| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 1198, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.001669449081803005, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0000000000000002e-07, | |
| "loss": 2.0754, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00333889816360601, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 1.8508, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.005008347245409015, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 1.7493, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00667779632721202, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 1.7768, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.008347245409015025, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.8859, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01001669449081803, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 2.1114, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.011686143572621035, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 1.8268, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01335559265442404, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 1.427, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.015025041736227046, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8000000000000001e-06, | |
| "loss": 1.8472, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.01669449081803005, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.7571, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.018363939899833055, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2e-06, | |
| "loss": 1.8135, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02003338898163606, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.6594, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.021702838063439065, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6e-06, | |
| "loss": 1.8615, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.02337228714524207, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 1.9803, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.025041736227045076, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3e-06, | |
| "loss": 1.9604, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02671118530884808, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 1.8476, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.028380634390651086, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.71, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03005008347245409, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 1.9865, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03171953255425709, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 1.9356, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0333889816360601, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.6531, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.035058430717863104, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 1.6921, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.03672787979966611, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.7874, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.038397328881469114, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.600000000000001e-06, | |
| "loss": 1.8481, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.04006677796327212, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 1.6141, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.041736227045075125, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5e-06, | |
| "loss": 1.621, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04340567612687813, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.2e-06, | |
| "loss": 1.5873, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.045075125208681135, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 1.6264, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.04674457429048414, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 1.7444, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.048414023372287146, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.8e-06, | |
| "loss": 1.9601, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.05008347245409015, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6e-06, | |
| "loss": 1.7666, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05175292153589316, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.200000000000001e-06, | |
| "loss": 1.973, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.05342237061769616, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.6661, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.05509181969949917, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.600000000000001e-06, | |
| "loss": 1.9688, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.05676126878130217, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 1.9153, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.05843071786310518, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7e-06, | |
| "loss": 1.804, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06010016694490818, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 1.7318, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.06176961602671119, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.4e-06, | |
| "loss": 2.0672, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.06343906510851419, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 1.8864, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0651085141903172, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.800000000000002e-06, | |
| "loss": 1.7625, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0667779632721202, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.7466, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06844741235392321, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.2e-06, | |
| "loss": 1.5579, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.07011686143572621, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.83, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.07178631051752922, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.6e-06, | |
| "loss": 1.9467, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.07345575959933222, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.8e-06, | |
| "loss": 1.9879, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.07512520868113523, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9e-06, | |
| "loss": 1.9357, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.07679465776293823, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 1.7311, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.07846410684474124, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.658, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.08013355592654424, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 1.6306, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.08180300500834725, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.800000000000001e-06, | |
| "loss": 1.783, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.08347245409015025, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8001, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08514190317195326, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.999981277850585e-06, | |
| "loss": 1.6999, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.08681135225375626, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.999925111542544e-06, | |
| "loss": 1.7434, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.08848080133555926, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.999831501496497e-06, | |
| "loss": 1.8493, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.09015025041736227, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.999700448413483e-06, | |
| "loss": 2.01, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.09181969949916527, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.999531953274934e-06, | |
| "loss": 1.8475, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.09348914858096828, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.999326017342688e-06, | |
| "loss": 1.5294, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.09515859766277128, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.999082642158972e-06, | |
| "loss": 1.6968, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.09682804674457429, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.998801829546387e-06, | |
| "loss": 1.6576, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.09849749582637729, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.9984835816079e-06, | |
| "loss": 1.9567, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1001669449081803, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.998127900726825e-06, | |
| "loss": 1.5722, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1018363939899833, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.997734789566809e-06, | |
| "loss": 1.5099, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.10350584307178631, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.997304251071802e-06, | |
| "loss": 1.6942, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.10517529215358931, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.996836288466046e-06, | |
| "loss": 1.4786, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.10684474123539232, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.99633090525405e-06, | |
| "loss": 1.7518, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.10851419031719532, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.99578810522056e-06, | |
| "loss": 1.8813, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.11018363939899833, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.995207892430525e-06, | |
| "loss": 1.6325, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.11185308848080133, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.994590271229077e-06, | |
| "loss": 1.6373, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.11352253756260434, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.9939352462415e-06, | |
| "loss": 1.7568, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.11519198664440734, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.993242822373178e-06, | |
| "loss": 1.8238, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.11686143572621036, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.99251300480958e-06, | |
| "loss": 1.8395, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.11853088480801335, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.991745799016206e-06, | |
| "loss": 1.4879, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.12020033388981637, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.990941210738553e-06, | |
| "loss": 1.9031, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.12186978297161936, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.990099246002071e-06, | |
| "loss": 1.7939, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.12353923205342238, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.989219911112114e-06, | |
| "loss": 1.7135, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.12520868113522537, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.988303212653898e-06, | |
| "loss": 1.7721, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.12687813021702837, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.98734915749245e-06, | |
| "loss": 1.7121, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1285475792988314, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.986357752772555e-06, | |
| "loss": 1.6391, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1302170283806344, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.985329005918702e-06, | |
| "loss": 1.7974, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1318864774624374, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.984262924635036e-06, | |
| "loss": 1.5529, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1335559265442404, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.983159516905287e-06, | |
| "loss": 1.9062, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1352253756260434, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.982018790992722e-06, | |
| "loss": 1.7188, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.13689482470784642, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.980840755440075e-06, | |
| "loss": 1.6821, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.13856427378964942, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.979625419069495e-06, | |
| "loss": 2.0225, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.14023372287145242, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.978372790982457e-06, | |
| "loss": 1.7347, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1419031719532554, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.977082880559725e-06, | |
| "loss": 1.7869, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.14357262103505844, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.975755697461254e-06, | |
| "loss": 1.7912, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.14524207011686144, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.974391251626132e-06, | |
| "loss": 1.6523, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.14691151919866444, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.972989553272501e-06, | |
| "loss": 1.7847, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.14858096828046743, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.971550612897487e-06, | |
| "loss": 1.5771, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.15025041736227046, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.970074441277111e-06, | |
| "loss": 1.721, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.15191986644407346, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.968561049466214e-06, | |
| "loss": 1.7414, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.15358931552587646, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.967010448798376e-06, | |
| "loss": 1.6912, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.15525876460767946, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.965422650885829e-06, | |
| "loss": 1.5043, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.15692821368948248, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.963797667619368e-06, | |
| "loss": 1.654, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.15859766277128548, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.962135511168263e-06, | |
| "loss": 1.6316, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.16026711185308848, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.960436193980175e-06, | |
| "loss": 1.5251, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.16193656093489148, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.958699728781046e-06, | |
| "loss": 1.919, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.1636060100166945, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.956926128575026e-06, | |
| "loss": 1.8145, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.1652754590984975, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.955115406644357e-06, | |
| "loss": 1.7591, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.1669449081803005, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.953267576549279e-06, | |
| "loss": 1.7051, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1686143572621035, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.951382652127935e-06, | |
| "loss": 1.6331, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.17028380634390652, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.949460647496258e-06, | |
| "loss": 1.8922, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.17195325542570952, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.947501577047874e-06, | |
| "loss": 1.9153, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.17362270450751252, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.945505455453983e-06, | |
| "loss": 1.7728, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.17529215358931552, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.943472297663262e-06, | |
| "loss": 1.615, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.17696160267111852, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.941402118901743e-06, | |
| "loss": 1.6637, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.17863105175292154, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.939294934672707e-06, | |
| "loss": 1.6815, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.18030050083472454, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.93715076075656e-06, | |
| "loss": 1.7668, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.18196994991652754, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.934969613210718e-06, | |
| "loss": 1.4658, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.18363939899833054, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.932751508369492e-06, | |
| "loss": 1.7838, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.18530884808013356, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.930496462843954e-06, | |
| "loss": 1.666, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.18697829716193656, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.92820449352183e-06, | |
| "loss": 1.7945, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.18864774624373956, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.92587561756735e-06, | |
| "loss": 1.762, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.19031719532554256, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.923509852421144e-06, | |
| "loss": 1.5722, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.19198664440734559, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.921107215800095e-06, | |
| "loss": 1.7682, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.19365609348914858, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.91866772569721e-06, | |
| "loss": 1.8919, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.19532554257095158, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.91619140038149e-06, | |
| "loss": 1.9279, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.19699499165275458, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.913678258397785e-06, | |
| "loss": 1.784, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.1986644407345576, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.911128318566668e-06, | |
| "loss": 1.978, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.2003338898163606, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.908541599984276e-06, | |
| "loss": 1.6277, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2020033388981636, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.905918122022183e-06, | |
| "loss": 1.55, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.2036727879799666, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.90325790432725e-06, | |
| "loss": 1.6347, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.20534223706176963, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.90056096682147e-06, | |
| "loss": 1.786, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.20701168614357263, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.897827329701834e-06, | |
| "loss": 1.7375, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.20868113522537562, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.895057013440163e-06, | |
| "loss": 1.6394, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.21035058430717862, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.892250038782972e-06, | |
| "loss": 1.6817, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.21202003338898165, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.889406426751296e-06, | |
| "loss": 1.9121, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.21368948247078465, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.88652619864055e-06, | |
| "loss": 1.4702, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.21535893155258765, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.883609376020356e-06, | |
| "loss": 1.6333, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.21702838063439064, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.880655980734391e-06, | |
| "loss": 1.64, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.21869782971619364, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.877666034900216e-06, | |
| "loss": 1.6464, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.22036727879799667, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.874639560909118e-06, | |
| "loss": 1.7207, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.22203672787979967, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.871576581425937e-06, | |
| "loss": 1.8864, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.22370617696160267, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.868477119388897e-06, | |
| "loss": 1.7744, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.22537562604340566, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.86534119800943e-06, | |
| "loss": 1.592, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2270450751252087, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.862168840772018e-06, | |
| "loss": 1.7051, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.2287145242070117, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.858960071433994e-06, | |
| "loss": 1.7263, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.2303839732888147, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.855714914025386e-06, | |
| "loss": 1.6824, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.23205342237061768, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.852433392848718e-06, | |
| "loss": 1.8981, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.2337228714524207, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.849115532478848e-06, | |
| "loss": 1.4935, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2353923205342237, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.84576135776276e-06, | |
| "loss": 1.3096, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.2370617696160267, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.842370893819404e-06, | |
| "loss": 1.963, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.2387312186978297, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.838944166039486e-06, | |
| "loss": 1.7601, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.24040066777963273, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.83548120008529e-06, | |
| "loss": 1.694, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.24207011686143573, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.831982021890483e-06, | |
| "loss": 1.697, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.24373956594323873, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.828446657659919e-06, | |
| "loss": 1.3733, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.24540901502504173, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.824875133869447e-06, | |
| "loss": 1.9497, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.24707846410684475, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.821267477265705e-06, | |
| "loss": 1.8237, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.24874791318864775, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.817623714865931e-06, | |
| "loss": 1.782, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.25041736227045075, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.813943873957748e-06, | |
| "loss": 1.7703, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25208681135225375, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.810227982098968e-06, | |
| "loss": 1.7999, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.25375626043405675, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.806476067117384e-06, | |
| "loss": 1.5728, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.25542570951585974, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.802688157110564e-06, | |
| "loss": 1.5857, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.2570951585976628, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.798864280445633e-06, | |
| "loss": 1.6758, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.2587646076794658, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.795004465759067e-06, | |
| "loss": 1.5893, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.2604340567612688, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.791108741956476e-06, | |
| "loss": 1.8483, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2621035058430718, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.787177138212391e-06, | |
| "loss": 1.9552, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.2637729549248748, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.78320968397004e-06, | |
| "loss": 1.7145, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.2654424040066778, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.779206408941131e-06, | |
| "loss": 1.6563, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.2671118530884808, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.77516734310563e-06, | |
| "loss": 1.7699, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2687813021702838, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.771092516711538e-06, | |
| "loss": 1.6084, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.2704507512520868, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.766981960274653e-06, | |
| "loss": 1.6172, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.27212020033388984, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.76283570457836e-06, | |
| "loss": 1.7309, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.27378964941569284, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.758653780673381e-06, | |
| "loss": 1.8666, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.27545909849749584, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.754436219877564e-06, | |
| "loss": 1.9176, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.27712854757929883, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.750183053775625e-06, | |
| "loss": 1.9063, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.27879799666110183, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.745894314218933e-06, | |
| "loss": 1.8567, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.28046744574290483, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.741570033325254e-06, | |
| "loss": 1.861, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.28213689482470783, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.737210243478522e-06, | |
| "loss": 1.7261, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.2838063439065108, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.732814977328593e-06, | |
| "loss": 1.6596, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.2854757929883139, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.728384267790997e-06, | |
| "loss": 1.9513, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.2871452420701169, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.723918148046696e-06, | |
| "loss": 1.6785, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.2888146911519199, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.719416651541839e-06, | |
| "loss": 1.7828, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.2904841402337229, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.714879811987496e-06, | |
| "loss": 1.9663, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.2921535893155259, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.710307663359426e-06, | |
| "loss": 1.6448, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2938230383973289, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.705700239897809e-06, | |
| "loss": 1.5589, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.29549248747913187, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.701057576106991e-06, | |
| "loss": 1.9398, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.29716193656093487, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.69637970675523e-06, | |
| "loss": 1.7781, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.2988313856427379, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.691666666874438e-06, | |
| "loss": 1.7034, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.3005008347245409, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.686918491759904e-06, | |
| "loss": 1.9747, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3021702838063439, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.682135216970048e-06, | |
| "loss": 1.627, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.3038397328881469, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.677316878326144e-06, | |
| "loss": 1.6952, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3055091819699499, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.672463511912056e-06, | |
| "loss": 1.9787, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.3071786310517529, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.667575154073962e-06, | |
| "loss": 1.7576, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.3088480801335559, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.66265184142009e-06, | |
| "loss": 1.4266, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.3105175292153589, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.657693610820437e-06, | |
| "loss": 1.84, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.3121869782971619, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.652700499406497e-06, | |
| "loss": 1.9044, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.31385642737896496, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.647672544570981e-06, | |
| "loss": 1.9096, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.31552587646076796, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.642609783967539e-06, | |
| "loss": 1.9484, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.31719532554257096, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.637512255510475e-06, | |
| "loss": 1.832, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.31886477462437396, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.632379997374462e-06, | |
| "loss": 1.8468, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.32053422370617696, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.627213047994265e-06, | |
| "loss": 1.5239, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.32220367278797996, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.622011446064439e-06, | |
| "loss": 1.9019, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.32387312186978295, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.616775230539057e-06, | |
| "loss": 1.6285, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.32554257095158595, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.611504440631398e-06, | |
| "loss": 1.7333, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.327212020033389, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.606199115813672e-06, | |
| "loss": 1.4345, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.328881469115192, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.600859295816708e-06, | |
| "loss": 1.7107, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.330550918196995, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.595485020629676e-06, | |
| "loss": 1.7476, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.332220367278798, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.590076330499763e-06, | |
| "loss": 1.8789, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.333889816360601, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.584633265931894e-06, | |
| "loss": 1.8005, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.335559265442404, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.579155867688415e-06, | |
| "loss": 1.9923, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.337228714524207, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.573644176788795e-06, | |
| "loss": 1.827, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.33889816360601, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.568098234509312e-06, | |
| "loss": 1.8459, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.34056761268781305, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.562518082382751e-06, | |
| "loss": 1.6333, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.34223706176961605, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.55690376219809e-06, | |
| "loss": 1.7023, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.34390651085141904, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.551255316000183e-06, | |
| "loss": 1.7017, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.34557595993322204, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.545572786089452e-06, | |
| "loss": 1.7043, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.34724540901502504, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.539856215021568e-06, | |
| "loss": 1.7999, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.34891485809682804, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.53410564560713e-06, | |
| "loss": 1.409, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.35058430717863104, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.528321120911345e-06, | |
| "loss": 1.4238, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.35225375626043404, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.522502684253709e-06, | |
| "loss": 1.958, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.35392320534223703, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.516650379207677e-06, | |
| "loss": 1.8037, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.3555926544240401, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.510764249600339e-06, | |
| "loss": 1.8266, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.3572621035058431, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.504844339512096e-06, | |
| "loss": 1.9888, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.3589315525876461, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.498890693276326e-06, | |
| "loss": 1.8452, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.3606010016694491, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.492903355479047e-06, | |
| "loss": 1.7189, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.3622704507512521, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.486882370958596e-06, | |
| "loss": 1.7222, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.3639398998330551, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.480827784805278e-06, | |
| "loss": 1.8466, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.3656093489148581, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.474739642361043e-06, | |
| "loss": 1.7537, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.3672787979966611, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.468617989219136e-06, | |
| "loss": 1.9107, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.36894824707846413, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.462462871223755e-06, | |
| "loss": 1.5911, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.37061769616026713, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.45627433446972e-06, | |
| "loss": 1.5626, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.3722871452420701, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.450052425302112e-06, | |
| "loss": 1.6464, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.3739565943238731, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.443797190315938e-06, | |
| "loss": 1.8094, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.3756260434056761, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.437508676355774e-06, | |
| "loss": 1.5682, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.3772954924874791, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.431186930515419e-06, | |
| "loss": 1.5277, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.3789649415692821, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.424832000137542e-06, | |
| "loss": 1.8445, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.3806343906510851, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.418443932813328e-06, | |
| "loss": 1.6292, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.3823038397328882, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.412022776382113e-06, | |
| "loss": 2.0086, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.38397328881469117, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.405568578931042e-06, | |
| "loss": 1.9954, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.38564273789649417, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.399081388794688e-06, | |
| "loss": 1.9507, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.38731218697829717, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.392561254554712e-06, | |
| "loss": 1.9047, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.38898163606010017, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.386008225039486e-06, | |
| "loss": 1.81, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.39065108514190316, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.379422349323728e-06, | |
| "loss": 1.5756, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.39232053422370616, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.372803676728138e-06, | |
| "loss": 1.5665, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.39398998330550916, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.366152256819025e-06, | |
| "loss": 1.6799, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.39565943238731216, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.359468139407942e-06, | |
| "loss": 1.4389, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.3973288814691152, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.352751374551305e-06, | |
| "loss": 1.8924, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.3989983305509182, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.346002012550027e-06, | |
| "loss": 1.8713, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.4006677796327212, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.339220103949132e-06, | |
| "loss": 1.7269, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.4023372287145242, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.332405699537382e-06, | |
| "loss": 1.651, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.4040066777963272, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.325558850346897e-06, | |
| "loss": 1.6421, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.4056761268781302, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.318679607652768e-06, | |
| "loss": 1.6585, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.4073455759599332, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.311768022972682e-06, | |
| "loss": 1.8029, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.4090150250417362, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.304824148066526e-06, | |
| "loss": 1.5796, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.41068447412353926, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.297848034936007e-06, | |
| "loss": 1.7035, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.41235392320534225, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.290839735824254e-06, | |
| "loss": 1.8096, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.41402337228714525, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.283799303215442e-06, | |
| "loss": 1.7213, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.41569282136894825, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.276726789834378e-06, | |
| "loss": 1.8537, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.41736227045075125, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.269622248646124e-06, | |
| "loss": 1.7887, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.41903171953255425, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.262485732855597e-06, | |
| "loss": 1.6711, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.42070116861435725, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.255317295907158e-06, | |
| "loss": 1.7755, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.42237061769616024, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.24811699148423e-06, | |
| "loss": 1.8195, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.4240400667779633, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.240884873508876e-06, | |
| "loss": 2.0062, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.4257095158597663, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.233620996141421e-06, | |
| "loss": 1.7525, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.4273789649415693, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.226325413780021e-06, | |
| "loss": 1.338, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.4290484140233723, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.218998181060271e-06, | |
| "loss": 1.7117, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.4307178631051753, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.211639352854786e-06, | |
| "loss": 1.7551, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.4323873121869783, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.204248984272802e-06, | |
| "loss": 1.4557, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.4340567612687813, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.196827130659752e-06, | |
| "loss": 1.6702, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.4357262103505843, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.189373847596853e-06, | |
| "loss": 1.6316, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.4373956594323873, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.181889190900702e-06, | |
| "loss": 1.6081, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.43906510851419034, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.174373216622841e-06, | |
| "loss": 1.4578, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.44073455759599334, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.166825981049345e-06, | |
| "loss": 1.8555, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.44240400667779634, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.15924754070041e-06, | |
| "loss": 1.7215, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.44407345575959933, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.151637952329903e-06, | |
| "loss": 1.8248, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.44574290484140233, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.143997272924974e-06, | |
| "loss": 1.8428, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.44741235392320533, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.136325559705593e-06, | |
| "loss": 1.8746, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.44908180300500833, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.128622870124147e-06, | |
| "loss": 1.7128, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.4507512520868113, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.120889261864999e-06, | |
| "loss": 1.569, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.4524207011686144, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.113124792844053e-06, | |
| "loss": 1.6879, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.4540901502504174, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.105329521208334e-06, | |
| "loss": 1.7053, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.4557595993322204, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.097503505335534e-06, | |
| "loss": 1.763, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.4574290484140234, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.089646803833589e-06, | |
| "loss": 1.6282, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.4590984974958264, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.081759475540236e-06, | |
| "loss": 1.7139, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.4607679465776294, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.073841579522571e-06, | |
| "loss": 1.7466, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.46243739565943237, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.065893175076604e-06, | |
| "loss": 1.7338, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.46410684474123537, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.057914321726824e-06, | |
| "loss": 1.9141, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.4657762938230384, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.049905079225744e-06, | |
| "loss": 1.6576, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.4674457429048414, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.041865507553458e-06, | |
| "loss": 1.7862, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4691151919866444, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.033795666917191e-06, | |
| "loss": 1.7305, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.4707846410684474, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.025695617750848e-06, | |
| "loss": 1.9305, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.4724540901502504, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.01756542071456e-06, | |
| "loss": 1.6688, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.4741235392320534, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.009405136694234e-06, | |
| "loss": 1.9435, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.4757929883138564, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.001214826801092e-06, | |
| "loss": 2.0056, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.4774624373956594, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.992994552371217e-06, | |
| "loss": 1.5892, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.4791318864774624, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.98474437496509e-06, | |
| "loss": 1.6205, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.48080133555926546, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.976464356367133e-06, | |
| "loss": 1.7236, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.48247078464106846, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.968154558585244e-06, | |
| "loss": 1.646, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.48414023372287146, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.959815043850336e-06, | |
| "loss": 1.5001, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.48580968280467446, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.951445874615862e-06, | |
| "loss": 1.7625, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.48747913188647746, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.943047113557358e-06, | |
| "loss": 1.848, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.48914858096828046, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.934618823571968e-06, | |
| "loss": 1.6042, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.49081803005008345, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.926161067777973e-06, | |
| "loss": 1.6858, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.49248747913188645, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.917673909514321e-06, | |
| "loss": 2.0782, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.4941569282136895, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.90915741234015e-06, | |
| "loss": 1.6056, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.4958263772954925, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.900611640034313e-06, | |
| "loss": 1.9565, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.4974958263772955, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.892036656594898e-06, | |
| "loss": 1.4701, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.4991652754590985, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.883432526238757e-06, | |
| "loss": 1.4169, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.5008347245409015, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.874799313401014e-06, | |
| "loss": 1.6465, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5025041736227045, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.866137082734591e-06, | |
| "loss": 1.5715, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.5041736227045075, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.857445899109716e-06, | |
| "loss": 1.848, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.5058430717863105, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.848725827613445e-06, | |
| "loss": 1.7313, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.5075125208681135, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.839976933549173e-06, | |
| "loss": 1.8501, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.5091819699499165, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.831199282436136e-06, | |
| "loss": 1.5231, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.5108514190317195, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.822392940008937e-06, | |
| "loss": 1.4616, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.5125208681135225, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.813557972217038e-06, | |
| "loss": 1.5747, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.5141903171953256, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.804694445224274e-06, | |
| "loss": 1.5613, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.5158597662771286, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.795802425408352e-06, | |
| "loss": 1.3077, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.5175292153589316, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.786881979360368e-06, | |
| "loss": 1.5789, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.5191986644407346, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.777933173884288e-06, | |
| "loss": 1.7127, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.5208681135225376, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.76895607599646e-06, | |
| "loss": 1.7422, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.5225375626043406, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.759950752925114e-06, | |
| "loss": 1.6902, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.5242070116861436, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.750917272109849e-06, | |
| "loss": 1.763, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.5258764607679466, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.741855701201138e-06, | |
| "loss": 1.7908, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.5275459098497496, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.732766108059814e-06, | |
| "loss": 1.6755, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.5292153589315526, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.723648560756565e-06, | |
| "loss": 1.7597, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.5308848080133556, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.714503127571425e-06, | |
| "loss": 1.9694, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.5325542570951586, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.705329876993262e-06, | |
| "loss": 1.6419, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.5342237061769616, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.696128877719258e-06, | |
| "loss": 1.8573, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5358931552587646, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.686900198654413e-06, | |
| "loss": 1.7608, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.5375626043405676, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.677643908911007e-06, | |
| "loss": 2.0053, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.5392320534223706, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.668360077808093e-06, | |
| "loss": 1.8578, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.5409015025041736, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.659048774870986e-06, | |
| "loss": 1.6681, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.5425709515859767, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.649710069830723e-06, | |
| "loss": 1.6123, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5442404006677797, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.64034403262356e-06, | |
| "loss": 1.7475, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.5459098497495827, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.630950733390434e-06, | |
| "loss": 1.3078, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.5475792988313857, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.621530242476446e-06, | |
| "loss": 1.6937, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.5492487479131887, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.612082630430333e-06, | |
| "loss": 1.9197, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.5509181969949917, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.602607968003935e-06, | |
| "loss": 1.7944, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5525876460767947, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.593106326151672e-06, | |
| "loss": 1.83, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.5542570951585977, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.583577776030005e-06, | |
| "loss": 2.0895, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.5559265442404007, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.574022388996913e-06, | |
| "loss": 1.6171, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.5575959933222037, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.564440236611344e-06, | |
| "loss": 1.486, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.5592654424040067, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.5548313906327e-06, | |
| "loss": 1.7145, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.5609348914858097, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.545195923020273e-06, | |
| "loss": 1.713, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.5626043405676127, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.535533905932739e-06, | |
| "loss": 1.756, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.5642737896494157, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.525845411727581e-06, | |
| "loss": 1.5199, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.5659432387312187, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.516130512960576e-06, | |
| "loss": 1.7866, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.5676126878130217, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.506389282385242e-06, | |
| "loss": 1.9604, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.5692821368948247, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.49662179295228e-06, | |
| "loss": 1.8083, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.5709515859766278, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.486828117809057e-06, | |
| "loss": 1.8973, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.5726210350584308, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.47700833029903e-06, | |
| "loss": 1.6784, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.5742904841402338, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.467162503961209e-06, | |
| "loss": 1.7076, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.5759599332220368, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.45729071252961e-06, | |
| "loss": 1.4877, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.5776293823038398, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.447393029932692e-06, | |
| "loss": 1.8807, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.5792988313856428, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.43746953029281e-06, | |
| "loss": 1.8924, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.5809682804674458, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.427520287925669e-06, | |
| "loss": 1.7638, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.5826377295492488, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.417545377339739e-06, | |
| "loss": 1.8664, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.5843071786310517, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.407544873235736e-06, | |
| "loss": 1.6039, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5859766277128547, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.39751885050603e-06, | |
| "loss": 1.8483, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.5876460767946577, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.387467384234096e-06, | |
| "loss": 1.9636, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.5893155258764607, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.377390549693959e-06, | |
| "loss": 1.3699, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.5909849749582637, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.367288422349617e-06, | |
| "loss": 1.7837, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.5926544240400667, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.35716107785449e-06, | |
| "loss": 1.7749, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.5943238731218697, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.347008592050834e-06, | |
| "loss": 1.6986, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.5959933222036727, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.336831040969196e-06, | |
| "loss": 1.7904, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.5976627712854758, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.326628500827826e-06, | |
| "loss": 1.9549, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.5993322203672788, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.316401048032121e-06, | |
| "loss": 1.7828, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.6010016694490818, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.306148759174036e-06, | |
| "loss": 1.7061, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6026711185308848, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.295871711031527e-06, | |
| "loss": 1.5943, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.6043405676126878, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.285569980567965e-06, | |
| "loss": 1.7353, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.6060100166944908, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.275243644931565e-06, | |
| "loss": 1.7265, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.6076794657762938, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.264892781454807e-06, | |
| "loss": 1.9438, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.6093489148580968, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.254517467653858e-06, | |
| "loss": 1.6295, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.6110183639398998, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.244117781227982e-06, | |
| "loss": 2.0601, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.6126878130217028, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.23369380005898e-06, | |
| "loss": 1.6861, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.6143572621035058, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.22324560221058e-06, | |
| "loss": 1.68, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.6160267111853088, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.21277326592787e-06, | |
| "loss": 1.683, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.6176961602671118, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.202276869636713e-06, | |
| "loss": 1.4055, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6193656093489148, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.191756491943146e-06, | |
| "loss": 1.7286, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.6210350584307178, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.1812122116328e-06, | |
| "loss": 1.7372, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.6227045075125208, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.170644107670313e-06, | |
| "loss": 1.5938, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.6243739565943238, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.160052259198737e-06, | |
| "loss": 1.5759, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.6260434056761269, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.149436745538934e-06, | |
| "loss": 1.6956, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.6277128547579299, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.138797646189e-06, | |
| "loss": 1.5602, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.6293823038397329, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.128135040823661e-06, | |
| "loss": 1.5855, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.6310517529215359, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.117449009293668e-06, | |
| "loss": 1.9672, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.6327212020033389, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.106739631625216e-06, | |
| "loss": 1.8238, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.6343906510851419, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.096006988019331e-06, | |
| "loss": 1.8443, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6360601001669449, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.085251158851278e-06, | |
| "loss": 1.7355, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.6377295492487479, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.074472224669952e-06, | |
| "loss": 1.6855, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.6393989983305509, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.063670266197278e-06, | |
| "loss": 1.9201, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.6410684474123539, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.052845364327609e-06, | |
| "loss": 1.6821, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.6427378964941569, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.041997600127118e-06, | |
| "loss": 1.8732, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.6444073455759599, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.031127054833192e-06, | |
| "loss": 1.5744, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.6460767946577629, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.020233809853815e-06, | |
| "loss": 1.6747, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.6477462437395659, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.009317946766975e-06, | |
| "loss": 1.6959, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.6494156928213689, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.998379547320038e-06, | |
| "loss": 1.464, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.6510851419031719, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.987418693429145e-06, | |
| "loss": 1.7419, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.6527545909849749, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.976435467178592e-06, | |
| "loss": 1.801, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.654424040066778, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.965429950820222e-06, | |
| "loss": 1.7309, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.656093489148581, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.954402226772804e-06, | |
| "loss": 1.6584, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.657762938230384, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.943352377621414e-06, | |
| "loss": 1.8112, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.659432387312187, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.932280486116825e-06, | |
| "loss": 1.9561, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.66110183639399, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.92118663517488e-06, | |
| "loss": 1.7664, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.662771285475793, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.910070907875871e-06, | |
| "loss": 1.5639, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.664440734557596, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.898933387463924e-06, | |
| "loss": 1.401, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.666110183639399, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.887774157346365e-06, | |
| "loss": 1.5964, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.667779632721202, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.876593301093104e-06, | |
| "loss": 1.6001, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.669449081803005, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.865390902436005e-06, | |
| "loss": 1.7532, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.671118530884808, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.854167045268265e-06, | |
| "loss": 1.6746, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.672787979966611, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.842921813643767e-06, | |
| "loss": 1.9211, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.674457429048414, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.831655291776484e-06, | |
| "loss": 1.9482, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.676126878130217, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.82036756403981e-06, | |
| "loss": 1.8791, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.67779632721202, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.809058714965962e-06, | |
| "loss": 1.8097, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.679465776293823, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.797728829245321e-06, | |
| "loss": 1.6008, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.6811352253756261, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.786377991725813e-06, | |
| "loss": 1.5568, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.6828046744574291, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.775006287412268e-06, | |
| "loss": 1.5637, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.6844741235392321, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.763613801465785e-06, | |
| "loss": 1.8548, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.6861435726210351, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.752200619203094e-06, | |
| "loss": 1.6665, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.6878130217028381, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.740766826095918e-06, | |
| "loss": 1.8327, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.6894824707846411, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.729312507770326e-06, | |
| "loss": 1.7214, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.6911519198664441, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.717837750006106e-06, | |
| "loss": 1.4923, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.6928213689482471, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.706342638736108e-06, | |
| "loss": 1.9765, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.6944908180300501, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.694827260045608e-06, | |
| "loss": 1.8433, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.6961602671118531, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.683291700171663e-06, | |
| "loss": 1.7273, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.6978297161936561, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.671736045502462e-06, | |
| "loss": 1.9375, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.6994991652754591, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.660160382576683e-06, | |
| "loss": 1.8932, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.7011686143572621, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.648564798082842e-06, | |
| "loss": 1.8344, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7028380634390651, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.636949378858647e-06, | |
| "loss": 1.888, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.7045075125208681, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.625314211890342e-06, | |
| "loss": 1.6684, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.7061769616026711, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.613659384312062e-06, | |
| "loss": 1.6708, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.7078464106844741, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.601984983405173e-06, | |
| "loss": 1.619, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.7095158597662772, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.590291096597631e-06, | |
| "loss": 1.7182, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.7111853088480802, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.57857781146331e-06, | |
| "loss": 1.8545, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.7128547579298832, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.566845215721362e-06, | |
| "loss": 1.8239, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.7145242070116862, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.555093397235553e-06, | |
| "loss": 1.7683, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.7161936560934892, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.543322444013601e-06, | |
| "loss": 1.8621, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.7178631051752922, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.531532444206524e-06, | |
| "loss": 1.916, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7195325542570952, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.519723486107977e-06, | |
| "loss": 1.6987, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.7212020033388982, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.507895658153594e-06, | |
| "loss": 1.6218, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.7228714524207012, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.496049048920317e-06, | |
| "loss": 1.6171, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.7245409015025042, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.484183747125743e-06, | |
| "loss": 1.8034, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.7262103505843072, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.472299841627452e-06, | |
| "loss": 1.8408, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.7278797996661102, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.460397421422346e-06, | |
| "loss": 1.857, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.7295492487479132, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.448476575645982e-06, | |
| "loss": 1.6, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.7312186978297162, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.4365373935719e-06, | |
| "loss": 1.9061, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.7328881469115192, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.424579964610963e-06, | |
| "loss": 1.8586, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.7345575959933222, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.412604378310677e-06, | |
| "loss": 2.0079, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7362270450751253, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.400610724354531e-06, | |
| "loss": 1.7882, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.7378964941569283, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.388599092561315e-06, | |
| "loss": 1.8505, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.7395659432387313, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.376569572884457e-06, | |
| "loss": 1.5792, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.7412353923205343, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.364522255411342e-06, | |
| "loss": 1.7987, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.7429048414023373, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.3524572303626415e-06, | |
| "loss": 1.814, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.7445742904841403, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.340374588091638e-06, | |
| "loss": 1.54, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.7462437395659433, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.328274419083541e-06, | |
| "loss": 1.9338, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.7479131886477463, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.316156813954821e-06, | |
| "loss": 1.81, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.7495826377295493, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.304021863452525e-06, | |
| "loss": 1.7657, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.7512520868113522, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.291869658453594e-06, | |
| "loss": 1.8334, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7529215358931552, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.279700289964187e-06, | |
| "loss": 1.6198, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.7545909849749582, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.267513849119001e-06, | |
| "loss": 1.8972, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.7562604340567612, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.255310427180579e-06, | |
| "loss": 1.819, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.7579298831385642, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.243090115538639e-06, | |
| "loss": 1.5967, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.7595993322203672, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.230853005709378e-06, | |
| "loss": 1.8483, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.7612687813021702, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.218599189334799e-06, | |
| "loss": 1.7642, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.7629382303839732, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.206328758182013e-06, | |
| "loss": 1.6935, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.7646076794657763, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.194041804142556e-06, | |
| "loss": 1.6039, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.7662771285475793, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.181738419231708e-06, | |
| "loss": 1.4567, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.7679465776293823, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.169418695587791e-06, | |
| "loss": 1.5594, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.7696160267111853, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.157082725471488e-06, | |
| "loss": 1.8546, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.7712854757929883, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.144730601265148e-06, | |
| "loss": 1.8287, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.7729549248747913, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.132362415472099e-06, | |
| "loss": 1.6402, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.7746243739565943, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.1199782607159494e-06, | |
| "loss": 1.549, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.7762938230383973, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.107578229739895e-06, | |
| "loss": 1.5773, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.7779632721202003, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.095162415406034e-06, | |
| "loss": 2.0086, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.7796327212020033, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.082730910694655e-06, | |
| "loss": 1.635, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.7813021702838063, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.070283808703553e-06, | |
| "loss": 1.7928, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.7829716193656093, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.057821202647332e-06, | |
| "loss": 1.839, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.7846410684474123, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.045343185856701e-06, | |
| "loss": 1.6673, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.7863105175292153, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.032849851777774e-06, | |
| "loss": 1.7299, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.7879799666110183, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.020341293971383e-06, | |
| "loss": 1.8863, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.7896494156928213, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.0078176061123595e-06, | |
| "loss": 1.8666, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.7913188647746243, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.995278881988847e-06, | |
| "loss": 1.3643, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.7929883138564274, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.982725215501592e-06, | |
| "loss": 1.6456, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.7946577629382304, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.970156700663244e-06, | |
| "loss": 1.8537, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.7963272120200334, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.957573431597646e-06, | |
| "loss": 1.8776, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.7979966611018364, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.9449755025391355e-06, | |
| "loss": 1.8815, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.7996661101836394, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.932363007831837e-06, | |
| "loss": 1.8574, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.8013355592654424, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.919736041928956e-06, | |
| "loss": 1.8733, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.8030050083472454, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.907094699392066e-06, | |
| "loss": 1.7342, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.8046744574290484, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.894439074890413e-06, | |
| "loss": 1.6841, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.8063439065108514, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.881769263200192e-06, | |
| "loss": 1.6834, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.8080133555926544, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.869085359203844e-06, | |
| "loss": 1.8617, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.8096828046744574, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.8563874578893505e-06, | |
| "loss": 1.8515, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.8113522537562604, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.843675654349513e-06, | |
| "loss": 1.672, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.8130217028380634, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.830950043781245e-06, | |
| "loss": 1.8821, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.8146911519198664, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.818210721484859e-06, | |
| "loss": 1.9313, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.8163606010016694, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.805457782863354e-06, | |
| "loss": 1.8146, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.8180300500834724, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.792691323421698e-06, | |
| "loss": 1.6788, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.8196994991652755, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.779911438766117e-06, | |
| "loss": 1.8528, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.8213689482470785, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.767118224603374e-06, | |
| "loss": 1.6756, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.8230383973288815, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.754311776740057e-06, | |
| "loss": 1.4272, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.8247078464106845, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.741492191081856e-06, | |
| "loss": 1.7705, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.8263772954924875, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.728659563632853e-06, | |
| "loss": 1.5004, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.8280467445742905, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.715813990494793e-06, | |
| "loss": 2.0629, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.8297161936560935, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.702955567866372e-06, | |
| "loss": 2.0329, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.8313856427378965, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.690084392042514e-06, | |
| "loss": 1.7926, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.8330550918196995, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.677200559413652e-06, | |
| "loss": 1.6044, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.8347245409015025, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.664304166465e-06, | |
| "loss": 1.2786, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8363939899833055, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.651395309775837e-06, | |
| "loss": 1.8032, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.8380634390651085, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.638474086018778e-06, | |
| "loss": 1.807, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.8397328881469115, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.62554059195906e-06, | |
| "loss": 1.5938, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.8414023372287145, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.612594924453801e-06, | |
| "loss": 1.5486, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.8430717863105175, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.599637180451295e-06, | |
| "loss": 1.7077, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.8447412353923205, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.5866674569902676e-06, | |
| "loss": 1.4935, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.8464106844741235, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.5736858511991585e-06, | |
| "loss": 1.5449, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.8480801335559266, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.5606924602953925e-06, | |
| "loss": 1.3814, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.8497495826377296, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.547687381584653e-06, | |
| "loss": 2.0371, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.8514190317195326, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.534670712460151e-06, | |
| "loss": 1.8953, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.8530884808013356, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.521642550401894e-06, | |
| "loss": 1.63, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.8547579298831386, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.508602992975963e-06, | |
| "loss": 1.7057, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.8564273789649416, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.495552137833774e-06, | |
| "loss": 1.6038, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.8580968280467446, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.4824900827113506e-06, | |
| "loss": 1.6273, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.8597662771285476, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.469416925428593e-06, | |
| "loss": 1.8133, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.8614357262103506, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.456332763888544e-06, | |
| "loss": 1.6532, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.8631051752921536, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.443237696076652e-06, | |
| "loss": 1.9142, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.8647746243739566, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.430131820060043e-06, | |
| "loss": 1.7401, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.8664440734557596, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.417015233986786e-06, | |
| "loss": 1.5529, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.8681135225375626, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.403888036085155e-06, | |
| "loss": 1.5511, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.8697829716193656, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.390750324662895e-06, | |
| "loss": 1.7932, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.8714524207011686, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.3776021981064825e-06, | |
| "loss": 1.7507, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.8731218697829716, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.364443754880395e-06, | |
| "loss": 1.8488, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.8747913188647746, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.3512750935263664e-06, | |
| "loss": 1.4568, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.8764607679465777, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.338096312662658e-06, | |
| "loss": 1.8529, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.8781302170283807, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.32490751098331e-06, | |
| "loss": 1.6563, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.8797996661101837, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.311708787257408e-06, | |
| "loss": 1.6838, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.8814691151919867, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.298500240328342e-06, | |
| "loss": 1.7945, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.8831385642737897, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.285281969113072e-06, | |
| "loss": 1.4784, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.8848080133555927, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.272054072601374e-06, | |
| "loss": 1.7104, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.8864774624373957, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.258816649855109e-06, | |
| "loss": 1.5431, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.8881469115191987, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.245569800007484e-06, | |
| "loss": 1.5852, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.8898163606010017, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.232313622262297e-06, | |
| "loss": 1.7488, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.8914858096828047, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.219048215893204e-06, | |
| "loss": 1.6947, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.8931552587646077, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.2057736802429724e-06, | |
| "loss": 1.6526, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.8948247078464107, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.192490114722741e-06, | |
| "loss": 1.4266, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.8964941569282137, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.179197618811267e-06, | |
| "loss": 1.4984, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.8981636060100167, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.1658962920541875e-06, | |
| "loss": 1.7565, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.8998330550918197, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.152586234063277e-06, | |
| "loss": 1.6057, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.9015025041736227, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.139267544515689e-06, | |
| "loss": 1.4978, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9031719532554258, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.125940323153223e-06, | |
| "loss": 1.5634, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.9048414023372288, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.112604669781572e-06, | |
| "loss": 1.8368, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.9065108514190318, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.0992606842695745e-06, | |
| "loss": 1.4745, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.9081803005008348, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.0859084665484645e-06, | |
| "loss": 1.6676, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.9098497495826378, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.07254811661113e-06, | |
| "loss": 1.7473, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.9115191986644408, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.059179734511357e-06, | |
| "loss": 1.8217, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.9131886477462438, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.045803420363085e-06, | |
| "loss": 2.0666, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.9148580968280468, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.032419274339654e-06, | |
| "loss": 1.7439, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.9165275459098498, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.019027396673058e-06, | |
| "loss": 1.6459, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.9181969949916527, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.005627887653189e-06, | |
| "loss": 1.6537, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.9198664440734557, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.9922208476270914e-06, | |
| "loss": 1.4779, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.9215358931552587, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.978806376998209e-06, | |
| "loss": 1.6659, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.9232053422370617, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.965384576225632e-06, | |
| "loss": 1.8208, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.9248747913188647, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.951955545823342e-06, | |
| "loss": 1.7005, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.9265442404006677, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.938519386359466e-06, | |
| "loss": 1.7261, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.9282136894824707, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.925076198455517e-06, | |
| "loss": 1.8591, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.9298831385642737, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.911626082785644e-06, | |
| "loss": 1.9504, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.9315525876460768, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.898169140075878e-06, | |
| "loss": 1.8146, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.9332220367278798, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.884705471103376e-06, | |
| "loss": 1.7633, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.9348914858096828, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.871235176695664e-06, | |
| "loss": 1.7236, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.9365609348914858, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.857758357729892e-06, | |
| "loss": 1.9265, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.9382303839732888, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.844275115132064e-06, | |
| "loss": 1.5696, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.9398998330550918, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.830785549876296e-06, | |
| "loss": 1.8268, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.9415692821368948, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.817289762984048e-06, | |
| "loss": 1.5622, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.9432387312186978, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.803787855523377e-06, | |
| "loss": 1.7594, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.9449081803005008, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.790279928608173e-06, | |
| "loss": 1.6731, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.9465776293823038, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.776766083397409e-06, | |
| "loss": 1.5842, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.9482470784641068, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.763246421094373e-06, | |
| "loss": 1.8225, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.9499165275459098, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.749721042945924e-06, | |
| "loss": 1.591, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.9515859766277128, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.736190050241719e-06, | |
| "loss": 1.7563, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.9532554257095158, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.722653544313467e-06, | |
| "loss": 1.707, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.9549248747913188, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.709111626534161e-06, | |
| "loss": 1.8462, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.9565943238731218, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.695564398317326e-06, | |
| "loss": 1.6125, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.9582637729549248, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.6820119611162515e-06, | |
| "loss": 1.6431, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.9599332220367279, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.668454416423243e-06, | |
| "loss": 1.9529, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.9616026711185309, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.65489186576885e-06, | |
| "loss": 1.8668, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.9632721202003339, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.64132441072111e-06, | |
| "loss": 1.9664, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.9649415692821369, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.627752152884794e-06, | |
| "loss": 1.6759, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.9666110183639399, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.614175193900639e-06, | |
| "loss": 1.8378, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.9682804674457429, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.600593635444583e-06, | |
| "loss": 1.2707, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.9699499165275459, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.587007579227014e-06, | |
| "loss": 1.6216, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.9716193656093489, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.573417126992004e-06, | |
| "loss": 1.688, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.9732888146911519, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.559822380516539e-06, | |
| "loss": 1.8716, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.9749582637729549, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.546223441609775e-06, | |
| "loss": 1.938, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.9766277128547579, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.532620412112255e-06, | |
| "loss": 1.5521, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.9782971619365609, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.51901339389516e-06, | |
| "loss": 1.7355, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.9799666110183639, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.5054024888595415e-06, | |
| "loss": 1.3318, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.9816360601001669, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.491787798935557e-06, | |
| "loss": 1.7539, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.9833055091819699, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.478169426081712e-06, | |
| "loss": 1.8847, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.9849749582637729, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.464547472284091e-06, | |
| "loss": 1.6883, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.986644407345576, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.450922039555594e-06, | |
| "loss": 1.7047, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.988313856427379, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.437293229935178e-06, | |
| "loss": 1.5539, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.989983305509182, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.4236611454870865e-06, | |
| "loss": 1.6321, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.991652754590985, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.4100258883000874e-06, | |
| "loss": 1.7593, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.993322203672788, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.39638756048671e-06, | |
| "loss": 1.7527, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.994991652754591, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.38274626418248e-06, | |
| "loss": 1.3562, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.996661101836394, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.3691021015451494e-06, | |
| "loss": 1.4055, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.998330550918197, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.355455174753941e-06, | |
| "loss": 1.7539, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.341805586008778e-06, | |
| "loss": 1.7956, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.001669449081803, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.328153437529512e-06, | |
| "loss": 2.0003, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.003338898163606, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.31449883155517e-06, | |
| "loss": 1.7132, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.005008347245409, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.300841870343183e-06, | |
| "loss": 1.3768, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.006677796327212, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.287182656168618e-06, | |
| "loss": 1.7136, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.008347245409015, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.273521291323411e-06, | |
| "loss": 1.7554, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.010016694490818, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.259857878115611e-06, | |
| "loss": 1.2793, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.011686143572621, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.2461925188686035e-06, | |
| "loss": 1.7101, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.013355592654424, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.232525315920346e-06, | |
| "loss": 1.9689, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.015025041736227, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.218856371622605e-06, | |
| "loss": 1.7394, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.01669449081803, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.205185788340189e-06, | |
| "loss": 1.7766, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.018363939899833, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.191513668450178e-06, | |
| "loss": 1.7382, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.020033388981636, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.17784011434116e-06, | |
| "loss": 1.6687, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.021702838063439, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.1641652284124645e-06, | |
| "loss": 1.7971, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.023372287145242, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.150489113073394e-06, | |
| "loss": 1.9121, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.025041736227045, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.136811870742462e-06, | |
| "loss": 1.9295, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.026711185308848, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.123133603846613e-06, | |
| "loss": 1.9907, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.0283806343906512, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.109454414820475e-06, | |
| "loss": 1.7408, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.0300500834724542, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.095774406105572e-06, | |
| "loss": 1.9591, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.0317195325542572, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.0820936801495716e-06, | |
| "loss": 1.8002, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.0333889816360602, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.068412339405514e-06, | |
| "loss": 1.3725, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.0350584307178632, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.054730486331041e-06, | |
| "loss": 1.7209, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.0367278797996662, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.0410482233876275e-06, | |
| "loss": 1.5783, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.0383973288814692, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.0273656530398285e-06, | |
| "loss": 1.6146, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.0400667779632722, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.013682877754491e-06, | |
| "loss": 1.6253, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.0417362270450752, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5e-06, | |
| "loss": 1.6338, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.0434056761268782, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.986317122245508e-06, | |
| "loss": 1.7018, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.0450751252086812, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.972634346960173e-06, | |
| "loss": 1.9059, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.0467445742904842, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.9589517766123725e-06, | |
| "loss": 1.6677, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.0484140233722872, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.945269513668962e-06, | |
| "loss": 1.8649, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.0500834724540902, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.931587660594488e-06, | |
| "loss": 1.503, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.0517529215358932, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.917906319850431e-06, | |
| "loss": 1.6686, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.0534223706176962, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.90422559389443e-06, | |
| "loss": 1.4966, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.0550918196994992, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.890545585179527e-06, | |
| "loss": 2.0538, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.0567612687813022, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.876866396153388e-06, | |
| "loss": 1.458, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.0584307178631052, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.863188129257539e-06, | |
| "loss": 1.7656, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.0601001669449082, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.849510886926606e-06, | |
| "loss": 1.8316, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.0617696160267112, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.835834771587537e-06, | |
| "loss": 1.6928, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.0634390651085142, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.822159885658843e-06, | |
| "loss": 1.7323, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.0651085141903172, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.808486331549824e-06, | |
| "loss": 1.7975, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.0667779632721202, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.794814211659813e-06, | |
| "loss": 1.819, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.0684474123539232, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.781143628377396e-06, | |
| "loss": 1.4956, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.0701168614357262, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.767474684079655e-06, | |
| "loss": 1.767, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.0717863105175292, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.753807481131398e-06, | |
| "loss": 1.9048, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.0734557595993321, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.74014212188439e-06, | |
| "loss": 1.7265, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.0751252086811351, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.72647870867659e-06, | |
| "loss": 1.4845, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.0767946577629381, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.712817343831384e-06, | |
| "loss": 1.9606, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.0784641068447411, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.699158129656818e-06, | |
| "loss": 1.6695, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.0801335559265441, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.685501168444831e-06, | |
| "loss": 1.7123, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.0818030050083474, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.671846562470489e-06, | |
| "loss": 1.7025, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.0834724540901504, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.658194413991224e-06, | |
| "loss": 1.6985, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.0851419031719534, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.644544825246059e-06, | |
| "loss": 1.6561, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.0868113522537564, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.630897898454852e-06, | |
| "loss": 1.6168, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.0884808013355594, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.617253735817522e-06, | |
| "loss": 1.8791, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.0901502504173624, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.603612439513293e-06, | |
| "loss": 1.7602, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.0918196994991654, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.589974111699914e-06, | |
| "loss": 1.7051, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.0934891485809684, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.576338854512916e-06, | |
| "loss": 1.5697, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.0951585976627713, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.562706770064824e-06, | |
| "loss": 1.6602, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.0968280467445743, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.549077960444407e-06, | |
| "loss": 1.6243, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.0984974958263773, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.535452527715911e-06, | |
| "loss": 1.902, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.1001669449081803, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.521830573918289e-06, | |
| "loss": 1.8197, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.1018363939899833, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.508212201064446e-06, | |
| "loss": 1.8519, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.1035058430717863, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.49459751114046e-06, | |
| "loss": 1.9204, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.1051752921535893, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.480986606104842e-06, | |
| "loss": 1.3943, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.1068447412353923, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.467379587887747e-06, | |
| "loss": 1.6344, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.1085141903171953, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.453776558390225e-06, | |
| "loss": 1.8236, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.1101836393989983, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.4401776194834615e-06, | |
| "loss": 1.6879, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.1118530884808013, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.426582873007999e-06, | |
| "loss": 1.7198, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.1135225375626043, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.412992420772988e-06, | |
| "loss": 1.9099, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.1151919866444073, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.3994063645554185e-06, | |
| "loss": 1.9572, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.1168614357262103, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.385824806099364e-06, | |
| "loss": 1.9276, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.1185308848080133, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.3722478471152065e-06, | |
| "loss": 1.6846, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.1202003338898163, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.3586755892788896e-06, | |
| "loss": 1.5223, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.1218697829716193, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.345108134231152e-06, | |
| "loss": 1.9143, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.1235392320534223, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.331545583576758e-06, | |
| "loss": 1.3943, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.1252086811352253, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.31798803888375e-06, | |
| "loss": 1.6374, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.1268781302170283, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.304435601682675e-06, | |
| "loss": 1.9188, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.1285475792988313, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.290888373465841e-06, | |
| "loss": 1.9335, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.1302170283806343, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.277346455686535e-06, | |
| "loss": 1.5135, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.1318864774624373, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.2638099497582835e-06, | |
| "loss": 1.6542, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.1335559265442403, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.2502789570540775e-06, | |
| "loss": 1.429, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.1352253756260433, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.236753578905627e-06, | |
| "loss": 1.3608, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.1368948247078463, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.223233916602593e-06, | |
| "loss": 1.6528, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.1385642737896493, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.209720071391827e-06, | |
| "loss": 1.6002, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.1402337228714523, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.196212144476626e-06, | |
| "loss": 1.7644, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.1419031719532553, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.1827102370159526e-06, | |
| "loss": 1.5274, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.1435726210350585, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.169214450123706e-06, | |
| "loss": 1.5075, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.1452420701168615, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.155724884867937e-06, | |
| "loss": 1.6292, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.1469115191986645, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.142241642270109e-06, | |
| "loss": 1.752, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.1485809682804675, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.1287648233043366e-06, | |
| "loss": 1.7797, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.1502504173622705, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.115294528896627e-06, | |
| "loss": 1.8398, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.1519198664440735, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.1018308599241245e-06, | |
| "loss": 1.6796, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.1535893155258765, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.088373917214357e-06, | |
| "loss": 1.8816, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.1552587646076795, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.074923801544485e-06, | |
| "loss": 1.7933, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.1569282136894825, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.0614806136405355e-06, | |
| "loss": 1.6684, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.1585976627712855, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.048044454176658e-06, | |
| "loss": 1.3224, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.1602671118530885, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.034615423774369e-06, | |
| "loss": 1.787, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.1619365609348915, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.0211936230017915e-06, | |
| "loss": 1.6381, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.1636060100166945, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.00777915237291e-06, | |
| "loss": 1.7108, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.1652754590984975, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.9943721123468124e-06, | |
| "loss": 1.8407, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.1669449081803005, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.980972603326945e-06, | |
| "loss": 1.7378, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.1686143572621035, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.967580725660348e-06, | |
| "loss": 1.8234, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.1702838063439065, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.954196579636918e-06, | |
| "loss": 1.7271, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.1719532554257095, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.940820265488644e-06, | |
| "loss": 1.7063, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.1736227045075125, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.9274518833888704e-06, | |
| "loss": 1.6849, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.1752921535893155, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.914091533451537e-06, | |
| "loss": 1.7138, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.1769616026711185, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.900739315730426e-06, | |
| "loss": 1.8181, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.1786310517529215, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.887395330218429e-06, | |
| "loss": 1.6003, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.1803005008347245, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.874059676846778e-06, | |
| "loss": 1.775, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.1819699499165275, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.860732455484314e-06, | |
| "loss": 1.4583, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.1836393989983305, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.847413765936724e-06, | |
| "loss": 1.6664, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.1853088480801335, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.8341037079458125e-06, | |
| "loss": 1.7606, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.1869782971619365, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.820802381188735e-06, | |
| "loss": 1.4833, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.1886477462437395, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.8075098852772607e-06, | |
| "loss": 1.771, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.1903171953255425, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7942263197570297e-06, | |
| "loss": 1.6852, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.1919866444073457, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7809517841067976e-06, | |
| "loss": 1.8644, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.1936560934891487, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7676863777377055e-06, | |
| "loss": 1.7803, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.1953255425709517, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7544301999925176e-06, | |
| "loss": 1.8378, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.1969949916527547, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7411833501448924e-06, | |
| "loss": 1.7448, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.1986644407345577, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.727945927398628e-06, | |
| "loss": 1.5136, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.2003338898163607, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7147180308869296e-06, | |
| "loss": 1.9288, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.2020033388981637, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7014997596716596e-06, | |
| "loss": 1.606, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.2036727879799667, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.688291212742594e-06, | |
| "loss": 1.8199, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.2053422370617697, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.675092489016693e-06, | |
| "loss": 1.7053, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.2070116861435727, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.6619036873373435e-06, | |
| "loss": 1.7121, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.2086811352253757, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.6487249064736352e-06, | |
| "loss": 1.4813, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.2103505843071787, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.6355562451196065e-06, | |
| "loss": 1.7326, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.2120200333889817, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.622397801893518e-06, | |
| "loss": 1.6339, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.2136894824707847, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.6092496753371064e-06, | |
| "loss": 1.6062, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.2153589315525877, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.5961119639148443e-06, | |
| "loss": 1.784, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.2170283806343907, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.582984766013215e-06, | |
| "loss": 1.5648, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.2186978297161937, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.569868179939958e-06, | |
| "loss": 1.674, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.2203672787979967, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.556762303923351e-06, | |
| "loss": 1.6671, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.2220367278797997, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.543667236111458e-06, | |
| "loss": 1.4856, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.2237061769616027, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.530583074571407e-06, | |
| "loss": 1.7501, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.2253756260434057, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.5175099172886507e-06, | |
| "loss": 1.611, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.2270450751252087, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.504447862166227e-06, | |
| "loss": 2.1473, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.2287145242070117, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.4913970070240388e-06, | |
| "loss": 1.6351, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.2303839732888147, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.4783574495981075e-06, | |
| "loss": 1.7557, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.2320534223706177, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.4653292875398523e-06, | |
| "loss": 1.7185, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.2337228714524207, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.4523126184153483e-06, | |
| "loss": 1.6528, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.2353923205342237, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.4393075397046105e-06, | |
| "loss": 1.5345, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.2370617696160267, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.426314148800843e-06, | |
| "loss": 1.4437, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.2387312186978297, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.4133325430097337e-06, | |
| "loss": 1.8744, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.2404006677796326, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.400362819548706e-06, | |
| "loss": 1.625, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.2420701168614356, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.3874050755461984e-06, | |
| "loss": 1.8173, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.2437395659432386, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.374459408040942e-06, | |
| "loss": 1.9883, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.2454090150250416, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.3615259139812227e-06, | |
| "loss": 1.6616, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.2470784641068446, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.3486046902241663e-06, | |
| "loss": 1.5479, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.2487479131886476, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.335695833535001e-06, | |
| "loss": 1.943, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.2504173622704506, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.322799440586349e-06, | |
| "loss": 1.8239, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.2520868113522536, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.309915607957487e-06, | |
| "loss": 1.7892, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.2537562604340566, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2970444321336294e-06, | |
| "loss": 1.3755, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.2554257095158596, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2841860095052096e-06, | |
| "loss": 1.9403, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.2570951585976629, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.27134043636715e-06, | |
| "loss": 1.4275, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.2587646076794659, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.258507808918146e-06, | |
| "loss": 1.8757, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.2604340567612689, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.245688223259944e-06, | |
| "loss": 1.5396, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.2621035058430718, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.232881775396626e-06, | |
| "loss": 1.5604, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.2637729549248748, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2200885612338846e-06, | |
| "loss": 1.7815, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.2654424040066778, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2073086765783023e-06, | |
| "loss": 1.5858, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.2671118530884808, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1945422171366482e-06, | |
| "loss": 1.7903, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.2687813021702838, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1817892785151426e-06, | |
| "loss": 1.6163, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.2704507512520868, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1690499562187573e-06, | |
| "loss": 1.6819, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 1.2721202003338898, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1563243456504877e-06, | |
| "loss": 1.6804, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.2737896494156928, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1436125421106507e-06, | |
| "loss": 2.0157, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 1.2754590984974958, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1309146407961565e-06, | |
| "loss": 1.9822, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.2771285475792988, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.118230736799809e-06, | |
| "loss": 1.6525, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.2787979966611018, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1055609251095874e-06, | |
| "loss": 1.9026, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.2804674457429048, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0929053006079336e-06, | |
| "loss": 1.7447, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 1.2821368948247078, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0802639580710465e-06, | |
| "loss": 1.7471, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.2838063439065108, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.067636992168165e-06, | |
| "loss": 1.6875, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 1.2854757929883138, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0550244974608675e-06, | |
| "loss": 1.68, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.2871452420701168, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0424265684023556e-06, | |
| "loss": 1.6858, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 1.2888146911519198, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0298432993367577e-06, | |
| "loss": 1.9605, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.2904841402337228, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0172747844984098e-06, | |
| "loss": 1.6079, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.2921535893155258, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0047211180111537e-06, | |
| "loss": 1.6149, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.2938230383973288, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.9921823938876426e-06, | |
| "loss": 1.831, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.2954924874791318, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.979658706028619e-06, | |
| "loss": 1.5783, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.2971619365609348, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.9671501482222277e-06, | |
| "loss": 1.5288, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.298831385642738, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.9546568141433007e-06, | |
| "loss": 1.5322, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.300500834724541, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.9421787973526694e-06, | |
| "loss": 1.6689, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.302170283806344, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.9297161912964476e-06, | |
| "loss": 1.4969, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.303839732888147, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.917269089305347e-06, | |
| "loss": 1.7739, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.30550918196995, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.904837584593968e-06, | |
| "loss": 1.3092, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.307178631051753, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8924217702601048e-06, | |
| "loss": 1.6546, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.308848080133556, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.880021739284053e-06, | |
| "loss": 1.7141, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.310517529215359, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8676375845279013e-06, | |
| "loss": 1.9269, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.312186978297162, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8552693987348533e-06, | |
| "loss": 1.6174, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.313856427378965, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8429172745285127e-06, | |
| "loss": 1.672, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.315525876460768, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.83058130441221e-06, | |
| "loss": 1.8326, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.317195325542571, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8182615807682933e-06, | |
| "loss": 1.7942, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.318864774624374, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8059581958574434e-06, | |
| "loss": 1.6053, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.320534223706177, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.793671241817989e-06, | |
| "loss": 1.6082, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.32220367278798, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.781400810665201e-06, | |
| "loss": 1.7061, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.323873121869783, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.769146994290623e-06, | |
| "loss": 1.8037, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.325542570951586, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.7569098844613616e-06, | |
| "loss": 1.8475, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.327212020033389, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.744689572819421e-06, | |
| "loss": 1.854, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.328881469115192, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.7324861508810007e-06, | |
| "loss": 1.9131, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.330550918196995, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.7202997100358117e-06, | |
| "loss": 1.5491, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 1.332220367278798, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.708130341546407e-06, | |
| "loss": 1.4143, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.333889816360601, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.695978136547476e-06, | |
| "loss": 1.8685, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 1.335559265442404, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6838431860451797e-06, | |
| "loss": 1.7486, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.337228714524207, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6717255809164615e-06, | |
| "loss": 1.4641, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 1.33889816360601, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.659625411908366e-06, | |
| "loss": 1.6572, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.340567612687813, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6475427696373598e-06, | |
| "loss": 1.5809, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 1.342237061769616, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.635477744588658e-06, | |
| "loss": 1.5445, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.343906510851419, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6234304271155443e-06, | |
| "loss": 1.8947, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.345575959933222, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.611400907438685e-06, | |
| "loss": 1.8787, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.347245409015025, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.5993892756454702e-06, | |
| "loss": 1.9229, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 1.348914858096828, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.587395621689325e-06, | |
| "loss": 1.7786, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.350584307178631, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.57542003538904e-06, | |
| "loss": 1.9414, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 1.352253756260434, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.563462606428101e-06, | |
| "loss": 1.7625, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.353923205342237, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.5515234243540186e-06, | |
| "loss": 1.924, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 1.35559265442404, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.5396025785776545e-06, | |
| "loss": 1.8949, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 1.357262103505843, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.527700158372548e-06, | |
| "loss": 1.784, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 1.358931552587646, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.515816252874258e-06, | |
| "loss": 1.8006, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 1.360601001669449, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.5039509510796843e-06, | |
| "loss": 1.8571, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.362270450751252, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4921043418464085e-06, | |
| "loss": 1.5047, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 1.363939899833055, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4802765138920236e-06, | |
| "loss": 1.6812, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 1.365609348914858, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4684675557934766e-06, | |
| "loss": 1.8403, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 1.367278797996661, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.456677555986401e-06, | |
| "loss": 1.5725, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 1.3689482470784642, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4449066027644473e-06, | |
| "loss": 1.4778, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.3706176961602672, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.433154784278638e-06, | |
| "loss": 1.8657, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 1.3722871452420702, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4214221885366918e-06, | |
| "loss": 1.6946, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 1.3739565943238732, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4097089034023726e-06, | |
| "loss": 1.6673, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 1.3756260434056762, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.398015016594828e-06, | |
| "loss": 1.7846, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 1.3772954924874792, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.386340615687941e-06, | |
| "loss": 1.7574, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.3789649415692822, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.3746857881096586e-06, | |
| "loss": 1.6527, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 1.3806343906510852, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.363050621141354e-06, | |
| "loss": 1.8629, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 1.3823038397328882, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.351435201917159e-06, | |
| "loss": 1.7808, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 1.3839732888146912, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.339839617423318e-06, | |
| "loss": 1.6724, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 1.3856427378964942, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.32826395449754e-06, | |
| "loss": 1.4843, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.3873121869782972, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.316708299828338e-06, | |
| "loss": 1.7324, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 1.3889816360601002, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.3051727399543934e-06, | |
| "loss": 1.8341, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.3906510851419032, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2936573612638922e-06, | |
| "loss": 1.605, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 1.3923205342237062, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.282162249993895e-06, | |
| "loss": 1.6203, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 1.3939899833055092, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2706874922296756e-06, | |
| "loss": 1.6853, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.3956594323873122, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.259233173904084e-06, | |
| "loss": 1.6005, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.3973288814691152, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2477993807969074e-06, | |
| "loss": 1.7349, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 1.3989983305509182, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2363861985342156e-06, | |
| "loss": 1.7476, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 1.4006677796327212, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.224993712587734e-06, | |
| "loss": 1.7429, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 1.4023372287145242, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2136220082741876e-06, | |
| "loss": 1.6146, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.4040066777963272, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.20227117075468e-06, | |
| "loss": 1.7036, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 1.4056761268781301, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1909412850340395e-06, | |
| "loss": 1.4333, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 1.4073455759599331, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1796324359601896e-06, | |
| "loss": 1.9014, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 1.4090150250417361, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.168344708223519e-06, | |
| "loss": 1.7888, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 1.4106844741235394, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1570781863562328e-06, | |
| "loss": 1.726, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.4123539232053424, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1458329547317384e-06, | |
| "loss": 1.6039, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 1.4140233722871454, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.134609097563995e-06, | |
| "loss": 1.7142, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 1.4156928213689484, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1234066989068972e-06, | |
| "loss": 1.5549, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 1.4173622704507514, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1122258426536373e-06, | |
| "loss": 1.6325, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 1.4190317195325544, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1010666125360767e-06, | |
| "loss": 1.6929, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.4207011686143574, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.08992909212413e-06, | |
| "loss": 1.7992, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 1.4223706176961604, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0788133648251207e-06, | |
| "loss": 1.6507, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 1.4240400667779634, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.067719513883176e-06, | |
| "loss": 1.7738, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 1.4257095158597664, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0566476223785857e-06, | |
| "loss": 1.778, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 1.4273789649415694, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.045597773227199e-06, | |
| "loss": 1.9336, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.4290484140233723, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0345700491797786e-06, | |
| "loss": 1.642, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 1.4307178631051753, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0235645328214077e-06, | |
| "loss": 1.5683, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 1.4323873121869783, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0125813065708568e-06, | |
| "loss": 1.6782, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 1.4340567612687813, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.001620452679962e-06, | |
| "loss": 1.6515, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 1.4357262103505843, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9906820532330262e-06, | |
| "loss": 1.7665, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.4373956594323873, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.979766190146187e-06, | |
| "loss": 1.4809, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 1.4390651085141903, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9688729451668116e-06, | |
| "loss": 1.4875, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 1.4407345575959933, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9580023998728823e-06, | |
| "loss": 1.7605, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 1.4424040066777963, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.947154635672393e-06, | |
| "loss": 1.5669, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 1.4440734557595993, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9363297338027236e-06, | |
| "loss": 1.6956, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.4457429048414023, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.925527775330049e-06, | |
| "loss": 1.5113, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 1.4474123539232053, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9147488411487226e-06, | |
| "loss": 1.5448, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 1.4490818030050083, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9039930119806698e-06, | |
| "loss": 1.7054, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 1.4507512520868113, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8932603683747858e-06, | |
| "loss": 1.7207, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 1.4524207011686143, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8825509907063328e-06, | |
| "loss": 1.9051, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.4540901502504173, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8718649591763415e-06, | |
| "loss": 1.6744, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 1.4557595993322203, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8612023538109998e-06, | |
| "loss": 1.7186, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 1.4574290484140233, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8505632544610657e-06, | |
| "loss": 1.7795, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 1.4590984974958263, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8399477408012645e-06, | |
| "loss": 1.7003, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.4607679465776293, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8293558923296873e-06, | |
| "loss": 1.6446, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.4624373956594323, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8187877883672024e-06, | |
| "loss": 1.931, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 1.4641068447412353, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8082435080568556e-06, | |
| "loss": 1.8991, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 1.4657762938230383, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.797723130363288e-06, | |
| "loss": 1.6597, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 1.4674457429048413, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7872267340721289e-06, | |
| "loss": 1.6849, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 1.4691151919866443, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7767543977894198e-06, | |
| "loss": 1.6496, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.4707846410684473, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7663061999410209e-06, | |
| "loss": 1.6825, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 1.4724540901502503, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.755882218772018e-06, | |
| "loss": 1.3431, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.4741235392320533, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.745482532346145e-06, | |
| "loss": 1.8387, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 1.4757929883138563, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7351072185451934e-06, | |
| "loss": 1.857, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 1.4774624373956593, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7247563550684366e-06, | |
| "loss": 1.7855, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.4791318864774623, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7144300194320357e-06, | |
| "loss": 1.7008, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 1.4808013355592655, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7041282889684746e-06, | |
| "loss": 1.8663, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 1.4824707846410685, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6938512408259655e-06, | |
| "loss": 1.9139, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.4841402337228715, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6835989519678802e-06, | |
| "loss": 1.7092, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.4858096828046745, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6733714991721738e-06, | |
| "loss": 1.6333, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.4874791318864775, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6631689590308049e-06, | |
| "loss": 1.6067, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 1.4891485809682805, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.652991407949167e-06, | |
| "loss": 1.5876, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.4908180300500835, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6428389221455115e-06, | |
| "loss": 1.8228, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 1.4924874791318865, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6327115776503833e-06, | |
| "loss": 1.8408, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 1.4941569282136895, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.622609450306043e-06, | |
| "loss": 1.7807, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.4958263772954925, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6125326157659048e-06, | |
| "loss": 1.7149, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 1.4974958263772955, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6024811494939723e-06, | |
| "loss": 1.9643, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 1.4991652754590985, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5924551267642641e-06, | |
| "loss": 1.6214, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 1.5008347245409015, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5824546226602611e-06, | |
| "loss": 1.9501, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 1.5025041736227045, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.572479712074333e-06, | |
| "loss": 1.726, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.5041736227045075, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5625304697071897e-06, | |
| "loss": 1.7017, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 1.5058430717863105, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5526069700673108e-06, | |
| "loss": 1.5422, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 1.5075125208681135, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.542709287470393e-06, | |
| "loss": 1.6978, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 1.5091819699499165, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.532837496038792e-06, | |
| "loss": 1.7853, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 1.5108514190317195, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5229916697009706e-06, | |
| "loss": 1.5383, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.5125208681135225, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5131718821909435e-06, | |
| "loss": 1.6953, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 1.5141903171953257, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5033782070477192e-06, | |
| "loss": 1.6438, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 1.5158597662771287, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4936107176147606e-06, | |
| "loss": 1.8139, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 1.5175292153589317, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.483869487039425e-06, | |
| "loss": 1.5901, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 1.5191986644407347, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4741545882724213e-06, | |
| "loss": 1.5955, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.5208681135225377, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4644660940672628e-06, | |
| "loss": 1.5351, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 1.5225375626043407, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4548040769797255e-06, | |
| "loss": 1.7605, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 1.5242070116861437, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4451686093673028e-06, | |
| "loss": 1.6614, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 1.5258764607679467, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4355597633886576e-06, | |
| "loss": 1.8486, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 1.5275459098497497, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.425977611003091e-06, | |
| "loss": 1.7763, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.5292153589315527, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.416422223969996e-06, | |
| "loss": 1.7033, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 1.5308848080133557, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4068936738483302e-06, | |
| "loss": 1.5128, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 1.5325542570951587, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3973920319960654e-06, | |
| "loss": 1.8173, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 1.5342237061769617, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3879173695696668e-06, | |
| "loss": 1.7077, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 1.5358931552587647, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.378469757523554e-06, | |
| "loss": 1.6683, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.5375626043405677, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3690492666095672e-06, | |
| "loss": 1.6297, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 1.5392320534223707, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3596559673764421e-06, | |
| "loss": 1.5391, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 1.5409015025041737, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.350289930169278e-06, | |
| "loss": 1.7279, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 1.5425709515859767, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3409512251290164e-06, | |
| "loss": 1.9615, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 1.5442404006677797, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3316399221919075e-06, | |
| "loss": 1.5777, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.5459098497495827, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.322356091088996e-06, | |
| "loss": 1.6915, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 1.5475792988313857, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3130998013455875e-06, | |
| "loss": 1.6709, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 1.5492487479131887, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.303871122280742e-06, | |
| "loss": 1.5226, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 1.5509181969949917, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2946701230067405e-06, | |
| "loss": 1.736, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 1.5525876460767947, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2854968724285755e-06, | |
| "loss": 1.3861, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.5542570951585977, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.276351439243436e-06, | |
| "loss": 1.824, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 1.5559265442404007, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2672338919401866e-06, | |
| "loss": 1.8992, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 1.5575959933222037, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2581442987988635e-06, | |
| "loss": 1.6777, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 1.5592654424040067, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2490827278901513e-06, | |
| "loss": 1.5996, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 1.5609348914858097, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2400492470748877e-06, | |
| "loss": 1.4484, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.5626043405676127, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2310439240035415e-06, | |
| "loss": 1.6822, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 1.5642737896494157, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2220668261157132e-06, | |
| "loss": 1.6954, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 1.5659432387312187, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2131180206396331e-06, | |
| "loss": 1.88, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 1.5676126878130217, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2041975745916474e-06, | |
| "loss": 1.7492, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 1.5692821368948247, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.195305554775728e-06, | |
| "loss": 1.5302, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.5709515859766277, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.186442027782964e-06, | |
| "loss": 1.7966, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 1.5726210350584306, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.177607059991065e-06, | |
| "loss": 1.5941, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 1.5742904841402336, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1688007175638655e-06, | |
| "loss": 1.5734, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 1.5759599332220366, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1600230664508288e-06, | |
| "loss": 1.8075, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 1.5776293823038396, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1512741723865562e-06, | |
| "loss": 2.1487, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.5792988313856426, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1425541008902852e-06, | |
| "loss": 1.7166, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 1.5809682804674456, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.133862917265411e-06, | |
| "loss": 1.7968, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 1.5826377295492486, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1252006865989868e-06, | |
| "loss": 1.7527, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 1.5843071786310516, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1165674737612447e-06, | |
| "loss": 1.7028, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 1.5859766277128546, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.107963343405103e-06, | |
| "loss": 1.6813, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.5876460767946576, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0993883599656885e-06, | |
| "loss": 1.6206, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 1.5893155258764606, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0908425876598512e-06, | |
| "loss": 1.7438, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 1.5909849749582636, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0823260904856791e-06, | |
| "loss": 1.3356, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 1.5926544240400666, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0738389322220276e-06, | |
| "loss": 1.7507, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 1.5943238731218696, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0653811764280336e-06, | |
| "loss": 1.9094, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.5959933222036726, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0569528864426444e-06, | |
| "loss": 1.7581, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 1.5976627712854758, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0485541253841391e-06, | |
| "loss": 1.734, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 1.5993322203672788, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0401849561496647e-06, | |
| "loss": 1.7455, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 1.6010016694490818, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.031845441414756e-06, | |
| "loss": 1.6586, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 1.6026711185308848, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0235356436328675e-06, | |
| "loss": 1.5319, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.6043405676126878, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.015255625034911e-06, | |
| "loss": 1.6651, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 1.6060100166944908, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.007005447628785e-06, | |
| "loss": 1.4794, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 1.6076794657762938, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.987851731989096e-07, | |
| "loss": 1.9092, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 1.6093489148580968, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.905948633057666e-07, | |
| "loss": 1.6746, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 1.6110183639398998, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.82434579285441e-07, | |
| "loss": 1.7691, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.6126878130217028, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.743043822491528e-07, | |
| "loss": 1.8323, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 1.6143572621035058, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.662043330828086e-07, | |
| "loss": 1.7632, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 1.6160267111853088, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.58134492446543e-07, | |
| "loss": 1.6031, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 1.6176961602671118, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.500949207742566e-07, | |
| "loss": 1.9118, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 1.6193656093489148, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.420856782731774e-07, | |
| "loss": 1.7467, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.6210350584307178, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.341068249233964e-07, | |
| "loss": 1.7573, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 1.6227045075125208, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.26158420477431e-07, | |
| "loss": 1.757, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 1.6243739565943238, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.182405244597647e-07, | |
| "loss": 1.7199, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 1.626043405676127, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.10353196166412e-07, | |
| "loss": 1.5325, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 1.62771285475793, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.024964946644682e-07, | |
| "loss": 1.6199, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.629382303839733, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.946704787916676e-07, | |
| "loss": 1.6902, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 1.631051752921536, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.868752071559478e-07, | |
| "loss": 1.3334, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 1.632721202003339, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.791107381350028e-07, | |
| "loss": 1.9713, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 1.634390651085142, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.713771298758539e-07, | |
| "loss": 1.6646, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 1.636060100166945, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.636744402944075e-07, | |
| "loss": 1.8279, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.637729549248748, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.560027270750276e-07, | |
| "loss": 1.7358, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 1.639398998330551, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.483620476700977e-07, | |
| "loss": 2.0311, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 1.641068447412354, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.40752459299593e-07, | |
| "loss": 1.5454, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 1.642737896494157, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.33174018950656e-07, | |
| "loss": 1.782, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 1.64440734557596, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.256267833771608e-07, | |
| "loss": 1.6006, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.646076794657763, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.181108090993001e-07, | |
| "loss": 1.6482, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 1.647746243739566, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.106261524031473e-07, | |
| "loss": 1.6951, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 1.649415692821369, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.031728693402502e-07, | |
| "loss": 1.7212, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 1.651085141903172, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.957510157271991e-07, | |
| "loss": 1.7513, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 1.652754590984975, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.883606471452138e-07, | |
| "loss": 1.649, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.654424040066778, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.810018189397301e-07, | |
| "loss": 1.5066, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 1.656093489148581, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.736745862199785e-07, | |
| "loss": 1.6172, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 1.657762938230384, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.663790038585794e-07, | |
| "loss": 1.7814, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 1.659432387312187, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.59115126491124e-07, | |
| "loss": 1.5252, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 1.66110183639399, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.518830085157735e-07, | |
| "loss": 1.9539, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.662771285475793, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.446827040928439e-07, | |
| "loss": 1.7377, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 1.664440734557596, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.375142671444046e-07, | |
| "loss": 1.8891, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 1.666110183639399, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.303777513538762e-07, | |
| "loss": 1.9226, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 1.667779632721202, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.232732101656231e-07, | |
| "loss": 1.5819, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 1.669449081803005, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.162006967845602e-07, | |
| "loss": 1.7022, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.671118530884808, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.091602641757467e-07, | |
| "loss": 1.5175, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 1.672787979966611, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.021519650639952e-07, | |
| "loss": 1.8913, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 1.674457429048414, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.951758519334745e-07, | |
| "loss": 1.801, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 1.676126878130217, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.882319770273193e-07, | |
| "loss": 1.6252, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 1.67779632721202, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.813203923472328e-07, | |
| "loss": 1.6969, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.679465776293823, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.744411496531045e-07, | |
| "loss": 1.8072, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 1.681135225375626, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.67594300462619e-07, | |
| "loss": 1.6769, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 1.682804674457429, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.607798960508693e-07, | |
| "loss": 1.8645, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 1.684474123539232, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.539979874499747e-07, | |
| "loss": 1.8879, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 1.686143572621035, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.472486254486954e-07, | |
| "loss": 1.5467, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.687813021702838, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.405318605920602e-07, | |
| "loss": 1.5909, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 1.689482470784641, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.338477431809764e-07, | |
| "loss": 1.738, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 1.691151919866444, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.271963232718631e-07, | |
| "loss": 1.914, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 1.692821368948247, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.205776506762729e-07, | |
| "loss": 1.6797, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 1.69449081803005, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.139917749605151e-07, | |
| "loss": 1.7716, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.696160267111853, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.074387454452891e-07, | |
| "loss": 1.8465, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 1.697829716193656, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.009186112053134e-07, | |
| "loss": 2.0192, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 1.699499165275459, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.944314210689611e-07, | |
| "loss": 2.0092, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 1.701168614357262, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.879772236178871e-07, | |
| "loss": 1.9489, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 1.702838063439065, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.815560671866721e-07, | |
| "loss": 1.6819, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.704507512520868, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.751679998624571e-07, | |
| "loss": 1.3891, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 1.706176961602671, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.688130694845817e-07, | |
| "loss": 1.8861, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 1.707846410684474, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.624913236442287e-07, | |
| "loss": 1.441, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 1.7095158597662772, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.562028096840638e-07, | |
| "loss": 1.4711, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 1.7111853088480802, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.499475746978899e-07, | |
| "loss": 1.7767, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.7128547579298832, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.437256655302814e-07, | |
| "loss": 1.5217, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 1.7145242070116862, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.37537128776246e-07, | |
| "loss": 1.8079, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 1.7161936560934892, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.313820107808665e-07, | |
| "loss": 1.9587, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 1.7178631051752922, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.25260357638957e-07, | |
| "loss": 1.8453, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 1.7195325542570952, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.191722151947227e-07, | |
| "loss": 1.6909, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.7212020033388982, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.131176290414053e-07, | |
| "loss": 1.6264, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 1.7228714524207012, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.07096644520954e-07, | |
| "loss": 1.5964, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 1.7245409015025042, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.011093067236756e-07, | |
| "loss": 1.9016, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 1.7262103505843072, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.951556604879049e-07, | |
| "loss": 1.7335, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 1.7278797996661102, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.892357503996625e-07, | |
| "loss": 1.5909, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.7295492487479132, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.83349620792325e-07, | |
| "loss": 1.633, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 1.7312186978297162, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.77497315746292e-07, | |
| "loss": 1.7341, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 1.7328881469115192, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.716788790886545e-07, | |
| "loss": 1.632, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 1.7345575959933222, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.658943543928707e-07, | |
| "loss": 1.7613, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 1.7362270450751254, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.601437849784318e-07, | |
| "loss": 1.8181, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.7378964941569284, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.544272139105488e-07, | |
| "loss": 1.6785, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 1.7395659432387314, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.487446839998194e-07, | |
| "loss": 1.6502, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 1.7412353923205344, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.4309623780191214e-07, | |
| "loss": 1.919, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 1.7429048414023374, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.374819176172501e-07, | |
| "loss": 1.5251, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 1.7445742904841404, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.319017654906887e-07, | |
| "loss": 1.8641, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.7462437395659434, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.263558232112064e-07, | |
| "loss": 1.9693, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 1.7479131886477464, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.2084413231158473e-07, | |
| "loss": 1.5452, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 1.7495826377295494, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.153667340681067e-07, | |
| "loss": 1.8121, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 1.7512520868113524, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.099236695002379e-07, | |
| "loss": 1.8385, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 1.7529215358931554, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.045149793703257e-07, | |
| "loss": 1.9422, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.7545909849749584, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.9914070418329123e-07, | |
| "loss": 1.849, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 1.7562604340567614, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.938008841863289e-07, | |
| "loss": 1.712, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 1.7579298831385644, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.8849555936860296e-07, | |
| "loss": 1.5905, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 1.7595993322203674, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.832247694609442e-07, | |
| "loss": 1.8678, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 1.7612687813021703, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.779885539355621e-07, | |
| "loss": 1.665, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 1.7629382303839733, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7278695200573754e-07, | |
| "loss": 1.5783, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 1.7646076794657763, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.6762000262554e-07, | |
| "loss": 2.0622, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 1.7662771285475793, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.6248774448952695e-07, | |
| "loss": 1.6632, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 1.7679465776293823, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.5739021603246104e-07, | |
| "loss": 1.6519, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 1.7696160267111853, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.52327455429019e-07, | |
| "loss": 1.9502, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.7712854757929883, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.472995005935037e-07, | |
| "loss": 1.6183, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 1.7729549248747913, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.423063891795647e-07, | |
| "loss": 1.6387, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 1.7746243739565943, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.3734815857991155e-07, | |
| "loss": 1.5453, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 1.7762938230383973, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.324248459260393e-07, | |
| "loss": 1.8939, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 1.7779632721202003, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2753648808794505e-07, | |
| "loss": 1.8887, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.7796327212020033, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2268312167385687e-07, | |
| "loss": 1.7644, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 1.7813021702838063, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1786478302995305e-07, | |
| "loss": 1.5679, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 1.7829716193656093, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.1308150824009785e-07, | |
| "loss": 1.672, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 1.7846410684474123, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0833333312556446e-07, | |
| "loss": 1.6457, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 1.7863105175292153, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0362029324477015e-07, | |
| "loss": 1.4934, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.7879799666110183, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.9894242389301053e-07, | |
| "loss": 1.946, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 1.7896494156928213, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.942997601021924e-07, | |
| "loss": 1.7434, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 1.7913188647746243, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.896923366405746e-07, | |
| "loss": 1.7937, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 1.7929883138564273, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.851201880125043e-07, | |
| "loss": 2.0654, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 1.7946577629382303, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8058334845816214e-07, | |
| "loss": 1.8588, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.7963272120200333, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.760818519533037e-07, | |
| "loss": 1.9593, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 1.7979966611018363, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.716157322090041e-07, | |
| "loss": 1.8096, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 1.7996661101836393, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6718502267140844e-07, | |
| "loss": 1.3897, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 1.8013355592654423, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6278975652147875e-07, | |
| "loss": 1.6728, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 1.8030050083472453, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.584299666747475e-07, | |
| "loss": 1.6031, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.8046744574290483, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.541056857810681e-07, | |
| "loss": 1.6506, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 1.8063439065108513, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4981694622437546e-07, | |
| "loss": 1.6066, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 1.8080133555926543, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4556378012243807e-07, | |
| "loss": 1.5126, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 1.8096828046744573, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4134621932661916e-07, | |
| "loss": 1.6942, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 1.8113522537562603, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.3716429542164244e-07, | |
| "loss": 1.7853, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 1.8130217028380633, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.330180397253473e-07, | |
| "loss": 1.7304, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 1.8146911519198663, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.28907483288463e-07, | |
| "loss": 1.5907, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 1.8163606010016693, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2483265689436929e-07, | |
| "loss": 1.9272, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 1.8180300500834723, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2079359105886989e-07, | |
| "loss": 1.8972, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 1.8196994991652755, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.167903160299617e-07, | |
| "loss": 1.4341, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.8213689482470785, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1282286178761046e-07, | |
| "loss": 1.7007, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 1.8230383973288815, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0889125804352595e-07, | |
| "loss": 1.3616, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 1.8247078464106845, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.049955342409349e-07, | |
| "loss": 1.7606, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 1.8263772954924875, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0113571955436895e-07, | |
| "loss": 1.4619, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 1.8280467445742905, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9731184288943772e-07, | |
| "loss": 1.5691, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 1.8297161936560935, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9352393288261717e-07, | |
| "loss": 1.6799, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 1.8313856427378965, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8977201790103428e-07, | |
| "loss": 1.4127, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 1.8330550918196995, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8605612604225388e-07, | |
| "loss": 1.811, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 1.8347245409015025, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8237628513407046e-07, | |
| "loss": 1.8761, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 1.8363939899833055, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.787325227342951e-07, | |
| "loss": 1.7312, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.8380634390651085, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.751248661305538e-07, | |
| "loss": 1.6336, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 1.8397328881469115, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.715533423400817e-07, | |
| "loss": 1.9454, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 1.8414023372287145, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.680179781095187e-07, | |
| "loss": 1.6344, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 1.8430717863105175, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6451879991471186e-07, | |
| "loss": 1.3738, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 1.8447412353923205, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.610558339605156e-07, | |
| "loss": 1.5461, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.8464106844741235, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.576291061805979e-07, | |
| "loss": 1.939, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 1.8480801335559267, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.542386422372405e-07, | |
| "loss": 1.5495, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 1.8497495826377297, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5088446752115403e-07, | |
| "loss": 1.7424, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 1.8514190317195327, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4756660715128267e-07, | |
| "loss": 1.4635, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 1.8530884808013357, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4428508597461587e-07, | |
| "loss": 1.8355, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.8547579298831387, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4103992856600634e-07, | |
| "loss": 1.4809, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 1.8564273789649417, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.378311592279835e-07, | |
| "loss": 1.5775, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 1.8580968280467447, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.346588019905698e-07, | |
| "loss": 1.7013, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 1.8597662771285477, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3152288061110518e-07, | |
| "loss": 1.8002, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 1.8614357262103507, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.284234185740635e-07, | |
| "loss": 1.6495, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.8631051752921537, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.253604390908819e-07, | |
| "loss": 1.7394, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 1.8647746243739567, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2233396509978513e-07, | |
| "loss": 1.937, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 1.8664440734557597, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.193440192656109e-07, | |
| "loss": 1.7579, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 1.8681135225375627, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.163906239796453e-07, | |
| "loss": 1.9714, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 1.8697829716193657, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1347380135945108e-07, | |
| "loss": 1.576, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.8714524207011687, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1059357324870456e-07, | |
| "loss": 1.7355, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 1.8731218697829717, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0774996121702907e-07, | |
| "loss": 1.9441, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 1.8747913188647747, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.049429865598367e-07, | |
| "loss": 1.829, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 1.8764607679465777, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0217267029816736e-07, | |
| "loss": 1.827, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 1.8781302170283807, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.943903317853055e-08, | |
| "loss": 1.6159, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.8797996661101837, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.674209567275161e-08, | |
| "loss": 1.6928, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 1.8814691151919867, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.408187797781743e-08, | |
| "loss": 1.7798, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 1.8831385642737897, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.145840001572537e-08, | |
| "loss": 1.6211, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 1.8848080133555927, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.887168143333402e-08, | |
| "loss": 1.6973, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 1.8864774624373957, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.632174160221496e-08, | |
| "loss": 1.893, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.8881469115191987, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.380859961851174e-08, | |
| "loss": 1.6941, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 1.8898163606010017, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.133227430279055e-08, | |
| "loss": 1.6935, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 1.8914858096828047, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.889278419990598e-08, | |
| "loss": 1.7137, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 1.8931552587646077, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.649014757885597e-08, | |
| "loss": 1.5314, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 1.8948247078464107, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.41243824326504e-08, | |
| "loss": 1.8015, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.8964941569282137, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.179550647817224e-08, | |
| "loss": 1.9739, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 1.8981636060100167, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.950353715604597e-08, | |
| "loss": 1.7966, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 1.8998330550918197, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.724849163050995e-08, | |
| "loss": 1.5907, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 1.9015025041736227, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.50303867892832e-08, | |
| "loss": 1.5096, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 1.9031719532554257, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.28492392434421e-08, | |
| "loss": 1.5267, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.9048414023372287, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.070506532729393e-08, | |
| "loss": 2.113, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 1.9065108514190316, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.8597881098257924e-08, | |
| "loss": 1.7329, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 1.9081803005008346, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.652770233673943e-08, | |
| "loss": 1.512, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 1.9098497495826376, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.4494544546018216e-08, | |
| "loss": 1.7529, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 1.9115191986644406, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.249842295212748e-08, | |
| "loss": 1.6867, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 1.9131886477462436, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.0539352503741756e-08, | |
| "loss": 1.5203, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 1.9148580968280466, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.861734787206529e-08, | |
| "loss": 1.7982, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 1.9165275459098496, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.67324234507216e-08, | |
| "loss": 1.5475, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 1.9181969949916526, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.48845933556441e-08, | |
| "loss": 1.6388, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 1.9198664440734556, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.307387142497399e-08, | |
| "loss": 1.6519, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.9215358931552586, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.130027121895419e-08, | |
| "loss": 1.7686, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 1.9232053422370616, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.956380601982668e-08, | |
| "loss": 1.7067, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 1.9248747913188646, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.786448883173755e-08, | |
| "loss": 1.5475, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 1.9265442404006676, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.620233238063375e-08, | |
| "loss": 1.8219, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 1.9282136894824706, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.45773491141721e-08, | |
| "loss": 1.7498, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 1.9298831385642736, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2989551201624836e-08, | |
| "loss": 1.9549, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 1.9315525876460768, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.143895053378698e-08, | |
| "loss": 1.6767, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 1.9332220367278798, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.992555872289082e-08, | |
| "loss": 1.4431, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 1.9348914858096828, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.844938710251377e-08, | |
| "loss": 1.519, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 1.9365609348914858, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.7010446727498974e-08, | |
| "loss": 1.8625, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.9382303839732888, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.5608748373869285e-08, | |
| "loss": 1.7547, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 1.9398998330550918, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4244302538746765e-08, | |
| "loss": 1.8016, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 1.9415692821368948, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2917119440275524e-08, | |
| "loss": 1.5698, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 1.9432387312186978, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.162720901754234e-08, | |
| "loss": 1.6776, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 1.9449081803005008, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.037458093050726e-08, | |
| "loss": 1.5975, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 1.9465776293823038, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9159244559924795e-08, | |
| "loss": 2.009, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 1.9482470784641068, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7981209007278956e-08, | |
| "loss": 1.6987, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 1.9499165275459098, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6840483094713867e-08, | |
| "loss": 1.8003, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 1.9515859766277128, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.573707536496494e-08, | |
| "loss": 1.7123, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 1.9532554257095158, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4670994081297796e-08, | |
| "loss": 1.4862, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.9549248747913188, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3642247227446114e-08, | |
| "loss": 1.5492, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 1.9565943238731218, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2650842507550554e-08, | |
| "loss": 1.7895, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 1.9582637729549248, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1696787346102134e-08, | |
| "loss": 1.747, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 1.959933222036728, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.078008888788673e-08, | |
| "loss": 1.734, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 1.961602671118531, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.900753997929557e-09, | |
| "loss": 1.4903, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.963272120200334, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.058789261446876e-09, | |
| "loss": 1.71, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 1.964941569282137, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.254200983794369e-09, | |
| "loss": 1.4663, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 1.96661101836394, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.48699519042051e-09, | |
| "loss": 1.807, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 1.968280467445743, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.757177626822709e-09, | |
| "loss": 1.5051, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 1.969949916527546, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.0647537585017956e-09, | |
| "loss": 1.3409, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.971619365609349, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.409728770923162e-09, | |
| "loss": 1.7744, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 1.973288814691152, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.792107569476789e-09, | |
| "loss": 1.7346, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 1.974958263772955, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.211894779441727e-09, | |
| "loss": 1.5221, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 1.976627712854758, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.669094745950008e-09, | |
| "loss": 1.795, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 1.978297161936561, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.16371153395445e-09, | |
| "loss": 1.5513, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 1.979966611018364, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6957489281997927e-09, | |
| "loss": 1.687, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 1.981636060100167, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2652104331921664e-09, | |
| "loss": 1.7631, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 1.98330550918197, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.8720992731741104e-09, | |
| "loss": 1.5843, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 1.984974958263773, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5164183920995946e-09, | |
| "loss": 1.6599, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 1.986644407345576, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1981704536129234e-09, | |
| "loss": 1.8012, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.988313856427379, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.173578410281992e-10, | |
| "loss": 1.8889, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 1.989983305509182, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.739826573121111e-10, | |
| "loss": 1.7534, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 1.991652754590985, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.680467250672837e-10, | |
| "loss": 1.8212, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 1.993322203672788, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.995515865183984e-10, | |
| "loss": 1.6632, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 1.994991652754591, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6849850350275643e-10, | |
| "loss": 1.7037, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 1.996661101836394, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.48884574575115e-11, | |
| "loss": 1.8285, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 1.998330550918197, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.872214941633921e-11, | |
| "loss": 1.4735, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.8333, | |
| "step": 1198 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1198, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 300, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.3485232050772378e+19, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |