det-noun-lstm-2 / checkpoint-152640 /trainer_state.json
abhinavp's picture
Training in progress, epoch 1, checkpoint
48bb15b verified
raw
history blame
37.5 kB
{
"best_metric": 4.469255447387695,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/det-noun/lstm/2/checkpoints/checkpoint-152640",
"epoch": 1.0250006060157382,
"eval_steps": 10,
"global_step": 152640,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.999998362119627e-05,
"loss": 10.8216,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.999161405248948e-05,
"loss": 7.5604,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0668,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.999,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.9512,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.8869,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.7317,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994129836742636e-05,
"loss": 6.6248,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993291241991584e-05,
"loss": 6.5301,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992452647240532e-05,
"loss": 6.4596,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.99161405248948e-05,
"loss": 6.4014,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990775457738428e-05,
"loss": 6.3423,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989936862987376e-05,
"loss": 6.2755,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989098268236324e-05,
"loss": 6.2058,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988259673485272e-05,
"loss": 6.1542,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.98742107873422e-05,
"loss": 6.0875,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986582483983168e-05,
"loss": 6.0564,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985743889232116e-05,
"loss": 5.9992,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984905294481064e-05,
"loss": 5.9548,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984066699730012e-05,
"loss": 5.921,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.8852,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.9823927859886547e-05,
"loss": 5.8459,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.9815541912376026e-05,
"loss": 5.8177,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.9807155964865506e-05,
"loss": 5.7829,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.9798770017354986e-05,
"loss": 5.7633,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.9790384069844466e-05,
"loss": 5.7281,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.9782014501137675e-05,
"loss": 5.7072,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773628553627155e-05,
"loss": 5.676,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.9765242606116635e-05,
"loss": 5.6523,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.9756856658606115e-05,
"loss": 5.6343,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.9748470711095595e-05,
"loss": 5.6143,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.9740101142388804e-05,
"loss": 5.5968,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.9731715194878284e-05,
"loss": 5.5768,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.9723329247367764e-05,
"loss": 5.5583,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.9714943299857244e-05,
"loss": 5.5427,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.970657373115046e-05,
"loss": 5.5264,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.969818778363994e-05,
"loss": 5.5112,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.968980183612942e-05,
"loss": 5.4832,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.96814158886189e-05,
"loss": 5.4788,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.967304631991211e-05,
"loss": 5.439,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.966466037240159e-05,
"loss": 5.4393,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.965627442489107e-05,
"loss": 5.4291,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964788847738054e-05,
"loss": 5.4257,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963951890867376e-05,
"loss": 5.3887,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.963113296116324e-05,
"loss": 5.3912,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962274701365272e-05,
"loss": 5.3806,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.96143610661422e-05,
"loss": 5.3718,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.960597511863168e-05,
"loss": 5.3694,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.959758917112116e-05,
"loss": 5.3343,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.958920322361064e-05,
"loss": 5.3292,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958081727610012e-05,
"loss": 5.3335,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.957244770739333e-05,
"loss": 5.3224,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956406175988281e-05,
"loss": 5.3033,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.955567581237229e-05,
"loss": 5.2834,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.954728986486177e-05,
"loss": 5.2885,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.9538920296154976e-05,
"loss": 5.2691,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530534348644456e-05,
"loss": 5.2891,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.9522148401133936e-05,
"loss": 5.2383,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.951376245362342e-05,
"loss": 5.2493,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.950539288491663e-05,
"loss": 5.2374,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.949700693740611e-05,
"loss": 5.2352,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948862098989559e-05,
"loss": 5.2312,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.948023504238507e-05,
"loss": 5.2077,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947184909487455e-05,
"loss": 5.1907,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.946347952616776e-05,
"loss": 5.1964,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945509357865724e-05,
"loss": 5.2059,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944670763114672e-05,
"loss": 5.1761,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.94383216836362e-05,
"loss": 5.1756,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942995211492941e-05,
"loss": 5.1513,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942156616741889e-05,
"loss": 5.1521,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.9413196598712105e-05,
"loss": 5.1627,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.9404810651201585e-05,
"loss": 5.1606,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.9396424703691065e-05,
"loss": 5.1345,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.9388055134984274e-05,
"loss": 5.147,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.9379669187473754e-05,
"loss": 5.1447,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.9371283239963234e-05,
"loss": 5.134,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362897292452714e-05,
"loss": 5.1029,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.9354511344942194e-05,
"loss": 5.1099,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.9346125397431674e-05,
"loss": 5.0994,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.933775582872488e-05,
"loss": 5.0918,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.932936988121436e-05,
"loss": 5.0906,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.932098393370384e-05,
"loss": 5.0855,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.931259798619332e-05,
"loss": 5.087,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.930421203868281e-05,
"loss": 5.0747,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.929582609117229e-05,
"loss": 5.0577,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.928744014366177e-05,
"loss": 5.0579,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.927905419615125e-05,
"loss": 5.0578,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.927066824864073e-05,
"loss": 5.0482,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.92622823011302e-05,
"loss": 5.0441,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.925391273242342e-05,
"loss": 5.0293,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.92455267849129e-05,
"loss": 5.0282,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.923714083740238e-05,
"loss": 5.0182,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922875488989185e-05,
"loss": 5.0257,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.922036894238133e-05,
"loss": 5.013,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921199937367455e-05,
"loss": 5.011,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920361342616403e-05,
"loss": 5.0087,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919522747865351e-05,
"loss": 4.9951,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918684153114299e-05,
"loss": 4.991,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917845558363247e-05,
"loss": 4.9818,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.9170086014925676e-05,
"loss": 4.9792,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.9161700067415156e-05,
"loss": 4.98,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.9153314119904636e-05,
"loss": 4.9626,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.9144928172394116e-05,
"loss": 4.9672,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.9136542224883596e-05,
"loss": 4.9631,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.9128156277373076e-05,
"loss": 4.9494,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.9119770329862556e-05,
"loss": 4.946,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.9111384382352036e-05,
"loss": 4.9413,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.9103014813645245e-05,
"loss": 4.9388,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.909464524493846e-05,
"loss": 4.9313,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.908625929742794e-05,
"loss": 4.9328,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907787334991742e-05,
"loss": 4.9251,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.90694874024069e-05,
"loss": 4.9121,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906111783370011e-05,
"loss": 4.9256,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.905273188618959e-05,
"loss": 4.9075,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.904434593867907e-05,
"loss": 4.914,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903595999116855e-05,
"loss": 4.9065,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.902757404365803e-05,
"loss": 4.9005,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.901918809614751e-05,
"loss": 4.8972,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.901081852744072e-05,
"loss": 4.8914,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.90024325799302e-05,
"loss": 4.9066,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.8994046632419686e-05,
"loss": 4.893,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.8985660684909166e-05,
"loss": 4.8765,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.8977274737398646e-05,
"loss": 4.8732,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.8968888789888125e-05,
"loss": 4.87,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.8960502842377605e-05,
"loss": 4.8656,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.8952116894867085e-05,
"loss": 4.8584,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.8943730947356565e-05,
"loss": 4.8617,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.893534499984604e-05,
"loss": 4.8471,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.8926975431139254e-05,
"loss": 4.8471,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.8918589483628734e-05,
"loss": 4.8421,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.8910219914921943e-05,
"loss": 4.8615,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.8901833967411423e-05,
"loss": 4.8528,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.88934480199009e-05,
"loss": 4.8357,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888506207239038e-05,
"loss": 4.8335,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.887667612487986e-05,
"loss": 4.8278,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.886829017736934e-05,
"loss": 4.8382,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.885990422985882e-05,
"loss": 4.8293,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.88515182823483e-05,
"loss": 4.8278,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.884313233483778e-05,
"loss": 4.8235,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.883476276613099e-05,
"loss": 4.8192,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.882637681862047e-05,
"loss": 4.8152,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.881799087110995e-05,
"loss": 4.8045,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.880960492359943e-05,
"loss": 4.8117,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.880121897608891e-05,
"loss": 4.802,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879283302857839e-05,
"loss": 4.79,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.878444708106787e-05,
"loss": 4.802,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877606113355735e-05,
"loss": 4.7984,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876769156485057e-05,
"loss": 4.7985,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875930561734005e-05,
"loss": 4.7871,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.875093604863326e-05,
"loss": 4.7846,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.7475996017456055,
"eval_runtime": 296.429,
"eval_samples_per_second": 1287.293,
"eval_steps_per_second": 40.229,
"step": 76320
},
{
"epoch": 1.0,
"learning_rate": 4.874255010112274e-05,
"loss": 4.7784,
"step": 76800
},
{
"epoch": 1.0,
"learning_rate": 4.873416415361222e-05,
"loss": 4.775,
"step": 77312
},
{
"epoch": 1.0,
"learning_rate": 4.87257782061017e-05,
"loss": 4.788,
"step": 77824
},
{
"epoch": 1.0,
"learning_rate": 4.871739225859118e-05,
"loss": 4.7724,
"step": 78336
},
{
"epoch": 1.0,
"learning_rate": 4.8709006311080657e-05,
"loss": 4.7751,
"step": 78848
},
{
"epoch": 1.0,
"learning_rate": 4.8700620363570137e-05,
"loss": 4.7598,
"step": 79360
},
{
"epoch": 1.0,
"learning_rate": 4.8692234416059616e-05,
"loss": 4.7737,
"step": 79872
},
{
"epoch": 1.0,
"learning_rate": 4.8683864847352826e-05,
"loss": 4.7559,
"step": 80384
},
{
"epoch": 1.0,
"learning_rate": 4.867549527864604e-05,
"loss": 4.7576,
"step": 80896
},
{
"epoch": 1.0,
"learning_rate": 4.866710933113552e-05,
"loss": 4.755,
"step": 81408
},
{
"epoch": 1.0,
"learning_rate": 4.8658723383625e-05,
"loss": 4.7629,
"step": 81920
},
{
"epoch": 1.0,
"learning_rate": 4.865035381491821e-05,
"loss": 4.7565,
"step": 82432
},
{
"epoch": 1.0,
"learning_rate": 4.864198424621142e-05,
"loss": 4.7374,
"step": 82944
},
{
"epoch": 1.0,
"learning_rate": 4.86335982987009e-05,
"loss": 4.7425,
"step": 83456
},
{
"epoch": 1.0,
"learning_rate": 4.862521235119038e-05,
"loss": 4.7336,
"step": 83968
},
{
"epoch": 1.0,
"learning_rate": 4.861682640367986e-05,
"loss": 4.7349,
"step": 84480
},
{
"epoch": 1.0,
"learning_rate": 4.860844045616934e-05,
"loss": 4.7358,
"step": 84992
},
{
"epoch": 1.0,
"learning_rate": 4.860005450865882e-05,
"loss": 4.7282,
"step": 85504
},
{
"epoch": 1.0,
"learning_rate": 4.85916685611483e-05,
"loss": 4.7318,
"step": 86016
},
{
"epoch": 1.0,
"learning_rate": 4.858328261363778e-05,
"loss": 4.7425,
"step": 86528
},
{
"epoch": 1.0,
"learning_rate": 4.857489666612726e-05,
"loss": 4.7282,
"step": 87040
},
{
"epoch": 1.0,
"learning_rate": 4.8566527097420475e-05,
"loss": 4.7265,
"step": 87552
},
{
"epoch": 1.0,
"learning_rate": 4.8558141149909955e-05,
"loss": 4.7191,
"step": 88064
},
{
"epoch": 1.0,
"learning_rate": 4.8549755202399435e-05,
"loss": 4.7329,
"step": 88576
},
{
"epoch": 1.0,
"learning_rate": 4.8541369254888915e-05,
"loss": 4.7028,
"step": 89088
},
{
"epoch": 1.0,
"learning_rate": 4.8532983307378395e-05,
"loss": 4.716,
"step": 89600
},
{
"epoch": 1.0,
"learning_rate": 4.8524613738671604e-05,
"loss": 4.7028,
"step": 90112
},
{
"epoch": 1.0,
"learning_rate": 4.8516227791161084e-05,
"loss": 4.7057,
"step": 90624
},
{
"epoch": 1.0,
"learning_rate": 4.8507841843650564e-05,
"loss": 4.7015,
"step": 91136
},
{
"epoch": 1.01,
"learning_rate": 4.8499455896140044e-05,
"loss": 4.6946,
"step": 91648
},
{
"epoch": 1.01,
"learning_rate": 4.8491069948629524e-05,
"loss": 4.7015,
"step": 92160
},
{
"epoch": 1.01,
"learning_rate": 4.848270037992273e-05,
"loss": 4.7093,
"step": 92672
},
{
"epoch": 1.01,
"learning_rate": 4.847431443241221e-05,
"loss": 4.6955,
"step": 93184
},
{
"epoch": 1.01,
"learning_rate": 4.846592848490169e-05,
"loss": 4.7016,
"step": 93696
},
{
"epoch": 1.01,
"learning_rate": 4.845754253739117e-05,
"loss": 4.685,
"step": 94208
},
{
"epoch": 1.01,
"learning_rate": 4.844915658988065e-05,
"loss": 4.6933,
"step": 94720
},
{
"epoch": 1.01,
"learning_rate": 4.844077064237013e-05,
"loss": 4.6781,
"step": 95232
},
{
"epoch": 1.01,
"learning_rate": 4.843238469485961e-05,
"loss": 4.6877,
"step": 95744
},
{
"epoch": 1.01,
"learning_rate": 4.842399874734909e-05,
"loss": 4.6654,
"step": 96256
},
{
"epoch": 1.01,
"learning_rate": 4.841561279983857e-05,
"loss": 4.6752,
"step": 96768
},
{
"epoch": 1.01,
"learning_rate": 4.840722685232805e-05,
"loss": 4.6717,
"step": 97280
},
{
"epoch": 1.01,
"learning_rate": 4.839885728362126e-05,
"loss": 4.6839,
"step": 97792
},
{
"epoch": 1.01,
"learning_rate": 4.839047133611074e-05,
"loss": 4.6563,
"step": 98304
},
{
"epoch": 1.01,
"learning_rate": 4.838208538860022e-05,
"loss": 4.6734,
"step": 98816
},
{
"epoch": 1.01,
"learning_rate": 4.83736994410897e-05,
"loss": 4.6692,
"step": 99328
},
{
"epoch": 1.01,
"learning_rate": 4.836531349357918e-05,
"loss": 4.6624,
"step": 99840
},
{
"epoch": 1.01,
"learning_rate": 4.835692754606866e-05,
"loss": 4.6728,
"step": 100352
},
{
"epoch": 1.01,
"learning_rate": 4.834855797736188e-05,
"loss": 4.6479,
"step": 100864
},
{
"epoch": 1.01,
"learning_rate": 4.8340188408655086e-05,
"loss": 4.65,
"step": 101376
},
{
"epoch": 1.01,
"learning_rate": 4.8331802461144566e-05,
"loss": 4.6661,
"step": 101888
},
{
"epoch": 1.01,
"learning_rate": 4.8323416513634046e-05,
"loss": 4.6527,
"step": 102400
},
{
"epoch": 1.01,
"learning_rate": 4.8315030566123526e-05,
"loss": 4.6478,
"step": 102912
},
{
"epoch": 1.01,
"learning_rate": 4.8306644618613006e-05,
"loss": 4.641,
"step": 103424
},
{
"epoch": 1.01,
"learning_rate": 4.8298258671102486e-05,
"loss": 4.6473,
"step": 103936
},
{
"epoch": 1.01,
"learning_rate": 4.8289872723591966e-05,
"loss": 4.6375,
"step": 104448
},
{
"epoch": 1.01,
"learning_rate": 4.8281486776081446e-05,
"loss": 4.6562,
"step": 104960
},
{
"epoch": 1.01,
"learning_rate": 4.8273117207374655e-05,
"loss": 4.6255,
"step": 105472
},
{
"epoch": 1.01,
"learning_rate": 4.8264731259864135e-05,
"loss": 4.6356,
"step": 105984
},
{
"epoch": 1.01,
"learning_rate": 4.8256345312353615e-05,
"loss": 4.6344,
"step": 106496
},
{
"epoch": 1.01,
"learning_rate": 4.8247959364843095e-05,
"loss": 4.6335,
"step": 107008
},
{
"epoch": 1.01,
"learning_rate": 4.823958979613631e-05,
"loss": 4.6387,
"step": 107520
},
{
"epoch": 1.01,
"learning_rate": 4.823120384862579e-05,
"loss": 4.6275,
"step": 108032
},
{
"epoch": 1.01,
"learning_rate": 4.822281790111527e-05,
"loss": 4.6092,
"step": 108544
},
{
"epoch": 1.01,
"learning_rate": 4.821443195360475e-05,
"loss": 4.6164,
"step": 109056
},
{
"epoch": 1.01,
"learning_rate": 4.820604600609423e-05,
"loss": 4.64,
"step": 109568
},
{
"epoch": 1.01,
"learning_rate": 4.819767643738744e-05,
"loss": 4.6128,
"step": 110080
},
{
"epoch": 1.01,
"learning_rate": 4.818929048987692e-05,
"loss": 4.6141,
"step": 110592
},
{
"epoch": 1.01,
"learning_rate": 4.81809045423664e-05,
"loss": 4.6072,
"step": 111104
},
{
"epoch": 1.01,
"learning_rate": 4.817251859485588e-05,
"loss": 4.6083,
"step": 111616
},
{
"epoch": 1.01,
"learning_rate": 4.816413264734535e-05,
"loss": 4.6229,
"step": 112128
},
{
"epoch": 1.01,
"learning_rate": 4.815576307863857e-05,
"loss": 4.6232,
"step": 112640
},
{
"epoch": 1.01,
"learning_rate": 4.814737713112805e-05,
"loss": 4.6074,
"step": 113152
},
{
"epoch": 1.01,
"learning_rate": 4.8138991183617535e-05,
"loss": 4.6199,
"step": 113664
},
{
"epoch": 1.01,
"learning_rate": 4.813060523610701e-05,
"loss": 4.6226,
"step": 114176
},
{
"epoch": 1.01,
"learning_rate": 4.812221928859649e-05,
"loss": 4.6187,
"step": 114688
},
{
"epoch": 1.01,
"learning_rate": 4.8113849719889704e-05,
"loss": 4.5928,
"step": 115200
},
{
"epoch": 1.01,
"learning_rate": 4.8105480151182914e-05,
"loss": 4.6069,
"step": 115712
},
{
"epoch": 1.01,
"learning_rate": 4.8097094203672393e-05,
"loss": 4.5993,
"step": 116224
},
{
"epoch": 1.01,
"learning_rate": 4.8088708256161873e-05,
"loss": 4.6005,
"step": 116736
},
{
"epoch": 1.01,
"learning_rate": 4.808032230865135e-05,
"loss": 4.5964,
"step": 117248
},
{
"epoch": 1.01,
"learning_rate": 4.807195273994456e-05,
"loss": 4.5974,
"step": 117760
},
{
"epoch": 1.01,
"learning_rate": 4.806356679243404e-05,
"loss": 4.6049,
"step": 118272
},
{
"epoch": 1.01,
"learning_rate": 4.805518084492352e-05,
"loss": 4.5982,
"step": 118784
},
{
"epoch": 1.01,
"learning_rate": 4.8046794897413e-05,
"loss": 4.5804,
"step": 119296
},
{
"epoch": 1.01,
"learning_rate": 4.803840894990248e-05,
"loss": 4.586,
"step": 119808
},
{
"epoch": 1.01,
"learning_rate": 4.803002300239196e-05,
"loss": 4.5875,
"step": 120320
},
{
"epoch": 1.01,
"learning_rate": 4.802163705488144e-05,
"loss": 4.5909,
"step": 120832
},
{
"epoch": 1.01,
"learning_rate": 4.801325110737092e-05,
"loss": 4.5899,
"step": 121344
},
{
"epoch": 1.01,
"learning_rate": 4.800488153866413e-05,
"loss": 4.5762,
"step": 121856
},
{
"epoch": 1.02,
"learning_rate": 4.799649559115361e-05,
"loss": 4.5836,
"step": 122368
},
{
"epoch": 1.02,
"learning_rate": 4.798810964364309e-05,
"loss": 4.5686,
"step": 122880
},
{
"epoch": 1.02,
"learning_rate": 4.797972369613257e-05,
"loss": 4.5845,
"step": 123392
},
{
"epoch": 1.02,
"learning_rate": 4.797133774862205e-05,
"loss": 4.5797,
"step": 123904
},
{
"epoch": 1.02,
"learning_rate": 4.796296817991526e-05,
"loss": 4.5739,
"step": 124416
},
{
"epoch": 1.02,
"learning_rate": 4.795458223240474e-05,
"loss": 4.5809,
"step": 124928
},
{
"epoch": 1.02,
"learning_rate": 4.794619628489423e-05,
"loss": 4.572,
"step": 125440
},
{
"epoch": 1.02,
"learning_rate": 4.793781033738371e-05,
"loss": 4.5677,
"step": 125952
},
{
"epoch": 1.02,
"learning_rate": 4.792942438987319e-05,
"loss": 4.5682,
"step": 126464
},
{
"epoch": 1.02,
"learning_rate": 4.792103844236267e-05,
"loss": 4.5696,
"step": 126976
},
{
"epoch": 1.02,
"learning_rate": 4.791265249485215e-05,
"loss": 4.5658,
"step": 127488
},
{
"epoch": 1.02,
"learning_rate": 4.790426654734163e-05,
"loss": 4.5572,
"step": 128000
},
{
"epoch": 1.02,
"learning_rate": 4.7895880599831107e-05,
"loss": 4.5611,
"step": 128512
},
{
"epoch": 1.02,
"learning_rate": 4.7887494652320587e-05,
"loss": 4.5635,
"step": 129024
},
{
"epoch": 1.02,
"learning_rate": 4.7879108704810067e-05,
"loss": 4.5587,
"step": 129536
},
{
"epoch": 1.02,
"learning_rate": 4.7870739136103276e-05,
"loss": 4.5431,
"step": 130048
},
{
"epoch": 1.02,
"learning_rate": 4.7862353188592756e-05,
"loss": 4.5559,
"step": 130560
},
{
"epoch": 1.02,
"learning_rate": 4.7853967241082236e-05,
"loss": 4.556,
"step": 131072
},
{
"epoch": 1.02,
"learning_rate": 4.7845581293571715e-05,
"loss": 4.5493,
"step": 131584
},
{
"epoch": 1.02,
"learning_rate": 4.7837195346061195e-05,
"loss": 4.5488,
"step": 132096
},
{
"epoch": 1.02,
"learning_rate": 4.7828809398550675e-05,
"loss": 4.5484,
"step": 132608
},
{
"epoch": 1.02,
"learning_rate": 4.7820423451040155e-05,
"loss": 4.5364,
"step": 133120
},
{
"epoch": 1.02,
"learning_rate": 4.781205388233337e-05,
"loss": 4.5578,
"step": 133632
},
{
"epoch": 1.02,
"learning_rate": 4.780368431362658e-05,
"loss": 4.5425,
"step": 134144
},
{
"epoch": 1.02,
"learning_rate": 4.779529836611606e-05,
"loss": 4.5465,
"step": 134656
},
{
"epoch": 1.02,
"learning_rate": 4.778691241860554e-05,
"loss": 4.5421,
"step": 135168
},
{
"epoch": 1.02,
"learning_rate": 4.777852647109502e-05,
"loss": 4.5486,
"step": 135680
},
{
"epoch": 1.02,
"learning_rate": 4.777014052358449e-05,
"loss": 4.5388,
"step": 136192
},
{
"epoch": 1.02,
"learning_rate": 4.776175457607397e-05,
"loss": 4.5412,
"step": 136704
},
{
"epoch": 1.02,
"learning_rate": 4.775336862856345e-05,
"loss": 4.5519,
"step": 137216
},
{
"epoch": 1.02,
"learning_rate": 4.774498268105293e-05,
"loss": 4.5429,
"step": 137728
},
{
"epoch": 1.02,
"learning_rate": 4.773659673354241e-05,
"loss": 4.5326,
"step": 138240
},
{
"epoch": 1.02,
"learning_rate": 4.772822716483563e-05,
"loss": 4.5328,
"step": 138752
},
{
"epoch": 1.02,
"learning_rate": 4.771984121732511e-05,
"loss": 4.5275,
"step": 139264
},
{
"epoch": 1.02,
"learning_rate": 4.771145526981459e-05,
"loss": 4.5259,
"step": 139776
},
{
"epoch": 1.02,
"learning_rate": 4.770306932230407e-05,
"loss": 4.5267,
"step": 140288
},
{
"epoch": 1.02,
"learning_rate": 4.769468337479355e-05,
"loss": 4.5245,
"step": 140800
},
{
"epoch": 1.02,
"learning_rate": 4.768631380608676e-05,
"loss": 4.5164,
"step": 141312
},
{
"epoch": 1.02,
"learning_rate": 4.767792785857624e-05,
"loss": 4.5206,
"step": 141824
},
{
"epoch": 1.02,
"learning_rate": 4.766954191106572e-05,
"loss": 4.5169,
"step": 142336
},
{
"epoch": 1.02,
"learning_rate": 4.766117234235893e-05,
"loss": 4.5342,
"step": 142848
},
{
"epoch": 1.02,
"learning_rate": 4.765278639484841e-05,
"loss": 4.5353,
"step": 143360
},
{
"epoch": 1.02,
"learning_rate": 4.764440044733789e-05,
"loss": 4.5214,
"step": 143872
},
{
"epoch": 1.02,
"learning_rate": 4.763601449982737e-05,
"loss": 4.5115,
"step": 144384
},
{
"epoch": 1.02,
"learning_rate": 4.762762855231685e-05,
"loss": 4.5181,
"step": 144896
},
{
"epoch": 1.02,
"learning_rate": 4.761924260480633e-05,
"loss": 4.5231,
"step": 145408
},
{
"epoch": 1.02,
"learning_rate": 4.7610856657295813e-05,
"loss": 4.5234,
"step": 145920
},
{
"epoch": 1.02,
"learning_rate": 4.7602470709785293e-05,
"loss": 4.5159,
"step": 146432
},
{
"epoch": 1.02,
"learning_rate": 4.7594084762274773e-05,
"loss": 4.5205,
"step": 146944
},
{
"epoch": 1.02,
"learning_rate": 4.758569881476425e-05,
"loss": 4.5118,
"step": 147456
},
{
"epoch": 1.02,
"learning_rate": 4.757732924605746e-05,
"loss": 4.5136,
"step": 147968
},
{
"epoch": 1.02,
"learning_rate": 4.756894329854694e-05,
"loss": 4.5086,
"step": 148480
},
{
"epoch": 1.02,
"learning_rate": 4.756055735103642e-05,
"loss": 4.5128,
"step": 148992
},
{
"epoch": 1.02,
"learning_rate": 4.75521714035259e-05,
"loss": 4.5053,
"step": 149504
},
{
"epoch": 1.02,
"learning_rate": 4.7543785456015375e-05,
"loss": 4.5,
"step": 150016
},
{
"epoch": 1.02,
"learning_rate": 4.7535399508504855e-05,
"loss": 4.508,
"step": 150528
},
{
"epoch": 1.02,
"learning_rate": 4.752702993979807e-05,
"loss": 4.5098,
"step": 151040
},
{
"epoch": 1.02,
"learning_rate": 4.751864399228755e-05,
"loss": 4.5114,
"step": 151552
},
{
"epoch": 1.02,
"learning_rate": 4.751025804477703e-05,
"loss": 4.4993,
"step": 152064
},
{
"epoch": 1.02,
"learning_rate": 4.750187209726651e-05,
"loss": 4.5007,
"step": 152576
},
{
"epoch": 1.03,
"eval_loss": 4.469255447387695,
"eval_runtime": 295.4389,
"eval_samples_per_second": 1291.607,
"eval_steps_per_second": 40.364,
"step": 152640
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 6.308165527496251e+16,
"trial_name": null,
"trial_params": null
}