 0.0014641883681179024,
+      "epoch": 12.709037900874636,
+      "grad_norm": 0.16998404264450073,
+      "learning_rate": 1e-06,
+      "loss": -0.0531,
+      "step": 1228
+    },
+    {
+      "clip_ratio/high_max": 0.0024453268451907206,
+      "clip_ratio/high_mean": 0.0009915249593177577,
+      "clip_ratio/low_mean": 0.0007085805464157602,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001700105465715751,
+      "epoch": 12.718367346938775,
+      "grad_norm": 0.15147055685520172,
+      "learning_rate": 1e-06,
+      "loss": -0.0584,
+      "step": 1229
+    },
+    {
+      "clip_ratio/high_max": 0.0021441570315801073,
+      "clip_ratio/high_mean": 0.0008396260145673295,
+      "clip_ratio/low_mean": 0.0006074137527321,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014470397654804401,
+      "epoch": 12.727696793002915,
+      "grad_norm": 0.6910759210586548,
+      "learning_rate": 1e-06,
+      "loss": -0.0544,
+      "step": 1230
+    },
+    {
+      "clip_ratio/high_max": 0.0024313779358635657,
+      "clip_ratio/high_mean": 0.0009315318420703989,
+      "clip_ratio/low_mean": 0.0006912571616339847,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016227889827860054,
+      "epoch": 12.737026239067056,
+      "grad_norm": 0.20457857847213745,
+      "learning_rate": 1e-06,
+      "loss": -0.0413,
+      "step": 1231
+    },
+    {
+      "clip_ratio/high_max": 0.0026995801090379246,
+      "clip_ratio/high_mean": 0.001102203874324914,
+      "clip_ratio/low_mean": 0.000710038711076777,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018122425899491645,
+      "epoch": 12.746355685131196,
+      "grad_norm": 0.1747591495513916,
+      "learning_rate": 1e-06,
+      "loss": -0.0546,
+      "step": 1232
+    },
+    {
+      "clip_ratio/high_max": 0.0021721825978602283,
+      "clip_ratio/high_mean": 0.0008254947861132678,
+      "clip_ratio/low_mean": 0.001039720304106595,
+      "clip_ratio/low_min": 6.993312854319811e-05,
+      "clip_ratio/region_mean": 0.0018652151193236932,
+      "epoch": 12.755685131195335,
+      "grad_norm": 0.14564496278762817,
+      "learning_rate": 1e-06,
+      "loss": 0.0087,
+      "step": 1233
+    },
+    {
+      "clip_ratio/high_max": 0.002144414611393586,
+      "clip_ratio/high_mean": 0.0008651878433738602,
+      "clip_ratio/low_mean": 0.0008761578019402805,
+      "clip_ratio/low_min": 3.951798134949058e-05,
+      "clip_ratio/region_mean": 0.0017413456698704977,
+      "epoch": 12.765014577259475,
+      "grad_norm": 0.15314052999019623,
+      "learning_rate": 1e-06,
+      "loss": -0.04,
+      "step": 1234
+    },
+    {
+      "clip_ratio/high_max": 0.0026401054456073325,
+      "clip_ratio/high_mean": 0.0009981954117392888,
+      "clip_ratio/low_mean": 0.0006213458800630178,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016195412681554444,
+      "epoch": 12.774344023323614,
+      "grad_norm": 0.14276200532913208,
+      "learning_rate": 1e-06,
+      "loss": -0.0479,
+      "step": 1235
+    },
+    {
+      "clip_ratio/high_max": 0.0022905794394318946,
+      "clip_ratio/high_mean": 0.001022375319735147,
+      "clip_ratio/low_mean": 0.0008215699112952279,
+      "clip_ratio/low_min": 4.134868140681647e-05,
+      "clip_ratio/region_mean": 0.001843945232394617,
+      "epoch": 12.783673469387756,
+      "grad_norm": 0.292371541261673,
+      "learning_rate": 1e-06,
+      "loss": -0.0469,
+      "step": 1236
+    },
+    {
+      "clip_ratio/high_max": 0.002155434776796028,
+      "clip_ratio/high_mean": 0.0009097324527829187,
+      "clip_ratio/low_mean": 0.000905888849956682,
+      "clip_ratio/low_min": 5.2954881539335474e-05,
+      "clip_ratio/region_mean": 0.0018156213554902934,
+      "epoch": 12.793002915451895,
+      "grad_norm": 0.1603776216506958,
+      "learning_rate": 1e-06,
+      "loss": -0.0341,
+      "step": 1237
+    },
+    {
+      "clip_ratio/high_max": 0.002130109212885145,
+      "clip_ratio/high_mean": 0.0009396676505275536,
+      "clip_ratio/low_mean": 0.000692275805704412,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001631943461688934,
+      "epoch": 12.802332361516035,
+      "grad_norm": 0.1660233736038208,
+      "learning_rate": 1e-06,
+      "loss": -0.0454,
+      "step": 1238
+    },
+    {
+      "clip_ratio/high_max": 0.002717175379075343,
+      "clip_ratio/high_mean": 0.0011015345671694377,
+      "clip_ratio/low_mean": 0.0007200122481663129,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018215468116977718,
+      "epoch": 12.811661807580174,
+      "grad_norm": 0.16909345984458923,
+      "learning_rate": 1e-06,
+      "loss": -0.0476,
+      "step": 1239
+    },
+    {
+      "clip_ratio/high_max": 0.0025858399385469966,
+      "clip_ratio/high_mean": 0.0010903625588980503,
+      "clip_ratio/low_mean": 0.0007437107390160236,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001834073267673375,
+      "epoch": 12.820991253644316,
+      "grad_norm": 0.14618924260139465,
+      "learning_rate": 1e-06,
+      "loss": -0.0283,
+      "step": 1240
+    },
+    {
+      "clip_ratio/high_max": 0.002248534372483846,
+      "clip_ratio/high_mean": 0.0008539697155356407,
+      "clip_ratio/low_mean": 0.0009583699647919275,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018123396876035258,
+      "epoch": 12.830320699708455,
+      "grad_norm": 0.12900881469249725,
+      "learning_rate": 1e-06,
+      "loss": -0.0065,
+      "step": 1241
+    },
+    {
+      "clip_ratio/high_max": 0.002524964431358967,
+      "clip_ratio/high_mean": 0.0009423884275747696,
+      "clip_ratio/low_mean": 0.0007425225703627802,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016849110324983485,
+      "epoch": 12.839650145772595,
+      "grad_norm": 0.304208368062973,
+      "learning_rate": 1e-06,
+      "loss": -0.023,
+      "step": 1242
+    },
+    {
+      "clip_ratio/high_max": 0.0023080027385731228,
+      "clip_ratio/high_mean": 0.0010065719252452254,
+      "clip_ratio/low_mean": 0.0009284118095820304,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019349837166373618,
+      "epoch": 12.848979591836734,
+      "grad_norm": 0.16248719394207,
+      "learning_rate": 1e-06,
+      "loss": -0.0317,
+      "step": 1243
+    },
+    {
+      "clip_ratio/high_max": 0.0028574519892572425,
+      "clip_ratio/high_mean": 0.001081569425878115,
+      "clip_ratio/low_mean": 0.0010608002203298383,
+      "clip_ratio/low_min": 7.573154289275408e-05,
+      "clip_ratio/region_mean": 0.0021423696525744162,
+      "epoch": 12.858309037900874,
+      "grad_norm": 0.1720161885023117,
+      "learning_rate": 1e-06,
+      "loss": 0.0147,
+      "step": 1244
+    },
+    {
+      "clip_ratio/high_max": 0.0026311200199415907,
+      "clip_ratio/high_mean": 0.0010182366913795704,
+      "clip_ratio/low_mean": 0.0008118485729937674,
+      "clip_ratio/low_min": 7.324685884668725e-05,
+      "clip_ratio/region_mean": 0.001830085224355571,
+      "epoch": 12.867638483965015,
+      "grad_norm": 0.1342935860157013,
+      "learning_rate": 1e-06,
+      "loss": -0.0344,
+      "step": 1245
+    },
+    {
+      "clip_ratio/high_max": 0.0031472905902774073,
+      "clip_ratio/high_mean": 0.00119532509779674,
+      "clip_ratio/low_mean": 0.0007574910468974849,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019528161574271508,
+      "epoch": 12.876967930029155,
+      "grad_norm": 0.1567607969045639,
+      "learning_rate": 1e-06,
+      "loss": -0.0768,
+      "step": 1246
+    },
+    {
+      "clip_ratio/high_max": 0.002146186903701164,
+      "clip_ratio/high_mean": 0.0008842511306283996,
+      "clip_ratio/low_mean": 0.0008600137771281879,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017442649295844603,
+      "epoch": 12.886297376093294,
+      "grad_norm": 0.1506277322769165,
+      "learning_rate": 1e-06,
+      "loss": -0.0228,
+      "step": 1247
+    },
+    {
+      "clip_ratio/high_max": 0.0028605922707356513,
+      "clip_ratio/high_mean": 0.0011452191774878884,
+      "clip_ratio/low_mean": 0.0008477435312670423,
+      "clip_ratio/low_min": 7.152830585255288e-05,
+      "clip_ratio/region_mean": 0.0019929627160308883,
+      "epoch": 12.895626822157434,
+      "grad_norm": 0.167267844080925,
+      "learning_rate": 1e-06,
+      "loss": -0.0492,
+      "step": 1248
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0907156808035714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4070.0,
+      "completions/mean_length": 855.1385498046875,
+      "completions/mean_terminated_length": 531.8106689453125,
+      "completions/min_length": 102.0,
+      "completions/min_terminated_length": 102.0,
+      "epoch": 13.00932944606414,
+      "grad_norm": 0.14323611557483673,
+      "learning_rate": 1e-06,
+      "loss": -0.0155,
+      "num_tokens": 713649407.0,
+      "reward": 0.6462751626968384,
+      "reward_std": 0.1527000069618225,
+      "rewards/simpleverify_reward/mean": 0.6462751030921936,
+      "rewards/simpleverify_reward/std": 0.4781334400177002,
+      "step": 1249
+    },
+    {
+      "clip_ratio/high_max": 0.0018533071197452955,
+      "clip_ratio/high_mean": 0.0006961805702303536,
+      "clip_ratio/low_mean": 0.00040510214319056104,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001101282690797234,
+      "epoch": 13.018658892128279,
+      "grad_norm": 0.31508520245552063,
+      "learning_rate": 1e-06,
+      "loss": -0.0717,
+      "step": 1250
+    },
+    {
+      "clip_ratio/high_max": 0.0017398837335349526,
+      "clip_ratio/high_mean": 0.0006584467691936879,
+      "clip_ratio/low_mean": 0.0004901854458694288,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011486322073324118,
+      "epoch": 13.02798833819242,
+      "grad_norm": 0.1800042986869812,
+      "learning_rate": 1e-06,
+      "loss": -0.0162,
+      "step": 1251
+    },
+    {
+      "clip_ratio/high_max": 0.0020725686081277672,
+      "clip_ratio/high_mean": 0.0007888763993832981,
+      "clip_ratio/low_mean": 0.0005945292377873557,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001383405615342781,
+      "epoch": 13.03731778425656,
+      "grad_norm": 0.2943728566169739,
+      "learning_rate": 1e-06,
+      "loss": -0.0257,
+      "step": 1252
+    },
+    {
+      "clip_ratio/high_max": 0.0020031657149957027,
+      "clip_ratio/high_mean": 0.0007610687353007961,
+      "clip_ratio/low_mean": 0.0005245145239314297,
+      "clip_ratio/low_min": 1.3394770576269366e-05,
+      "clip_ratio/region_mean": 0.0012855832756031305,
+      "epoch": 13.0466472303207,
+      "grad_norm": 0.16200225055217743,
+      "learning_rate": 1e-06,
+      "loss": -0.0555,
+      "step": 1253
+    },
+    {
+      "clip_ratio/high_max": 0.0020169132403680123,
+      "clip_ratio/high_mean": 0.0007873434478824493,
+      "clip_ratio/low_mean": 0.0004928506391479459,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012801940538338386,
+      "epoch": 13.055976676384839,
+      "grad_norm": 0.16475985944271088,
+      "learning_rate": 1e-06,
+      "loss": -0.0493,
+      "step": 1254
+    },
+    {
+      "clip_ratio/high_max": 0.0020985237606510054,
+      "clip_ratio/high_mean": 0.0008507140464644181,
+      "clip_ratio/low_mean": 0.0005364021710647648,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013871162445866503,
+      "epoch": 13.06530612244898,
+      "grad_norm": 0.2778988480567932,
+      "learning_rate": 1e-06,
+      "loss": -0.0684,
+      "step": 1255
+    },
+    {
+      "clip_ratio/high_max": 0.0017238276814168785,
+      "clip_ratio/high_mean": 0.0007483742410840932,
+      "clip_ratio/low_mean": 0.0005778618916565392,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013262361899251118,
+      "epoch": 13.07463556851312,
+      "grad_norm": 0.16235233843326569,
+      "learning_rate": 1e-06,
+      "loss": -0.0543,
+      "step": 1256
+    },
+    {
+      "clip_ratio/high_max": 0.002413678630546201,
+      "clip_ratio/high_mean": 0.0009781181597645627,
+      "clip_ratio/low_mean": 0.0005532661152756191,
+      "clip_ratio/low_min": 2.146291262761224e-05,
+      "clip_ratio/region_mean": 0.0015313842850446235,
+      "epoch": 13.08396501457726,
+      "grad_norm": 0.15409797430038452,
+      "learning_rate": 1e-06,
+      "loss": -0.0921,
+      "step": 1257
+    },
+    {
+      "clip_ratio/high_max": 0.0023069537346600555,
+      "clip_ratio/high_mean": 0.0008356540129170753,
+      "clip_ratio/low_mean": 0.0005737879018852254,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014094419238972478,
+      "epoch": 13.093294460641399,
+      "grad_norm": 0.18945381045341492,
+      "learning_rate": 1e-06,
+      "loss": -0.0736,
+      "step": 1258
+    },
+    {
+      "clip_ratio/high_max": 0.0022070999548304826,
+      "clip_ratio/high_mean": 0.0007774781824991805,
+      "clip_ratio/low_mean": 0.0006468332212534733,
+      "clip_ratio/low_min": 1.7970098269870505e-05,
+      "clip_ratio/region_mean": 0.001424311372829834,
+      "epoch": 13.102623906705539,
+      "grad_norm": 0.16146373748779297,
+      "learning_rate": 1e-06,
+      "loss": -0.0265,
+      "step": 1259
+    },
+    {
+      "clip_ratio/high_max": 0.002119520322594326,
+      "clip_ratio/high_mean": 0.0008539093469153158,
+      "clip_ratio/low_mean": 0.000707914303347934,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015618236211594194,
+      "epoch": 13.11195335276968,
+      "grad_norm": 0.16552145779132843,
+      "learning_rate": 1e-06,
+      "loss": -0.0341,
+      "step": 1260
+    },
+    {
+      "clip_ratio/high_max": 0.0025728354521561414,
+      "clip_ratio/high_mean": 0.0009410246748302598,
+      "clip_ratio/low_mean": 0.0005817306664539501,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001522755337646231,
+      "epoch": 13.12128279883382,
+      "grad_norm": 0.18818584084510803,
+      "learning_rate": 1e-06,
+      "loss": -0.0624,
+      "step": 1261
+    },
+    {
+      "clip_ratio/high_max": 0.002249132656288566,
+      "clip_ratio/high_mean": 0.0008876810570654925,
+      "clip_ratio/low_mean": 0.0007052272694636486,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015929083529044874,
+      "epoch": 13.130612244897959,
+      "grad_norm": 0.14595019817352295,
+      "learning_rate": 1e-06,
+      "loss": -0.0468,
+      "step": 1262
+    },
+    {
+      "clip_ratio/high_max": 0.0024061270050879102,
+      "clip_ratio/high_mean": 0.0009578252847859403,
+      "clip_ratio/low_mean": 0.0007233730557345552,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016811983077786863,
+      "epoch": 13.139941690962099,
+      "grad_norm": 0.15412220358848572,
+      "learning_rate": 1e-06,
+      "loss": -0.047,
+      "step": 1263
+    },
+    {
+      "clip_ratio/high_max": 0.0023314416321227327,
+      "clip_ratio/high_mean": 0.0008429664931099978,
+      "clip_ratio/low_mean": 0.000708587554981932,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001551553992612753,
+      "epoch": 13.14927113702624,
+      "grad_norm": 0.14815962314605713,
+      "learning_rate": 1e-06,
+      "loss": -0.0617,
+      "step": 1264
+    },
+    {
+      "clip_ratio/high_max": 0.0024939264767454006,
+      "clip_ratio/high_mean": 0.0009565121981722768,
+      "clip_ratio/low_mean": 0.0008484964637318626,
+      "clip_ratio/low_min": 1.1405109034967609e-05,
+      "clip_ratio/region_mean": 0.0018050086800940335,
+      "epoch": 13.15860058309038,
+      "grad_norm": 0.16304215788841248,
+      "learning_rate": 1e-06,
+      "loss": -0.0139,
+      "step": 1265
+    },
+    {
+      "clip_ratio/high_max": 0.0022899250325281173,
+      "clip_ratio/high_mean": 0.0009441661386517808,
+      "clip_ratio/low_mean": 0.0008178508192031586,
+      "clip_ratio/low_min": 5.0785482017090544e-05,
+      "clip_ratio/region_mean": 0.0017620169310248457,
+      "epoch": 13.167930029154519,
+      "grad_norm": 0.15857693552970886,
+      "learning_rate": 1e-06,
+      "loss": -0.0279,
+      "step": 1266
+    },
+    {
+      "clip_ratio/high_max": 0.0022827775246696547,
+      "clip_ratio/high_mean": 0.0009474246144236531,
+      "clip_ratio/low_mean": 0.0008259005080617499,
+      "clip_ratio/low_min": 4.331254240241833e-05,
+      "clip_ratio/region_mean": 0.0017733250788296573,
+      "epoch": 13.177259475218658,
+      "grad_norm": 0.16857674717903137,
+      "learning_rate": 1e-06,
+      "loss": -0.0312,
+      "step": 1267
+    },
+    {
+      "clip_ratio/high_max": 0.0020531503105303273,
+      "clip_ratio/high_mean": 0.0008404118561884388,
+      "clip_ratio/low_mean": 0.0008659362006255833,
+      "clip_ratio/low_min": 3.568370084394701e-05,
+      "clip_ratio/region_mean": 0.0017063480445358437,
+      "epoch": 13.186588921282798,
+      "grad_norm": 1.5576626062393188,
+      "learning_rate": 1e-06,
+      "loss": -0.025,
+      "step": 1268
+    },
+    {
+      "clip_ratio/high_max": 0.002561094363045413,
+      "clip_ratio/high_mean": 0.0009574164905643556,
+      "clip_ratio/low_mean": 0.0007971743943926413,
+      "clip_ratio/low_min": 1.7212889360962436e-05,
+      "clip_ratio/region_mean": 0.0017545908558531664,
+      "epoch": 13.19591836734694,
+      "grad_norm": 0.1774628758430481,
+      "learning_rate": 1e-06,
+      "loss": -0.0443,
+      "step": 1269
+    },
+    {
+      "clip_ratio/high_max": 0.0029007500161242206,
+      "clip_ratio/high_mean": 0.001128241274273023,
+      "clip_ratio/low_mean": 0.000921056325751124,
+      "clip_ratio/low_min": 4.4034757593180984e-05,
+      "clip_ratio/region_mean": 0.0020492976036621258,
+      "epoch": 13.205247813411079,
+      "grad_norm": 0.18118013441562653,
+      "learning_rate": 1e-06,
+      "loss": -0.0644,
+      "step": 1270
+    },
+    {
+      "clip_ratio/high_max": 0.0022209560265764594,
+      "clip_ratio/high_mean": 0.0009575722651788965,
+      "clip_ratio/low_mean": 0.0008682124571350869,
+      "clip_ratio/low_min": 3.587143510230817e-05,
+      "clip_ratio/region_mean": 0.0018257847041240893,
+      "epoch": 13.214577259475218,
+      "grad_norm": 0.1728212535381317,
+      "learning_rate": 1e-06,
+      "loss": -0.0295,
+      "step": 1271
+    },
+    {
+      "clip_ratio/high_max": 0.0021065881410322618,
+      "clip_ratio/high_mean": 0.0008626566923339851,
+      "clip_ratio/low_mean": 0.0010264630072924774,
+      "clip_ratio/low_min": 9.42913302424131e-05,
+      "clip_ratio/region_mean": 0.001889119710540399,
+      "epoch": 13.223906705539358,
+      "grad_norm": 0.24277324974536896,
+      "learning_rate": 1e-06,
+      "loss": -0.0231,
+      "step": 1272
+    },
+    {
+      "clip_ratio/high_max": 0.002869507086870726,
+      "clip_ratio/high_mean": 0.0011621733756328467,
+      "clip_ratio/low_mean": 0.0009256251032638829,
+      "clip_ratio/low_min": 3.807492976193316e-05,
+      "clip_ratio/region_mean": 0.002087798471620772,
+      "epoch": 13.2332361516035,
+      "grad_norm": 0.17172911763191223,
+      "learning_rate": 1e-06,
+      "loss": -0.0539,
+      "step": 1273
+    },
+    {
+      "clip_ratio/high_max": 0.002434178502880968,
+      "clip_ratio/high_mean": 0.0010049845168396132,
+      "clip_ratio/low_mean": 0.0008125728973027435,
+      "clip_ratio/low_min": 5.372974919737317e-05,
+      "clip_ratio/region_mean": 0.0018175573932239786,
+      "epoch": 13.242565597667639,
+      "grad_norm": 0.13892962038516998,
+      "learning_rate": 1e-06,
+      "loss": -0.0129,
+      "step": 1274
+    },
+    {
+      "clip_ratio/high_max": 0.0024921852455008775,
+      "clip_ratio/high_mean": 0.0009185114486172097,
+      "clip_ratio/low_mean": 0.0009172616119030863,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018357730732532218,
+      "epoch": 13.251895043731778,
+      "grad_norm": 0.1514800786972046,
+      "learning_rate": 1e-06,
+      "loss": -0.0036,
+      "step": 1275
+    },
+    {
+      "clip_ratio/high_max": 0.0025304442169726826,
+      "clip_ratio/high_mean": 0.0010810512976604514,
+      "clip_ratio/low_mean": 0.0010384250945207896,
+      "clip_ratio/low_min": 5.361258990888018e-05,
+      "clip_ratio/region_mean": 0.002119476383086294,
+      "epoch": 13.261224489795918,
+      "grad_norm": 0.14814990758895874,
+      "learning_rate": 1e-06,
+      "loss": -0.0236,
+      "step": 1276
+    },
+    {
+      "clip_ratio/high_max": 0.002682504476979375,
+      "clip_ratio/high_mean": 0.0010731675320130307,
+      "clip_ratio/low_mean": 0.001041904573867214,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0021150721222511493,
+      "epoch": 13.270553935860057,
+      "grad_norm": 0.16266551613807678,
+      "learning_rate": 1e-06,
+      "loss": -0.0313,
+      "step": 1277
+    },
+    {
+      "clip_ratio/high_max": 0.002794327348965453,
+      "clip_ratio/high_mean": 0.0010381264983152505,
+      "clip_ratio/low_mean": 0.000878216490491468,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019163429315085523,
+      "epoch": 13.279883381924199,
+      "grad_norm": 0.13999363780021667,
+      "learning_rate": 1e-06,
+      "loss": -0.0444,
+      "step": 1278
+    },
+    {
+      "clip_ratio/high_max": 0.002684035564016085,
+      "clip_ratio/high_mean": 0.0011079498035542201,
+      "clip_ratio/low_mean": 0.0008522837451891974,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019602335887611844,
+      "epoch": 13.289212827988338,
+      "grad_norm": 0.22559887170791626,
+      "learning_rate": 1e-06,
+      "loss": -0.0134,
+      "step": 1279
+    },
+    {
+      "clip_ratio/high_max": 0.003137089923257008,
+      "clip_ratio/high_mean": 0.0012736490425595548,
+      "clip_ratio/low_mean": 0.0008657584639877314,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.002139407537470106,
+      "epoch": 13.298542274052478,
+      "grad_norm": 0.15872757136821747,
+      "learning_rate": 1e-06,
+      "loss": -0.0499,
+      "step": 1280
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0867745535714286,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 3974.0,
+      "completions/mean_length": 833.7069702148438,
+      "completions/mean_terminated_length": 523.7243041992188,
+      "completions/min_length": 5.0,
+      "completions/min_terminated_length": 5.0,
+      "epoch": 13.307871720116617,
+      "grad_norm": 0.1634041666984558,
+      "learning_rate": 1e-06,
+      "loss": -0.0482,
+      "num_tokens": 730220885.0,
+      "reward": 0.6594587564468384,
+      "reward_std": 0.14032572507858276,
+      "rewards/simpleverify_reward/mean": 0.6594586968421936,
+      "rewards/simpleverify_reward/std": 0.4738995432853699,
+      "step": 1281
+    },
+    {
+      "clip_ratio/high_max": 0.001808859349694103,
+      "clip_ratio/high_mean": 0.0006272067239478929,
+      "clip_ratio/low_mean": 0.00036182535313855624,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.000989032101642806,
+      "epoch": 13.317201166180759,
+      "grad_norm": 0.1884487271308899,
+      "learning_rate": 1e-06,
+      "loss": 0.0078,
+      "step": 1282
+    },
+    {
+      "clip_ratio/high_max": 0.0020336356028565206,
+      "clip_ratio/high_mean": 0.0007408132141790702,
+      "clip_ratio/low_mean": 0.0002650878468557494,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010059010492113885,
+      "epoch": 13.326530612244898,
+      "grad_norm": 0.15217967331409454,
+      "learning_rate": 1e-06,
+      "loss": -0.0742,
+      "step": 1283
+    },
+    {
+      "clip_ratio/high_max": 0.00186717539691017,
+      "clip_ratio/high_mean": 0.000662296330119716,
+      "clip_ratio/low_mean": 0.00041883934864017647,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001081135676940903,
+      "epoch": 13.335860058309038,
+      "grad_norm": 0.18114089965820312,
+      "learning_rate": 1e-06,
+      "loss": -0.0127,
+      "step": 1284
+    },
+    {
+      "clip_ratio/high_max": 0.002345685505133588,
+      "clip_ratio/high_mean": 0.0008602707530371845,
+      "clip_ratio/low_mean": 0.00040927010786617757,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001269540865905583,
+      "epoch": 13.345189504373177,
+      "grad_norm": 0.15168233215808868,
+      "learning_rate": 1e-06,
+      "loss": -0.0395,
+      "step": 1285
+    },
+    {
+      "clip_ratio/high_max": 0.001897586254926864,
+      "clip_ratio/high_mean": 0.0006921796993992757,
+      "clip_ratio/low_mean": 0.00040923389678937383,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011014136034646071,
+      "epoch": 13.354518950437317,
+      "grad_norm": 0.15989919006824493,
+      "learning_rate": 1e-06,
+      "loss": -0.0615,
+      "step": 1286
+    },
+    {
+      "clip_ratio/high_max": 0.001944827425177209,
+      "clip_ratio/high_mean": 0.0007232873549583019,
+      "clip_ratio/low_mean": 0.00046811776928734616,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011914051265193848,
+      "epoch": 13.363848396501458,
+      "grad_norm": 0.20336195826530457,
+      "learning_rate": 1e-06,
+      "loss": -0.0298,
+      "step": 1287
+    },
+    {
+      "clip_ratio/high_max": 0.0018089843360939994,
+      "clip_ratio/high_mean": 0.0007766396356601035,
+      "clip_ratio/low_mean": 0.0004975311251200765,
+      "clip_ratio/low_min": 1.8723787434282713e-05,
+      "clip_ratio/region_mean": 0.001274170761462301,
+      "epoch": 13.373177842565598,
+      "grad_norm": 1.639616847038269,
+      "learning_rate": 1e-06,
+      "loss": -0.0315,
+      "step": 1288
+    },
+    {
+      "clip_ratio/high_max": 0.0018768889276543632,
+      "clip_ratio/high_mean": 0.0007424584982800297,
+      "clip_ratio/low_mean": 0.0005555891229960253,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012980476531083696,
+      "epoch": 13.382507288629737,
+      "grad_norm": 0.16705983877182007,
+      "learning_rate": 1e-06,
+      "loss": -0.0339,
+      "step": 1289
+    },
+    {
+      "clip_ratio/high_max": 0.0019959039309469517,
+      "clip_ratio/high_mean": 0.0007260613092512358,
+      "clip_ratio/low_mean": 0.0005967703255009837,
+      "clip_ratio/low_min": 1.3264007066027261e-05,
+      "clip_ratio/region_mean": 0.0013228316202003043,
+      "epoch": 13.391836734693877,
+      "grad_norm": 0.17171286046504974,
+      "learning_rate": 1e-06,
+      "loss": -0.0324,
+      "step": 1290
+    },
+    {
+      "clip_ratio/high_max": 0.0017491098660684656,
+      "clip_ratio/high_mean": 0.0007554136082035257,
+      "clip_ratio/low_mean": 0.0005566121276388003,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001312025746301515,
+      "epoch": 13.401166180758018,
+      "grad_norm": 0.154076486825943,
+      "learning_rate": 1e-06,
+      "loss": -0.0566,
+      "step": 1291
+    },
+    {
+      "clip_ratio/high_max": 0.002457838891132269,
+      "clip_ratio/high_mean": 0.0009031400022649905,
+      "clip_ratio/low_mean": 0.0008218799384849262,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001725019967125263,
+      "epoch": 13.410495626822158,
+      "grad_norm": 0.3098447620868683,
+      "learning_rate": 1e-06,
+      "loss": -0.0338,
+      "step": 1292
+    },
+    {
+      "clip_ratio/high_max": 0.002612185176985804,
+      "clip_ratio/high_mean": 0.0009527155234536622,
+      "clip_ratio/low_mean": 0.0007108504405550775,
+      "clip_ratio/low_min": 1.6512549336766824e-05,
+      "clip_ratio/region_mean": 0.0016635659776511602,
+      "epoch": 13.419825072886297,
+      "grad_norm": 0.15453490614891052,
+      "learning_rate": 1e-06,
+      "loss": -0.0544,
+      "step": 1293
+    },
+    {
+      "clip_ratio/high_max": 0.002358254940190818,
+      "clip_ratio/high_mean": 0.0008521508279955015,
+      "clip_ratio/low_mean": 0.0004965492491919576,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013487000614986755,
+      "epoch": 13.429154518950437,
+      "grad_norm": 0.1930791139602661,
+      "learning_rate": 1e-06,
+      "loss": -0.02,
+      "step": 1294
+    },
+    {
+      "clip_ratio/high_max": 0.0019142381934216246,
+      "clip_ratio/high_mean": 0.00085461691924138,
+      "clip_ratio/low_mean": 0.0005969883497982664,
+      "clip_ratio/low_min": 1.1363636076566763e-05,
+      "clip_ratio/region_mean": 0.0014516052324324846,
+      "epoch": 13.438483965014576,
+      "grad_norm": 0.16649490594863892,
+      "learning_rate": 1e-06,
+      "loss": -0.0216,
+      "step": 1295
+    },
+    {
+      "clip_ratio/high_max": 0.0023658930076635443,
+      "clip_ratio/high_mean": 0.0010460661578690633,
+      "clip_ratio/low_mean": 0.0006774481298634782,
+      "clip_ratio/low_min": 1.3202366062614601e-05,
+      "clip_ratio/region_mean": 0.0017235142659046687,
+      "epoch": 13.447813411078718,
+      "grad_norm": 0.20801228284835815,
+      "learning_rate": 1e-06,
+      "loss": -0.0699,
+      "step": 1296
+    },
+    {
+      "clip_ratio/high_max": 0.0022757874612580054,
+      "clip_ratio/high_mean": 0.0009783147088455735,
+      "clip_ratio/low_mean": 0.0006769907358830096,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016553054811083712,
+      "epoch": 13.457142857142857,
+      "grad_norm": 0.1578143984079361,
+      "learning_rate": 1e-06,
+      "loss": -0.0621,
+      "step": 1297
+    },
+    {
+      "clip_ratio/high_max": 0.002449881583743263,
+      "clip_ratio/high_mean": 0.0009233583696186543,
+      "clip_ratio/low_mean": 0.0007409365225612419,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001664294886722928,
+      "epoch": 13.466472303206997,
+      "grad_norm": 0.14478129148483276,
+      "learning_rate": 1e-06,
+      "loss": -0.0374,
+      "step": 1298
+    },
+    {
+      "clip_ratio/high_max": 0.0024123422044795007,
+      "clip_ratio/high_mean": 0.0010598309600027278,
+      "clip_ratio/low_mean": 0.000828198169983807,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018880291463574395,
+      "epoch": 13.475801749271136,
+      "grad_norm": 0.15566128492355347,
+      "learning_rate": 1e-06,
+      "loss": -0.0644,
+      "step": 1299
+    },
+    {
+      "clip_ratio/high_max": 0.0022581193697988056,
+      "clip_ratio/high_mean": 0.0008295270599774085,
+      "clip_ratio/low_mean": 0.0007799461964168586,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001609473256394267,
+      "epoch": 13.485131195335278,
+      "grad_norm": 0.1817512959241867,
+      "learning_rate": 1e-06,
+      "loss": -0.0132,
+      "step": 1300
+    },
+    {
+      "clip_ratio/high_max": 0.0022852260008221492,
+      "clip_ratio/high_mean": 0.0008699650170456152,
+      "clip_ratio/low_mean": 0.0007498297668462328,
+      "clip_ratio/low_min": 1.2918560969410464e-05,
+      "clip_ratio/region_mean": 0.001619794773432659,
+      "epoch": 13.494460641399417,
+      "grad_norm": 0.12788450717926025,
+      "learning_rate": 1e-06,
+      "loss": -0.0413,
+      "step": 1301
+    },
+    {
+      "clip_ratio/high_max": 0.002921771061664913,
+      "clip_ratio/high_mean": 0.0011562288818822708,
+      "clip_ratio/low_mean": 0.0007411051501549082,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018973340338561684,
+      "epoch": 13.503790087463557,
+      "grad_norm": 0.14873087406158447,
+      "learning_rate": 1e-06,
+      "loss": -0.0697,
+      "step": 1302
+    },
+    {
+      "clip_ratio/high_max": 0.0025456506955379155,
+      "clip_ratio/high_mean": 0.0009445292271266226,
+      "clip_ratio/low_mean": 0.0009287504954045289,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018732797107077204,
+      "epoch": 13.513119533527696,
+      "grad_norm": 0.15014441311359406,
+      "learning_rate": 1e-06,
+      "loss": -0.0002,
+      "step": 1303
+    },
+    {
+      "clip_ratio/high_max": 0.0029857042609364726,
+      "clip_ratio/high_mean": 0.001153793593402952,
+      "clip_ratio/low_mean": 0.0007656593706997228,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019194529886590317,
+      "epoch": 13.522448979591836,
+      "grad_norm": 0.16597695648670197,
+      "learning_rate": 1e-06,
+      "loss": -0.021,
+      "step": 1304
+    },
+    {
+      "clip_ratio/high_max": 0.0023489712039008737,
+      "clip_ratio/high_mean": 0.0009593406102794688,
+      "clip_ratio/low_mean": 0.0010226008926110808,
+      "clip_ratio/low_min": 4.99433335789945e-05,
+      "clip_ratio/region_mean": 0.00198194153199438,
+      "epoch": 13.531778425655977,
+      "grad_norm": 0.1690942943096161,
+      "learning_rate": 1e-06,
+      "loss": 0.0098,
+      "step": 1305
+    },
+    {
+      "clip_ratio/high_max": 0.0021892231088713743,
+      "clip_ratio/high_mean": 0.0008942062304413412,
+      "clip_ratio/low_mean": 0.0008410560567426728,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017352623035549186,
+      "epoch": 13.541107871720117,
+      "grad_norm": 0.15915672481060028,
+      "learning_rate": 1e-06,
+      "loss": 0.0015,
+      "step": 1306
+    },
+    {
+      "clip_ratio/high_max": 0.002514030653401278,
+      "clip_ratio/high_mean": 0.0008825074801279698,
+      "clip_ratio/low_mean": 0.0007410275356960483,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016235350412898697,
+      "epoch": 13.550437317784256,
+      "grad_norm": 0.1376136839389801,
+      "learning_rate": 1e-06,
+      "loss": -0.0317,
+      "step": 1307
+    },
+    {
+      "clip_ratio/high_max": 0.0026857103148358874,
+      "clip_ratio/high_mean": 0.0009508218008704716,
+      "clip_ratio/low_mean": 0.0008803735017863801,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018311953099328093,
+      "epoch": 13.559766763848396,
+      "grad_norm": 5.408728122711182,
+      "learning_rate": 1e-06,
+      "loss": 0.0036,
+      "step": 1308
+    },
+    {
+      "clip_ratio/high_max": 0.0023494828928960487,
+      "clip_ratio/high_mean": 0.0009465143348279526,
+      "clip_ratio/low_mean": 0.0008140747913785162,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001760589104378596,
+      "epoch": 13.569096209912537,
+      "grad_norm": 0.1753227561712265,
+      "learning_rate": 1e-06,
+      "loss": -0.0453,
+      "step": 1309
+    },
+    {
+      "clip_ratio/high_max": 0.002443296463752631,
+      "clip_ratio/high_mean": 0.0010179600867559202,
+      "clip_ratio/low_mean": 0.0009900964123517042,
+      "clip_ratio/low_min": 1.70068033185089e-05,
+      "clip_ratio/region_mean": 0.0020080564900126774,
+      "epoch": 13.578425655976677,
+      "grad_norm": 0.2076674997806549,
+      "learning_rate": 1e-06,
+      "loss": -0.0315,
+      "step": 1310
+    },
+    {
+      "clip_ratio/high_max": 0.002646667773660738,
+      "clip_ratio/high_mean": 0.0010991272793035023,
+      "clip_ratio/low_mean": 0.0007010152285147342,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001800142530555604,
+      "epoch": 13.587755102040816,
+      "grad_norm": 0.14079692959785461,
+      "learning_rate": 1e-06,
+      "loss": -0.0639,
+      "step": 1311
+    },
+    {
+      "clip_ratio/high_max": 0.002614323457237333,
+      "clip_ratio/high_mean": 0.0010439149227750022,
+      "clip_ratio/low_mean": 0.0008334641343026306,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018773790143313818,
+      "epoch": 13.597084548104956,
+      "grad_norm": 0.15647540986537933,
+      "learning_rate": 1e-06,
+      "loss": -0.0666,
+      "step": 1312
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.09716796875,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4091.0,
+      "completions/mean_length": 881.06884765625,
+      "completions/mean_terminated_length": 535.0593872070312,
+      "completions/min_length": 108.0,
+      "completions/min_terminated_length": 108.0,
+      "epoch": 13.606413994169095,
+      "grad_norm": 0.16756437718868256,
+      "learning_rate": 1e-06,
+      "loss": -0.052,
+      "num_tokens": 746936521.0,
+      "reward": 0.655831515789032,
+      "reward_std": 0.14602485299110413,
+      "rewards/simpleverify_reward/mean": 0.6558314561843872,
+      "rewards/simpleverify_reward/std": 0.4751046895980835,
+      "step": 1313
+    },
+    {
+      "clip_ratio/high_max": 0.001813500355638098,
+      "clip_ratio/high_mean": 0.0007773894612910226,
+      "clip_ratio/low_mean": 0.0003756572200472874,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011530467236298136,
+      "epoch": 13.615743440233237,
+      "grad_norm": 0.16500411927700043,
+      "learning_rate": 1e-06,
+      "loss": -0.0605,
+      "step": 1314
+    },
+    {
+      "clip_ratio/high_max": 0.0015950457909639226,
+      "clip_ratio/high_mean": 0.0005659756370732794,
+      "clip_ratio/low_mean": 0.00043099519257339125,
+      "clip_ratio/low_min": 1.5644554878235795e-05,
+      "clip_ratio/region_mean": 0.000996970833512023,
+      "epoch": 13.625072886297376,
+      "grad_norm": 0.17962244153022766,
+      "learning_rate": 1e-06,
+      "loss": 0.0066,
+      "step": 1315
+    },
+    {
+      "clip_ratio/high_max": 0.002201358074671589,
+      "clip_ratio/high_mean": 0.0008172746202035341,
+      "clip_ratio/low_mean": 0.0003837636122625554,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012010382415610366,
+      "epoch": 13.634402332361516,
+      "grad_norm": 0.14784713089466095,
+      "learning_rate": 1e-06,
+      "loss": -0.0393,
+      "step": 1316
+    },
+    {
+      "clip_ratio/high_max": 0.0017767321260180324,
+      "clip_ratio/high_mean": 0.0006811254061176442,
+      "clip_ratio/low_mean": 0.0004224134347623476,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011035388452000916,
+      "epoch": 13.643731778425655,
+      "grad_norm": 0.14827978610992432,
+      "learning_rate": 1e-06,
+      "loss": -0.025,
+      "step": 1317
+    },
+    {
+      "clip_ratio/high_max": 0.0019569822106859647,
+      "clip_ratio/high_mean": 0.0006786110207031015,
+      "clip_ratio/low_mean": 0.0004180717965027725,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010966828122036532,
+      "epoch": 13.653061224489797,
+      "grad_norm": 0.1614934653043747,
+      "learning_rate": 1e-06,
+      "loss": -0.0406,
+      "step": 1318
+    },
+    {
+      "clip_ratio/high_max": 0.0021182683885854203,
+      "clip_ratio/high_mean": 0.0008049517764447955,
+      "clip_ratio/low_mean": 0.00043997185821353924,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012449236237443984,
+      "epoch": 13.662390670553936,
+      "grad_norm": 0.18093058466911316,
+      "learning_rate": 1e-06,
+      "loss": -0.0635,
+      "step": 1319
+    },
+    {
+      "clip_ratio/high_max": 0.0018480188482499216,
+      "clip_ratio/high_mean": 0.0007299711505766027,
+      "clip_ratio/low_mean": 0.0006094280352044734,
+      "clip_ratio/low_min": 1.2781186342181172e-05,
+      "clip_ratio/region_mean": 0.0013393991757766344,
+      "epoch": 13.671720116618076,
+      "grad_norm": 0.1490190625190735,
+      "learning_rate": 1e-06,
+      "loss": -0.0172,
+      "step": 1320
+    },
+    {
+      "clip_ratio/high_max": 0.0015769346318847965,
+      "clip_ratio/high_mean": 0.0005846622361787013,
+      "clip_ratio/low_mean": 0.0005528841193154221,
+      "clip_ratio/low_min": 2.5237230147467926e-05,
+      "clip_ratio/region_mean": 0.001137546347308671,
+      "epoch": 13.681049562682215,
+      "grad_norm": 0.1591527760028839,
+      "learning_rate": 1e-06,
+      "loss": 0.0132,
+      "step": 1321
+    },
+    {
+      "clip_ratio/high_max": 0.0022073958898545243,
+      "clip_ratio/high_mean": 0.000830552562547382,
+      "clip_ratio/low_mean": 0.0006038271894794889,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014343797702167649,
+      "epoch": 13.690379008746355,
+      "grad_norm": 0.16553795337677002,
+      "learning_rate": 1e-06,
+      "loss": -0.0402,
+      "step": 1322
+    },
+    {
+      "clip_ratio/high_max": 0.002165704070648644,
+      "clip_ratio/high_mean": 0.0007843224639145774,
+      "clip_ratio/low_mean": 0.0005567349894590734,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001341057457466377,
+      "epoch": 13.699708454810496,
+      "grad_norm": 0.16234004497528076,
+      "learning_rate": 1e-06,
+      "loss": -0.071,
+      "step": 1323
+    },
+    {
+      "clip_ratio/high_max": 0.002194331907958258,
+      "clip_ratio/high_mean": 0.0009102580761464196,
+      "clip_ratio/low_mean": 0.0005638780476147076,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001474136151955463,
+      "epoch": 13.709037900874636,
+      "grad_norm": 0.1552676409482956,
+      "learning_rate": 1e-06,
+      "loss": -0.033,
+      "step": 1324
+    },
+    {
+      "clip_ratio/high_max": 0.0021396054071374238,
+      "clip_ratio/high_mean": 0.0008660626008349936,
+      "clip_ratio/low_mean": 0.0005403208729148901,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001406383489666041,
+      "epoch": 13.718367346938775,
+      "grad_norm": 0.1293756663799286,
+      "learning_rate": 1e-06,
+      "loss": -0.0791,
+      "step": 1325
+    },
+    {
+      "clip_ratio/high_max": 0.0022590512526221573,
+      "clip_ratio/high_mean": 0.0009599628792784642,
+      "clip_ratio/low_mean": 0.0005910327890887856,
+      "clip_ratio/low_min": 1.0257672329316847e-05,
+      "clip_ratio/region_mean": 0.0015509956574533135,
+      "epoch": 13.727696793002915,
+      "grad_norm": 0.183794304728508,
+      "learning_rate": 1e-06,
+      "loss": -0.0439,
+      "step": 1326
+    },
+    {
+      "clip_ratio/high_max": 0.002349718088225927,
+      "clip_ratio/high_mean": 0.000808860871984507,
+      "clip_ratio/low_mean": 0.0008203762936318526,
+      "clip_ratio/low_min": 2.071594281005673e-05,
+      "clip_ratio/region_mean": 0.001629237140150508,
+      "epoch": 13.737026239067056,
+      "grad_norm": 0.22887979447841644,
+      "learning_rate": 1e-06,
+      "loss": -0.0194,
+      "step": 1327
+    },
+    {
+      "clip_ratio/high_max": 0.0021565420902334154,
+      "clip_ratio/high_mean": 0.0008322687081090407,
+      "clip_ratio/low_mean": 0.0005570681873905414,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001389336885040393,
+      "epoch": 13.746355685131196,
+      "grad_norm": 0.14333762228488922,
+      "learning_rate": 1e-06,
+      "loss": -0.0549,
+      "step": 1328
+    },
+    {
+      "clip_ratio/high_max": 0.002397045012912713,
+      "clip_ratio/high_mean": 0.0008801318854239071,
+      "clip_ratio/low_mean": 0.000892540143468068,
+      "clip_ratio/low_min": 1.3748350284004118e-05,
+      "clip_ratio/region_mean": 0.0017726720252539963,
+      "epoch": 13.755685131195335,
+      "grad_norm": 0.15640684962272644,
+      "learning_rate": 1e-06,
+      "loss": 0.0016,
+      "step": 1329
+    },
+    {
+      "clip_ratio/high_max": 0.002037891717918683,
+      "clip_ratio/high_mean": 0.0008360920219274703,
+      "clip_ratio/low_mean": 0.0007439319815603085,
+      "clip_ratio/low_min": 3.0599756428273395e-05,
+      "clip_ratio/region_mean": 0.0015800239780219272,
+      "epoch": 13.765014577259475,
+      "grad_norm": 0.22396045923233032,
+      "learning_rate": 1e-06,
+      "loss": -0.0367,
+      "step": 1330
+    },
+    {
+      "clip_ratio/high_max": 0.00227334935698309,
+      "clip_ratio/high_mean": 0.0008723421869945014,
+      "clip_ratio/low_mean": 0.0007686912222197861,
+      "clip_ratio/low_min": 1.7189218851854093e-05,
+      "clip_ratio/region_mean": 0.0016410334392276127,
+      "epoch": 13.774344023323614,
+      "grad_norm": 0.17985431849956512,
+      "learning_rate": 1e-06,
+      "loss": 0.0089,
+      "step": 1331
+    },
+    {
+      "clip_ratio/high_max": 0.0027121722741867416,
+      "clip_ratio/high_mean": 0.001032087289786432,
+      "clip_ratio/low_mean": 0.0006574166673090076,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016895039443625137,
+      "epoch": 13.783673469387756,
+      "grad_norm": 0.1498870998620987,
+      "learning_rate": 1e-06,
+      "loss": -0.0675,
+      "step": 1332
+    },
+    {
+      "clip_ratio/high_max": 0.0025090702292800415,
+      "clip_ratio/high_mean": 0.0009700507780507905,
+      "clip_ratio/low_mean": 0.0007487067487090826,
+      "clip_ratio/low_min": 2.805206531775184e-05,
+      "clip_ratio/region_mean": 0.001718757499475032,
+      "epoch": 13.793002915451895,
+      "grad_norm": 0.14936555922031403,
+      "learning_rate": 1e-06,
+      "loss": -0.0544,
+      "step": 1333
+    },
+    {
+      "clip_ratio/high_max": 0.0024419114088232163,
+      "clip_ratio/high_mean": 0.0008547629472559493,
+      "clip_ratio/low_mean": 0.000663753350636398,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015185162701527588,
+      "epoch": 13.802332361516035,
+      "grad_norm": 0.43814796209335327,
+      "learning_rate": 1e-06,
+      "loss": -0.0105,
+      "step": 1334
+    },
+    {
+      "clip_ratio/high_max": 0.0023229664475366008,
+      "clip_ratio/high_mean": 0.000955230130784912,
+      "clip_ratio/low_mean": 0.0008407178538618609,
+      "clip_ratio/low_min": 5.260942634777166e-05,
+      "clip_ratio/region_mean": 0.0017959480537683703,
+      "epoch": 13.811661807580174,
+      "grad_norm": 0.16806772351264954,
+      "learning_rate": 1e-06,
+      "loss": -0.0677,
+      "step": 1335
+    },
+    {
+      "clip_ratio/high_max": 0.0024982161485240795,
+      "clip_ratio/high_mean": 0.000988750336546218,
+      "clip_ratio/low_mean": 0.0006580239614777383,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016467743189423345,
+      "epoch": 13.820991253644316,
+      "grad_norm": 0.1731717586517334,
+      "learning_rate": 1e-06,
+      "loss": -0.0163,
+      "step": 1336
+    },
+    {
+      "clip_ratio/high_max": 0.002182349213398993,
+      "clip_ratio/high_mean": 0.0009284431289415807,
+      "clip_ratio/low_mean": 0.0005774185074187699,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001505861640907824,
+      "epoch": 13.830320699708455,
+      "grad_norm": 0.16114822030067444,
+      "learning_rate": 1e-06,
+      "loss": -0.0542,
+      "step": 1337
+    },
+    {
+      "clip_ratio/high_max": 0.002774575310468208,
+      "clip_ratio/high_mean": 0.0009930949527188204,
+      "clip_ratio/low_mean": 0.0010404394051874988,
+      "clip_ratio/low_min": 4.161118340562098e-05,
+      "clip_ratio/region_mean": 0.002033534339716425,
+      "epoch": 13.839650145772595,
+      "grad_norm": 0.1805349588394165,
+      "learning_rate": 1e-06,
+      "loss": -0.0034,
+      "step": 1338
+    },
+    {
+      "clip_ratio/high_max": 0.0025615940176066943,
+      "clip_ratio/high_mean": 0.001067853288986953,
+      "clip_ratio/low_mean": 0.0006013203164911829,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00166917362366803,
+      "epoch": 13.848979591836734,
+      "grad_norm": 0.1403028815984726,
+      "learning_rate": 1e-06,
+      "loss": -0.0863,
+      "step": 1339
+    },
+    {
+      "clip_ratio/high_max": 0.0025039605316123925,
+      "clip_ratio/high_mean": 0.0009914779948303476,
+      "clip_ratio/low_mean": 0.0007381564028037246,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001729634383082157,
+      "epoch": 13.858309037900874,
+      "grad_norm": 0.1365215927362442,
+      "learning_rate": 1e-06,
+      "loss": -0.0316,
+      "step": 1340
+    },
+    {
+      "clip_ratio/high_max": 0.0021068060996185523,
+      "clip_ratio/high_mean": 0.0009505476118647493,
+      "clip_ratio/low_mean": 0.0007266292923304718,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016771769223851152,
+      "epoch": 13.867638483965015,
+      "grad_norm": 0.1598242223262787,
+      "learning_rate": 1e-06,
+      "loss": -0.0463,
+      "step": 1341
+    },
+    {
+      "clip_ratio/high_max": 0.002830190620443318,
+      "clip_ratio/high_mean": 0.0011102137050329475,
+      "clip_ratio/low_mean": 0.0008166018869815161,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019268155701865908,
+      "epoch": 13.876967930029155,
+      "grad_norm": 0.16801099479198456,
+      "learning_rate": 1e-06,
+      "loss": -0.0532,
+      "step": 1342
+    },
+    {
+      "clip_ratio/high_max": 0.002465410318109207,
+      "clip_ratio/high_mean": 0.0010882965070777573,
+      "clip_ratio/low_mean": 0.0008112803934636759,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001899576913274359,
+      "epoch": 13.886297376093294,
+      "grad_norm": 0.17822015285491943,
+      "learning_rate": 1e-06,
+      "loss": -0.045,
+      "step": 1343
+    },
+    {
+      "clip_ratio/high_max": 0.0022442466652137227,
+      "clip_ratio/high_mean": 0.0009439130590180866,
+      "clip_ratio/low_mean": 0.0006400478459909209,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015839609040995128,
+      "epoch": 13.895626822157434,
+      "grad_norm": 0.13033932447433472,
+      "learning_rate": 1e-06,
+      "loss": -0.0604,
+      "step": 1344
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0910993303571429,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 3996.0,
+      "completions/mean_length": 852.6739501953125,
+      "completions/mean_terminated_length": 527.5945434570312,
+      "completions/min_length": 87.0,
+      "completions/min_terminated_length": 87.0,
+      "epoch": 14.00932944606414,
+      "grad_norm": 0.1300303041934967,
+      "learning_rate": 1e-06,
+      "loss": -0.05,
+      "num_tokens": 763522171.0,
+      "reward": 0.670235812664032,
+      "reward_std": 0.1391279101371765,
+      "rewards/simpleverify_reward/mean": 0.6702357530593872,
+      "rewards/simpleverify_reward/std": 0.4701355993747711,
+      "step": 1345
+    },
+    {
+      "clip_ratio/high_max": 0.0018163523091061506,
+      "clip_ratio/high_mean": 0.0007031030163489049,
+      "clip_ratio/low_mean": 0.0002659307915564568,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0009690338156360667,
+      "epoch": 14.018658892128279,
+      "grad_norm": 0.14012062549591064,
+      "learning_rate": 1e-06,
+      "loss": -0.0438,
+      "step": 1346
+    },
+    {
+      "clip_ratio/high_max": 0.0017441005111322738,
+      "clip_ratio/high_mean": 0.0005964986430626595,
+      "clip_ratio/low_mean": 0.0004069341050580988,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001003432771540247,
+      "epoch": 14.02798833819242,
+      "grad_norm": 0.17307274043560028,
+      "learning_rate": 1e-06,
+      "loss": -0.0041,
+      "step": 1347
+    },
+    {
+      "clip_ratio/high_max": 0.001794335101294564,
+      "clip_ratio/high_mean": 0.0006799416469220887,
+      "clip_ratio/low_mean": 0.00041512002508170553,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001095061665182584,
+      "epoch": 14.03731778425656,
+      "grad_norm": 0.13101916015148163,
+      "learning_rate": 1e-06,
+      "loss": -0.0462,
+      "step": 1348
+    },
+    {
+      "clip_ratio/high_max": 0.0018801286933012307,
+      "clip_ratio/high_mean": 0.0006800919891247759,
+      "clip_ratio/low_mean": 0.00046491717193930526,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011450091806182172,
+      "epoch": 14.0466472303207,
+      "grad_norm": 0.16502954065799713,
+      "learning_rate": 1e-06,
+      "loss": -0.0388,
+      "step": 1349
+    },
+    {
+      "clip_ratio/high_max": 0.001899451766803395,
+      "clip_ratio/high_mean": 0.0006883819296490401,
+      "clip_ratio/low_mean": 0.0004995909202989424,
+      "clip_ratio/low_min": 1.700217580946628e-05,
+      "clip_ratio/region_mean": 0.0011879728626809083,
+      "epoch": 14.055976676384839,
+      "grad_norm": 0.15088720619678497,
+      "learning_rate": 1e-06,
+      "loss": 0.0022,
+      "step": 1350
+    },
+    {
+      "clip_ratio/high_max": 0.0017837504274211824,
+      "clip_ratio/high_mean": 0.0007263548359333072,
+      "clip_ratio/low_mean": 0.00040178661856771214,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00112814143722062,
+      "epoch": 14.06530612244898,
+      "grad_norm": 0.17774386703968048,
+      "learning_rate": 1e-06,
+      "loss": -0.0412,
+      "step": 1351
+    },
+    {
+      "clip_ratio/high_max": 0.0021136966097401455,
+      "clip_ratio/high_mean": 0.0007951533916639164,
+      "clip_ratio/low_mean": 0.0004988097116438439,
+      "clip_ratio/low_min": 1.525506468169624e-05,
+      "clip_ratio/region_mean": 0.0012939631051267497,
+      "epoch": 14.07463556851312,
+      "grad_norm": 0.3521977961063385,
+      "learning_rate": 1e-06,
+      "loss": -0.0395,
+      "step": 1352
+    },
+    {
+      "clip_ratio/high_max": 0.0018548462649050634,
+      "clip_ratio/high_mean": 0.0006804034910601331,
+      "clip_ratio/low_mean": 0.0005834443672938505,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012638478438020684,
+      "epoch": 14.08396501457726,
+      "grad_norm": 0.1781509667634964,
+      "learning_rate": 1e-06,
+      "loss": -0.0464,
+      "step": 1353
+    },
+    {
+      "clip_ratio/high_max": 0.0018683207890717313,
+      "clip_ratio/high_mean": 0.0006873378051750478,
+      "clip_ratio/low_mean": 0.00047870125990812085,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011660390428005485,
+      "epoch": 14.093294460641399,
+      "grad_norm": 0.1524149775505066,
+      "learning_rate": 1e-06,
+      "loss": -0.0494,
+      "step": 1354
+    },
+    {
+      "clip_ratio/high_max": 0.0018336370158067439,
+      "clip_ratio/high_mean": 0.0007193510482466081,
+      "clip_ratio/low_mean": 0.0005639771597998333,
+      "clip_ratio/low_min": 1.5284911569324322e-05,
+      "clip_ratio/region_mean": 0.0012833282198698726,
+      "epoch": 14.102623906705539,
+      "grad_norm": 0.1644119769334793,
+      "learning_rate": 1e-06,
+      "loss": -0.0085,
+      "step": 1355
+    },
+    {
+      "clip_ratio/high_max": 0.001847194303991273,
+      "clip_ratio/high_mean": 0.0007451445108017651,
+      "clip_ratio/low_mean": 0.0004804327027159161,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012255772344360594,
+      "epoch": 14.11195335276968,
+      "grad_norm": 0.1366289258003235,
+      "learning_rate": 1e-06,
+      "loss": -0.0327,
+      "step": 1356
+    },
+    {
+      "clip_ratio/high_max": 0.0023650554940104485,
+      "clip_ratio/high_mean": 0.0008509068684361409,
+      "clip_ratio/low_mean": 0.0005658402469634893,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014167471090331674,
+      "epoch": 14.12128279883382,
+      "grad_norm": 0.30652573704719543,
+      "learning_rate": 1e-06,
+      "loss": -0.0531,
+      "step": 1357
+    },
+    {
+      "clip_ratio/high_max": 0.0019843487862090115,
+      "clip_ratio/high_mean": 0.0008625020309409592,
+      "clip_ratio/low_mean": 0.000608655179348716,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001471157174819382,
+      "epoch": 14.130612244897959,
+      "grad_norm": 0.15605638921260834,
+      "learning_rate": 1e-06,
+      "loss": -0.0816,
+      "step": 1358
+    },
+    {
+      "clip_ratio/high_max": 0.0025308848707936704,
+      "clip_ratio/high_mean": 0.0009464774957450572,
+      "clip_ratio/low_mean": 0.0007858975827730319,
+      "clip_ratio/low_min": 1.7250897144549526e-05,
+      "clip_ratio/region_mean": 0.0017323750798823312,
+      "epoch": 14.139941690962099,
+      "grad_norm": 0.15378928184509277,
+      "learning_rate": 1e-06,
+      "loss": -0.0388,
+      "step": 1359
+    },
+    {
+      "clip_ratio/high_max": 0.002428557629173156,
+      "clip_ratio/high_mean": 0.000973530753981322,
+      "clip_ratio/low_mean": 0.0007160219192883233,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016895526860025711,
+      "epoch": 14.14927113702624,
+      "grad_norm": 0.15807275474071503,
+      "learning_rate": 1e-06,
+      "loss": -0.0692,
+      "step": 1360
+    },
+    {
+      "clip_ratio/high_max": 0.002050202587270178,
+      "clip_ratio/high_mean": 0.0008070806870819069,
+      "clip_ratio/low_mean": 0.0004673327671298466,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012744134473905433,
+      "epoch": 14.15860058309038,
+      "grad_norm": 0.1502304971218109,
+      "learning_rate": 1e-06,
+      "loss": -0.0773,
+      "step": 1361
+    },
+    {
+      "clip_ratio/high_max": 0.0023081238941813353,
+      "clip_ratio/high_mean": 0.0008741438014112646,
+      "clip_ratio/low_mean": 0.0006113954441389069,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014855392691970337,
+      "epoch": 14.167930029154519,
+      "grad_norm": 0.40585288405418396,
+      "learning_rate": 1e-06,
+      "loss": -0.0404,
+      "step": 1362
+    },
+    {
+      "clip_ratio/high_max": 0.0023766415106365457,
+      "clip_ratio/high_mean": 0.0009106923444051063,
+      "clip_ratio/low_mean": 0.0006825111149737495,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015932035021251068,
+      "epoch": 14.177259475218658,
+      "grad_norm": 0.15575088560581207,
+      "learning_rate": 1e-06,
+      "loss": -0.0312,
+      "step": 1363
+    },
+    {
+      "clip_ratio/high_max": 0.0023336290396400727,
+      "clip_ratio/high_mean": 0.0009512256656307727,
+      "clip_ratio/low_mean": 0.00057078834515778,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015220140521705616,
+      "epoch": 14.186588921282798,
+      "grad_norm": 0.1504908800125122,
+      "learning_rate": 1e-06,
+      "loss": -0.0739,
+      "step": 1364
+    },
+    {
+      "clip_ratio/high_max": 0.0026486451097298414,
+      "clip_ratio/high_mean": 0.0010069512300105998,
+      "clip_ratio/low_mean": 0.0005726203116864781,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015795715517015196,
+      "epoch": 14.19591836734694,
+      "grad_norm": 0.26770541071891785,
+      "learning_rate": 1e-06,
+      "loss": -0.0983,
+      "step": 1365
+    },
+    {
+      "clip_ratio/high_max": 0.002441513533995021,
+      "clip_ratio/high_mean": 0.0009189674128720071,
+      "clip_ratio/low_mean": 0.0006258882276597433,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015448556332557928,
+      "epoch": 14.205247813411079,
+      "grad_norm": 0.12824857234954834,
+      "learning_rate": 1e-06,
+      "loss": -0.0496,
+      "step": 1366
+    },
+    {
+      "clip_ratio/high_max": 0.002254344057291746,
+      "clip_ratio/high_mean": 0.0007987569788383553,
+      "clip_ratio/low_mean": 0.0008254318108811276,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016241887715295888,
+      "epoch": 14.214577259475218,
+      "grad_norm": 0.19126002490520477,
+      "learning_rate": 1e-06,
+      "loss": -0.0071,
+      "step": 1367
+    },
+    {
+      "clip_ratio/high_max": 0.0025836834611254744,
+      "clip_ratio/high_mean": 0.0010134078765986487,
+      "clip_ratio/low_mean": 0.0008518906288372818,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001865298552729655,
+      "epoch": 14.223906705539358,
+      "grad_norm": 0.14358840882778168,
+      "learning_rate": 1e-06,
+      "loss": -0.0553,
+      "step": 1368
+    },
+    {
+      "clip_ratio/high_max": 0.0023644188040634617,
+      "clip_ratio/high_mean": 0.0009145492313109571,
+      "clip_ratio/low_mean": 0.0006746485578332795,
+      "clip_ratio/low_min": 3.703703623614274e-05,
+      "clip_ratio/region_mean": 0.001589197781868279,
+      "epoch": 14.2332361516035,
+      "grad_norm": 0.1665046513080597,
+      "learning_rate": 1e-06,
+      "loss": -0.0569,
+      "step": 1369
+    },
+    {
+      "clip_ratio/high_max": 0.002611239324323833,
+      "clip_ratio/high_mean": 0.0009500922642473597,
+      "clip_ratio/low_mean": 0.0006333776145766024,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001583469900651835,
+      "epoch": 14.242565597667639,
+      "grad_norm": 0.13827988505363464,
+      "learning_rate": 1e-06,
+      "loss": -0.0529,
+      "step": 1370
+    },
+    {
+      "clip_ratio/high_max": 0.002228705616289517,
+      "clip_ratio/high_mean": 0.0008472597346553812,
+      "clip_ratio/low_mean": 0.0006761608569831878,
+      "clip_ratio/low_min": 2.270250661240425e-05,
+      "clip_ratio/region_mean": 0.001523420583907864,
+      "epoch": 14.251895043731778,
+      "grad_norm": 0.1589655876159668,
+      "learning_rate": 1e-06,
+      "loss": -0.0345,
+      "step": 1371
+    },
+    {
+      "clip_ratio/high_max": 0.0024966281562228687,
+      "clip_ratio/high_mean": 0.0010061951270472491,
+      "clip_ratio/low_mean": 0.000924932025554881,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019311271789774764,
+      "epoch": 14.261224489795918,
+      "grad_norm": 0.15875262022018433,
+      "learning_rate": 1e-06,
+      "loss": -0.0035,
+      "step": 1372
+    },
+    {
+      "clip_ratio/high_max": 0.002191486484662164,
+      "clip_ratio/high_mean": 0.0008106059503916185,
+      "clip_ratio/low_mean": 0.0007900780310592381,
+      "clip_ratio/low_min": 2.001921893679537e-05,
+      "clip_ratio/region_mean": 0.0016006839941837825,
+      "epoch": 14.270553935860057,
+      "grad_norm": 0.21720443665981293,
+      "learning_rate": 1e-06,
+      "loss": -0.0067,
+      "step": 1373
+    },
+    {
+      "clip_ratio/high_max": 0.002182364354666788,
+      "clip_ratio/high_mean": 0.0009078256771317683,
+      "clip_ratio/low_mean": 0.0007980483455867216,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017058740049833432,
+      "epoch": 14.279883381924199,
+      "grad_norm": 0.15851877629756927,
+      "learning_rate": 1e-06,
+      "loss": 0.0024,
+      "step": 1374
+    },
+    {
+      "clip_ratio/high_max": 0.0023586457391502336,
+      "clip_ratio/high_mean": 0.0008919053434510715,
+      "clip_ratio/low_mean": 0.0006079379568291188,
+      "clip_ratio/low_min": 3.053248656215146e-05,
+      "clip_ratio/region_mean": 0.0014998432670836337,
+      "epoch": 14.289212827988338,
+      "grad_norm": 0.12755772471427917,
+      "learning_rate": 1e-06,
+      "loss": -0.0466,
+      "step": 1375
+    },
+    {
+      "clip_ratio/high_max": 0.0026586229214444757,
+      "clip_ratio/high_mean": 0.0010218377410637913,
+      "clip_ratio/low_mean": 0.001001206077489769,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0020230438822181895,
+      "epoch": 14.298542274052478,
+      "grad_norm": 0.15385538339614868,
+      "learning_rate": 1e-06,
+      "loss": -0.039,
+      "step": 1376
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0970633370535714,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4004.0,
+      "completions/mean_length": 877.1220092773438,
+      "completions/mean_terminated_length": 531.1010131835938,
+      "completions/min_length": 118.0,
+      "completions/min_terminated_length": 118.0,
+      "epoch": 14.307871720116617,
+      "grad_norm": 0.1781497597694397,
+      "learning_rate": 1e-06,
+      "loss": -0.0439,
+      "num_tokens": 780128109.0,
+      "reward": 0.6571568250656128,
+      "reward_std": 0.14410173892974854,
+      "rewards/simpleverify_reward/mean": 0.6571568250656128,
+      "rewards/simpleverify_reward/std": 0.4746679365634918,
+      "step": 1377
+    },
+    {
+      "clip_ratio/high_max": 0.001786439632269321,
+      "clip_ratio/high_mean": 0.0006511402298201574,
+      "clip_ratio/low_mean": 0.000377095008843753,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010282352268404793,
+      "epoch": 14.317201166180759,
+      "grad_norm": 0.15552456676959991,
+      "learning_rate": 1e-06,
+      "loss": -0.044,
+      "step": 1378
+    },
+    {
+      "clip_ratio/high_max": 0.0018996015351149254,
+      "clip_ratio/high_mean": 0.0008098720663838321,
+      "clip_ratio/low_mean": 0.0003660948013930465,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011759668777813204,
+      "epoch": 14.326530612244898,
+      "grad_norm": 0.1618681401014328,
+      "learning_rate": 1e-06,
+      "loss": -0.0803,
+      "step": 1379
+    },
+    {
+      "clip_ratio/high_max": 0.0019648383240564726,
+      "clip_ratio/high_mean": 0.0007293696417036699,
+      "clip_ratio/low_mean": 0.00035925227757616085,
+      "clip_ratio/low_min": 1.5628907931386493e-05,
+      "clip_ratio/region_mean": 0.0010886219097301364,
+      "epoch": 14.335860058309038,
+      "grad_norm": 0.16378606855869293,
+      "learning_rate": 1e-06,
+      "loss": -0.0573,
+      "step": 1380
+    },
+    {
+      "clip_ratio/high_max": 0.0018145204485335853,
+      "clip_ratio/high_mean": 0.0006706941640004516,
+      "clip_ratio/low_mean": 0.0003518197438552306,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010225139230897184,
+      "epoch": 14.345189504373177,
+      "grad_norm": 0.13272294402122498,
+      "learning_rate": 1e-06,
+      "loss": -0.0506,
+      "step": 1381
+    },
+    {
+      "clip_ratio/high_max": 0.0017250901291845366,
+      "clip_ratio/high_mean": 0.0007156628980737878,
+      "clip_ratio/low_mean": 0.0004089378194294113,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011246007143199677,
+      "epoch": 14.354518950437317,
+      "grad_norm": 0.14052458107471466,
+      "learning_rate": 1e-06,
+      "loss": -0.0374,
+      "step": 1382
+    },
+    {
+      "clip_ratio/high_max": 0.002041730855125934,
+      "clip_ratio/high_mean": 0.0007437054646288743,
+      "clip_ratio/low_mean": 0.0005045933776273159,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0012482988240662962,
+      "epoch": 14.363848396501458,
+      "grad_norm": 0.17788098752498627,
+      "learning_rate": 1e-06,
+      "loss": 0.0069,
+      "step": 1383
+    },
+    {
+      "clip_ratio/high_max": 0.002104928469634615,
+      "clip_ratio/high_mean": 0.0007452774207195034,
+      "clip_ratio/low_mean": 0.0005182259255889221,
+      "clip_ratio/low_min": 2.5830004233284853e-05,
+      "clip_ratio/region_mean": 0.0012635033381229732,
+      "epoch": 14.373177842565598,
+      "grad_norm": 0.17931057512760162,
+      "learning_rate": 1e-06,
+      "loss": -0.0405,
+      "step": 1384
+    },
+    {
+      "clip_ratio/high_max": 0.0022468803799711168,
+      "clip_ratio/high_mean": 0.0007708484299655538,
+      "clip_ratio/low_mean": 0.00039266398516701884,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011635124210442882,
+      "epoch": 14.382507288629737,
+      "grad_norm": 0.16331049799919128,
+      "learning_rate": 1e-06,
+      "loss": -0.0562,
+      "step": 1385
+    },
+    {
+      "clip_ratio/high_max": 0.002127086067048367,
+      "clip_ratio/high_mean": 0.000811380059531075,
+      "clip_ratio/low_mean": 0.0005335072269190277,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013448872887238394,
+      "epoch": 14.391836734693877,
+      "grad_norm": 0.1949557065963745,
+      "learning_rate": 1e-06,
+      "loss": -0.0606,
+      "step": 1386
+    },
+    {
+      "clip_ratio/high_max": 0.0019874968929798342,
+      "clip_ratio/high_mean": 0.0007955819983180845,
+      "clip_ratio/low_mean": 0.0005172160235815682,
+      "clip_ratio/low_min": 1.713267556624487e-05,
+      "clip_ratio/region_mean": 0.001312798045546515,
+      "epoch": 14.401166180758018,
+      "grad_norm": 0.16476590931415558,
+      "learning_rate": 1e-06,
+      "loss": -0.0652,
+      "step": 1387
+    },
+    {
+      "clip_ratio/high_max": 0.0023038855761114974,
+      "clip_ratio/high_mean": 0.0009133996863965876,
+      "clip_ratio/low_mean": 0.0005296056394854531,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014430053379328456,
+      "epoch": 14.410495626822158,
+      "grad_norm": 0.1466214656829834,
+      "learning_rate": 1e-06,
+      "loss": -0.0357,
+      "step": 1388
+    },
+    {
+      "clip_ratio/high_max": 0.002041217085206881,
+      "clip_ratio/high_mean": 0.0008059848660195712,
+      "clip_ratio/low_mean": 0.0005780181381851435,
+      "clip_ratio/low_min": 1.3097233932057861e-05,
+      "clip_ratio/region_mean": 0.0013840030296705663,
+      "epoch": 14.419825072886297,
+      "grad_norm": 0.1603439897298813,
+      "learning_rate": 1e-06,
+      "loss": -0.029,
+      "step": 1389
+    },
+    {
+      "clip_ratio/high_max": 0.0022164858019095846,
+      "clip_ratio/high_mean": 0.0008501516831529443,
+      "clip_ratio/low_mean": 0.0006408127346730907,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014909643723513,
+      "epoch": 14.429154518950437,
+      "grad_norm": 0.16874945163726807,
+      "learning_rate": 1e-06,
+      "loss": -0.0158,
+      "step": 1390
+    },
+    {
+      "clip_ratio/high_max": 0.0021246844407869503,
+      "clip_ratio/high_mean": 0.0008615278256911552,
+      "clip_ratio/low_mean": 0.0006316156723187305,
+      "clip_ratio/low_min": 3.2260839361697435e-05,
+      "clip_ratio/region_mean": 0.0014931435143807903,
+      "epoch": 14.438483965014576,
+      "grad_norm": 0.1571662575006485,
+      "learning_rate": 1e-06,
+      "loss": -0.0419,
+      "step": 1391
+    },
+    {
+      "clip_ratio/high_max": 0.0020351516250229906,
+      "clip_ratio/high_mean": 0.0007117465365809039,
+      "clip_ratio/low_mean": 0.0007588785028929124,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014706250276503852,
+      "epoch": 14.447813411078718,
+      "grad_norm": 0.20177142322063446,
+      "learning_rate": 1e-06,
+      "loss": 0.0545,
+      "step": 1392
+    },
+    {
+      "clip_ratio/high_max": 0.0027564527263166383,
+      "clip_ratio/high_mean": 0.0010131545877811732,
+      "clip_ratio/low_mean": 0.0007468863486792543,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017600409373699222,
+      "epoch": 14.457142857142857,
+      "grad_norm": 0.20664866268634796,
+      "learning_rate": 1e-06,
+      "loss": -0.0576,
+      "step": 1393
+    },
+    {
+      "clip_ratio/high_max": 0.0023366313107544556,
+      "clip_ratio/high_mean": 0.000984400196102797,
+      "clip_ratio/low_mean": 0.00073258942393295,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017169896818813868,
+      "epoch": 14.466472303206997,
+      "grad_norm": 0.15278011560440063,
+      "learning_rate": 1e-06,
+      "loss": -0.0363,
+      "step": 1394
+    },
+    {
+      "clip_ratio/high_max": 0.0024514257893315516,
+      "clip_ratio/high_mean": 0.0009990524886234198,
+      "clip_ratio/low_mean": 0.0005893639181522303,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015884164204180706,
+      "epoch": 14.475801749271136,
+      "grad_norm": 0.15815916657447815,
+      "learning_rate": 1e-06,
+      "loss": -0.0869,
+      "step": 1395
+    },
+    {
+      "clip_ratio/high_max": 0.0022904781653778628,
+      "clip_ratio/high_mean": 0.00089939329882327,
+      "clip_ratio/low_mean": 0.0007771529744786676,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001676546293310821,
+      "epoch": 14.485131195335278,
+      "grad_norm": 0.16466347873210907,
+      "learning_rate": 1e-06,
+      "loss": -0.0243,
+      "step": 1396
+    },
+    {
+      "clip_ratio/high_max": 0.0023122845523175783,
+      "clip_ratio/high_mean": 0.0008836859469738556,
+      "clip_ratio/low_mean": 0.0006575505867658649,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015412365246447735,
+      "epoch": 14.494460641399417,
+      "grad_norm": 0.1602303832769394,
+      "learning_rate": 1e-06,
+      "loss": 0.0024,
+      "step": 1397
+    },
+    {
+      "clip_ratio/high_max": 0.0022690760961268097,
+      "clip_ratio/high_mean": 0.0008907983501558192,
+      "clip_ratio/low_mean": 0.0007349894349317765,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001625787750526797,
+      "epoch": 14.503790087463557,
+      "grad_norm": 0.17472094297409058,
+      "learning_rate": 1e-06,
+      "loss": -0.0504,
+      "step": 1398
+    },
+    {
+      "clip_ratio/high_max": 0.002355555865506176,
+      "clip_ratio/high_mean": 0.0009016182575578569,
+      "clip_ratio/low_mean": 0.0007545262560597621,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016561444717808627,
+      "epoch": 14.513119533527696,
+      "grad_norm": 0.209573432803154,
+      "learning_rate": 1e-06,
+      "loss": -0.0315,
+      "step": 1399
+    },
+    {
+      "clip_ratio/high_max": 0.002152741792087909,
+      "clip_ratio/high_mean": 0.0009140385282080388,
+      "clip_ratio/low_mean": 0.000828500468742277,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001742538977850927,
+      "epoch": 14.522448979591836,
+      "grad_norm": 0.15524740517139435,
+      "learning_rate": 1e-06,
+      "loss": -0.0312,
+      "step": 1400
+    },
+    {
+      "clip_ratio/high_max": 0.002396453226538142,
+      "clip_ratio/high_mean": 0.0009471032299188664,
+      "clip_ratio/low_mean": 0.0008003411373920244,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001747444344800897,
+      "epoch": 14.531778425655977,
+      "grad_norm": 0.15095356106758118,
+      "learning_rate": 1e-06,
+      "loss": -0.033,
+      "step": 1401
+    },
+    {
+      "clip_ratio/high_max": 0.0026893775029748213,
+      "clip_ratio/high_mean": 0.0011095104559899482,
+      "clip_ratio/low_mean": 0.000741342244509724,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018508526918594725,
+      "epoch": 14.541107871720117,
+      "grad_norm": 0.16207027435302734,
+      "learning_rate": 1e-06,
+      "loss": -0.0258,
+      "step": 1402
+    },
+    {
+      "clip_ratio/high_max": 0.002336189449124504,
+      "clip_ratio/high_mean": 0.0010394274722784758,
+      "clip_ratio/low_mean": 0.0008122254566842457,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018516529598855413,
+      "epoch": 14.550437317784256,
+      "grad_norm": 0.16433963179588318,
+      "learning_rate": 1e-06,
+      "loss": -0.1248,
+      "step": 1403
+    },
+    {
+      "clip_ratio/high_max": 0.0024447282921755686,
+      "clip_ratio/high_mean": 0.0010036249555014365,
+      "clip_ratio/low_mean": 0.0008792593635007506,
+      "clip_ratio/low_min": 3.404757444513962e-05,
+      "clip_ratio/region_mean": 0.0018828843094524927,
+      "epoch": 14.559766763848396,
+      "grad_norm": 0.7684108018875122,
+      "learning_rate": 1e-06,
+      "loss": -0.0189,
+      "step": 1404
+    },
+    {
+      "clip_ratio/high_max": 0.0021612151176668704,
+      "clip_ratio/high_mean": 0.0009288510373153258,
+      "clip_ratio/low_mean": 0.0007889671178418212,
+      "clip_ratio/low_min": 3.855644536088221e-05,
+      "clip_ratio/region_mean": 0.0017178181442432106,
+      "epoch": 14.569096209912537,
+      "grad_norm": 0.15791532397270203,
+      "learning_rate": 1e-06,
+      "loss": -0.0711,
+      "step": 1405
+    },
+    {
+      "clip_ratio/high_max": 0.0031253333363565616,
+      "clip_ratio/high_mean": 0.0011079810101364274,
+      "clip_ratio/low_mean": 0.0008127800501824822,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019207610675948672,
+      "epoch": 14.578425655976677,
+      "grad_norm": 0.14668971300125122,
+      "learning_rate": 1e-06,
+      "loss": -0.062,
+      "step": 1406
+    },
+    {
+      "clip_ratio/high_max": 0.0027687831898219883,
+      "clip_ratio/high_mean": 0.0009778712283150526,
+      "clip_ratio/low_mean": 0.0007738993808743544,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017517706437502056,
+      "epoch": 14.587755102040816,
+      "grad_norm": 0.14398270845413208,
+      "learning_rate": 1e-06,
+      "loss": -0.04,
+      "step": 1407
+    },
+    {
+      "clip_ratio/high_max": 0.0026752289486466907,
+      "clip_ratio/high_mean": 0.0009468498101341538,
+      "clip_ratio/low_mean": 0.0006932692103873705,
+      "clip_ratio/low_min": 1.0568143807176966e-05,
+      "clip_ratio/region_mean": 0.0016401190259784926,
+      "epoch": 14.597084548104956,
+      "grad_norm": 0.1568804234266281,
+      "learning_rate": 1e-06,
+      "loss": -0.0588,
+      "step": 1408
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0955636160714286,
+      "completions/max_length": 4096.0,
+      "completions/max_terminated_length": 4062.0,
+      "completions/mean_length": 861.3428344726562,
+      "completions/mean_terminated_length": 519.5657348632812,
+      "completions/min_length": 6.0,
+      "completions/min_terminated_length": 6.0,
+      "epoch": 14.606413994169095,
+      "grad_norm": 0.16777855157852173,
+      "learning_rate": 1e-06,
+      "loss": -0.0105,
+      "num_tokens": 796516440.0,
+      "reward": 0.664794921875,
+      "reward_std": 0.13781213760375977,
+      "rewards/simpleverify_reward/mean": 0.664794921875,
+      "rewards/simpleverify_reward/std": 0.4720703065395355,
+      "step": 1409
+    },
+    {
+      "clip_ratio/high_max": 0.001371023325191345,
+      "clip_ratio/high_mean": 0.0005251198062978801,
+      "clip_ratio/low_mean": 0.00037971350002408144,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0009048333049577195,
+      "epoch": 14.615743440233237,
+      "grad_norm": 0.1606350988149643,
+      "learning_rate": 1e-06,
+      "loss": -0.0141,
+      "step": 1410
+    },
+    {
+      "clip_ratio/high_max": 0.002084295505483169,
+      "clip_ratio/high_mean": 0.000694572858265019,
+      "clip_ratio/low_mean": 0.00038588593542954186,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0010804587818711298,
+      "epoch": 14.625072886297376,
+      "grad_norm": 0.15839152038097382,
+      "learning_rate": 1e-06,
+      "loss": -0.0144,
+      "step": 1411
+    },
+    {
+      "clip_ratio/high_max": 0.0017357024589728098,
+      "clip_ratio/high_mean": 0.0006295709554251516,
+      "clip_ratio/low_mean": 0.0003676174446809455,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0009971884137485176,
+      "epoch": 14.634402332361516,
+      "grad_norm": 0.15450921654701233,
+      "learning_rate": 1e-06,
+      "loss": -0.0387,
+      "step": 1412
+    },
+    {
+      "clip_ratio/high_max": 0.0017522241450933507,
+      "clip_ratio/high_mean": 0.0007708320003985136,
+      "clip_ratio/low_mean": 0.00040037452549768204,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.00117120652430458,
+      "epoch": 14.643731778425655,
+      "grad_norm": 0.15621048212051392,
+      "learning_rate": 1e-06,
+      "loss": -0.0716,
+      "step": 1413
+    },
+    {
+      "clip_ratio/high_max": 0.0018941683301818557,
+      "clip_ratio/high_mean": 0.0006009948883729521,
+      "clip_ratio/low_mean": 0.00038278234660538146,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.000983777252258733,
+      "epoch": 14.653061224489797,
+      "grad_norm": 0.7600335478782654,
+      "learning_rate": 1e-06,
+      "loss": -0.0209,
+      "step": 1414
+    },
+    {
+      "clip_ratio/high_max": 0.001902294548926875,
+      "clip_ratio/high_mean": 0.0007294602355614188,
+      "clip_ratio/low_mean": 0.0004647338855647831,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0011941941284021595,
+      "epoch": 14.662390670553936,
+      "grad_norm": 528.1058959960938,
+      "learning_rate": 1e-06,
+      "loss": 0.0261,
+      "step": 1415
+    },
+    {
+      "clip_ratio/high_max": 0.0018748468319245148,
+      "clip_ratio/high_mean": 0.0006354565593937878,
+      "clip_ratio/low_mean": 0.0003740206661859702,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001009477233310463,
+      "epoch": 14.671720116618076,
+      "grad_norm": 0.216713085770607,
+      "learning_rate": 1e-06,
+      "loss": -0.0705,
+      "step": 1416
+    },
+    {
+      "clip_ratio/high_max": 0.00203963217427372,
+      "clip_ratio/high_mean": 0.0007368754249910126,
+      "clip_ratio/low_mean": 0.0007297716183529701,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001466647026973078,
+      "epoch": 14.681049562682215,
+      "grad_norm": 0.16934724152088165,
+      "learning_rate": 1e-06,
+      "loss": -0.0254,
+      "step": 1417
+    },
+    {
+      "clip_ratio/high_max": 0.0020112539386900607,
+      "clip_ratio/high_mean": 0.0008417380413447972,
+      "clip_ratio/low_mean": 0.000479844017945652,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001321582036325708,
+      "epoch": 14.690379008746355,
+      "grad_norm": 0.17261309921741486,
+      "learning_rate": 1e-06,
+      "loss": -0.0922,
+      "step": 1418
+    },
+    {
+      "clip_ratio/high_max": 0.0021192739659454674,
+      "clip_ratio/high_mean": 0.0009254170618078206,
+      "clip_ratio/low_mean": 0.0005467834498631419,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014722004816576373,
+      "epoch": 14.699708454810496,
+      "grad_norm": 0.14940941333770752,
+      "learning_rate": 1e-06,
+      "loss": -0.0979,
+      "step": 1419
+    },
+    {
+      "clip_ratio/high_max": 0.0021527031349251047,
+      "clip_ratio/high_mean": 0.0008098544603853952,
+      "clip_ratio/low_mean": 0.0005650116188462562,
+      "clip_ratio/low_min": 9.181724635709543e-06,
+      "clip_ratio/region_mean": 0.001374866085825488,
+      "epoch": 14.709037900874636,
+      "grad_norm": 0.16518554091453552,
+      "learning_rate": 1e-06,
+      "loss": -0.0605,
+      "step": 1420
+    },
+    {
+      "clip_ratio/high_max": 0.0026311096371500753,
+      "clip_ratio/high_mean": 0.000915269018150866,
+      "clip_ratio/low_mean": 0.0005642396790790372,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0014795086863159668,
+      "epoch": 14.718367346938775,
+      "grad_norm": 0.15047743916511536,
+      "learning_rate": 1e-06,
+      "loss": -0.0613,
+      "step": 1421
+    },
+    {
+      "clip_ratio/high_max": 0.0022607879727729596,
+      "clip_ratio/high_mean": 0.0008101974126475397,
+      "clip_ratio/low_mean": 0.0005676412843058642,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013778386928606778,
+      "epoch": 14.727696793002915,
+      "grad_norm": 0.14523451030254364,
+      "learning_rate": 1e-06,
+      "loss": -0.0506,
+      "step": 1422
+    },
+    {
+      "clip_ratio/high_max": 0.0022824135594419204,
+      "clip_ratio/high_mean": 0.0009598317810741719,
+      "clip_ratio/low_mean": 0.00062423257077171,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015840643463889137,
+      "epoch": 14.737026239067056,
+      "grad_norm": 0.17060302197933197,
+      "learning_rate": 1e-06,
+      "loss": -0.0328,
+      "step": 1423
+    },
+    {
+      "clip_ratio/high_max": 0.0022096964603406377,
+      "clip_ratio/high_mean": 0.000983294781690347,
+      "clip_ratio/low_mean": 0.000659456734865671,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016427514929091558,
+      "epoch": 14.746355685131196,
+      "grad_norm": 0.17045511305332184,
+      "learning_rate": 1e-06,
+      "loss": -0.069,
+      "step": 1424
+    },
+    {
+      "clip_ratio/high_max": 0.002079459918604698,
+      "clip_ratio/high_mean": 0.000781764821113029,
+      "clip_ratio/low_mean": 0.0006046573339517636,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0013864221546100453,
+      "epoch": 14.755685131195335,
+      "grad_norm": 0.14913129806518555,
+      "learning_rate": 1e-06,
+      "loss": -0.0277,
+      "step": 1425
+    },
+    {
+      "clip_ratio/high_max": 0.0026301128673367202,
+      "clip_ratio/high_mean": 0.0009390130126121221,
+      "clip_ratio/low_mean": 0.0008366364982066443,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017756495144567452,
+      "epoch": 14.765014577259475,
+      "grad_norm": 0.17709794640541077,
+      "learning_rate": 1e-06,
+      "loss": -0.0301,
+      "step": 1426
+    },
+    {
+      "clip_ratio/high_max": 0.002419726486550644,
+      "clip_ratio/high_mean": 0.001029123322950909,
+      "clip_ratio/low_mean": 0.0007320142758544534,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001761137587891426,
+      "epoch": 14.774344023323614,
+      "grad_norm": 101.44778442382812,
+      "learning_rate": 1e-06,
+      "loss": -0.0231,
+      "step": 1427
+    },
+    {
+      "clip_ratio/high_max": 0.0025511273415759206,
+      "clip_ratio/high_mean": 0.0009499411698925542,
+      "clip_ratio/low_mean": 0.0006135852263469133,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015635264207958244,
+      "epoch": 14.783673469387756,
+      "grad_norm": 0.16930541396141052,
+      "learning_rate": 1e-06,
+      "loss": -0.0656,
+      "step": 1428
+    },
+    {
+      "clip_ratio/high_max": 0.0028346305116428994,
+      "clip_ratio/high_mean": 0.000991210403299192,
+      "clip_ratio/low_mean": 0.0007385825665551238,
+      "clip_ratio/low_min": 1.4789398846914992e-05,
+      "clip_ratio/region_mean": 0.001729792966216337,
+      "epoch": 14.793002915451895,
+      "grad_norm": 0.18170827627182007,
+      "learning_rate": 1e-06,
+      "loss": -0.0524,
+      "step": 1429
+    },
+    {
+      "clip_ratio/high_max": 0.0024006419407669455,
+      "clip_ratio/high_mean": 0.000863458439198439,
+      "clip_ratio/low_mean": 0.0007311092631425709,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0015945676932460628,
+      "epoch": 14.802332361516035,
+      "grad_norm": 0.4562612473964691,
+      "learning_rate": 1e-06,
+      "loss": -0.0094,
+      "step": 1430
+    },
+    {
+      "clip_ratio/high_max": 0.002215027474449016,
+      "clip_ratio/high_mean": 0.0008813750164335943,
+      "clip_ratio/low_mean": 0.0008591822560219953,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017405572580173612,
+      "epoch": 14.811661807580174,
+      "grad_norm": 0.2189190536737442,
+      "learning_rate": 1e-06,
+      "loss": -0.0213,
+      "step": 1431
+    },
+    {
+      "clip_ratio/high_max": 0.002640083766891621,
+      "clip_ratio/high_mean": 0.000979864773398731,
+      "clip_ratio/low_mean": 0.0006226883015187923,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016025530530896503,
+      "epoch": 14.820991253644316,
+      "grad_norm": 0.13267762959003448,
+      "learning_rate": 1e-06,
+      "loss": -0.0534,
+      "step": 1432
+    },
+    {
+      "clip_ratio/high_max": 0.002463354256178718,
+      "clip_ratio/high_mean": 0.0009932165776262991,
+      "clip_ratio/low_mean": 0.0009031323679664638,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0018963489783345722,
+      "epoch": 14.830320699708455,
+      "grad_norm": 2.523435115814209,
+      "learning_rate": 1e-06,
+      "loss": -0.0491,
+      "step": 1433
+    },
+    {
+      "clip_ratio/high_max": 0.0026850850190385245,
+      "clip_ratio/high_mean": 0.0009352686010970501,
+      "clip_ratio/low_mean": 0.0007063232415021048,
+      "clip_ratio/low_min": 2.6829791750060394e-05,
+      "clip_ratio/region_mean": 0.0016415918653365225,
+      "epoch": 14.839650145772595,
+      "grad_norm": 0.3484272062778473,
+      "learning_rate": 1e-06,
+      "loss": -0.0161,
+      "step": 1434
+    },
+    {
+      "clip_ratio/high_max": 0.0025000876557896845,
+      "clip_ratio/high_mean": 0.0010419811460451456,
+      "clip_ratio/low_mean": 0.0006713745046909025,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0017133556902990676,
+      "epoch": 14.848979591836734,
+      "grad_norm": 0.1631457507610321,
+      "learning_rate": 1e-06,
+      "loss": -0.0532,
+      "step": 1435
+    },
+    {
+      "clip_ratio/high_max": 0.0023943624473758973,
+      "clip_ratio/high_mean": 0.0009258485661121085,
+      "clip_ratio/low_mean": 0.0009387403497385094,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.001864588906755671,
+      "epoch": 14.858309037900874,
+      "grad_norm": 0.1668511927127838,
+      "learning_rate": 1e-06,
+      "loss": -0.0054,
+      "step": 1436
+    },
+    {
+      "clip_ratio/high_max": 0.002261575617012568,
+      "clip_ratio/high_mean": 0.0009211516025970923,
+      "clip_ratio/low_mean": 0.0007193780056695687,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0016405296300945338,
+      "epoch": 14.867638483965015,
+      "grad_norm": 0.14407925307750702,
+      "learning_rate": 1e-06,
+      "loss": -0.0579,
+      "step": 1437
+    },
+    {
+      "clip_ratio/high_max": 0.0024793751508696005,
+      "clip_ratio/high_mean": 0.0011043389058613684,
+      "clip_ratio/low_mean": 0.0008443706365142134,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019487095560180023,
+      "epoch": 14.876967930029155,
+      "grad_norm": 0.1657928228378296,
+      "learning_rate": 1e-06,
+      "loss": -0.0696,
+      "step": 1438
+    },
+    {
+      "clip_ratio/high_max": 0.0021965775558783207,
+      "clip_ratio/high_mean": 0.0009189502807203098,
+      "clip_ratio/low_mean": 0.001032198974826315,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0019511492391757201,
+      "epoch": 14.886297376093294,
+      "grad_norm": 0.20805539190769196,
+      "learning_rate": 1e-06,
+      "loss": -0.0308,
+      "step": 1439
+    },
+    {
+      "clip_ratio/high_max": 0.0024939083596109413,
+      "clip_ratio/high_mean": 0.0010812456621351885,
+      "clip_ratio/low_mean": 0.0010336533523513936,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0021148989617358893,
+      "epoch": 14.895626822157434,
+      "grad_norm": 0.1499452441930771,
+      "learning_rate": 1e-06,
+      "loss": -0.0266,
+      "step": 1440
+    },
+    {
+      "epoch": 14.895626822157434,
+      "step": 1440,
+      "total_flos": 0.0,
+      "train_loss": -0.017012172211714947,
+      "train_runtime": 87883.361,
+      "train_samples_per_second": 16.313,
+      "train_steps_per_second": 0.018
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1600,
+  "num_input_tokens_seen": 796516440,
+  "num_train_epochs": 15,
+  "save_steps": 160,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}