| { | |
| "best_global_step": 1500, | |
| "best_metric": 0.8288407325744629, | |
| "best_model_checkpoint": "output/reasoning-model_v2/checkpoint-1500", | |
| "epoch": 0.13859373556315255, | |
| "eval_steps": 500, | |
| "global_step": 1500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0004619791185438418, | |
| "grad_norm": 10.1875, | |
| "learning_rate": 2.4615384615384616e-07, | |
| "loss": 0.9733, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0009239582370876836, | |
| "grad_norm": 7.46875, | |
| "learning_rate": 5.53846153846154e-07, | |
| "loss": 1.0462, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0013859373556315254, | |
| "grad_norm": 10.0625, | |
| "learning_rate": 8.615384615384616e-07, | |
| "loss": 0.91, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0018479164741753672, | |
| "grad_norm": 9.5, | |
| "learning_rate": 1.1692307692307693e-06, | |
| "loss": 0.9938, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.002309895592719209, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.476923076923077e-06, | |
| "loss": 0.922, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0027718747112630508, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.7846153846153846e-06, | |
| "loss": 0.9497, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0032338538298068925, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 2.0923076923076926e-06, | |
| "loss": 0.9382, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0036958329483507343, | |
| "grad_norm": 5.75, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.9658, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004157812066894576, | |
| "grad_norm": 6.65625, | |
| "learning_rate": 2.7076923076923076e-06, | |
| "loss": 0.9687, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004619791185438418, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 3.0153846153846154e-06, | |
| "loss": 0.9775, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.00508177030398226, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 3.323076923076923e-06, | |
| "loss": 0.9061, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0055437494225261015, | |
| "grad_norm": 5.3125, | |
| "learning_rate": 3.630769230769231e-06, | |
| "loss": 0.9917, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.006005728541069943, | |
| "grad_norm": 4.09375, | |
| "learning_rate": 3.938461538461539e-06, | |
| "loss": 0.7981, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006467707659613785, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 4.246153846153846e-06, | |
| "loss": 0.9878, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.006929686778157627, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 4.553846153846154e-06, | |
| "loss": 0.9152, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.007391665896701469, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 4.861538461538462e-06, | |
| "loss": 1.0513, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.007853645015245311, | |
| "grad_norm": 6.40625, | |
| "learning_rate": 5.16923076923077e-06, | |
| "loss": 0.9603, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.008315624133789152, | |
| "grad_norm": 5.625, | |
| "learning_rate": 5.476923076923077e-06, | |
| "loss": 0.9008, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.008777603252332995, | |
| "grad_norm": 6.0, | |
| "learning_rate": 5.784615384615385e-06, | |
| "loss": 0.9536, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.009239582370876836, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 6.092307692307693e-06, | |
| "loss": 0.9726, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009701561489420679, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.0662, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.01016354060796452, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 6.707692307692308e-06, | |
| "loss": 0.8922, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.010625519726508362, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 7.015384615384616e-06, | |
| "loss": 0.8976, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.011087498845052203, | |
| "grad_norm": 4.71875, | |
| "learning_rate": 7.323076923076924e-06, | |
| "loss": 0.7963, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.011549477963596046, | |
| "grad_norm": 5.53125, | |
| "learning_rate": 7.630769230769232e-06, | |
| "loss": 0.8124, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.012011457082139887, | |
| "grad_norm": 8.3125, | |
| "learning_rate": 7.93846153846154e-06, | |
| "loss": 0.8409, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01247343620068373, | |
| "grad_norm": 5.375, | |
| "learning_rate": 8.246153846153848e-06, | |
| "loss": 0.9, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.01293541531922757, | |
| "grad_norm": 5.625, | |
| "learning_rate": 8.553846153846156e-06, | |
| "loss": 0.8444, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.013397394437771413, | |
| "grad_norm": 5.25, | |
| "learning_rate": 8.861538461538463e-06, | |
| "loss": 0.8523, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.013859373556315254, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 9.169230769230771e-06, | |
| "loss": 0.9402, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.014321352674859096, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 9.476923076923079e-06, | |
| "loss": 0.7411, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.014783331793402937, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 9.784615384615387e-06, | |
| "loss": 0.8616, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01524531091194678, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.0092307692307693e-05, | |
| "loss": 0.8655, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.015707290030490623, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.04e-05, | |
| "loss": 0.8105, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.016169269149034465, | |
| "grad_norm": 6.96875, | |
| "learning_rate": 1.0707692307692308e-05, | |
| "loss": 0.8538, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.016631248267578305, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.1015384615384616e-05, | |
| "loss": 0.8744, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.017093227386122147, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.1323076923076924e-05, | |
| "loss": 0.8101, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.01755520650466599, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 1.1630769230769231e-05, | |
| "loss": 0.831, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.018017185623209832, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.1938461538461539e-05, | |
| "loss": 0.835, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.01847916474175367, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.2246153846153847e-05, | |
| "loss": 0.7559, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.018941143860297514, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.2553846153846155e-05, | |
| "loss": 0.7849, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.019403122978841357, | |
| "grad_norm": 3.625, | |
| "learning_rate": 1.2861538461538462e-05, | |
| "loss": 0.9112, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.0198651020973852, | |
| "grad_norm": 4.125, | |
| "learning_rate": 1.316923076923077e-05, | |
| "loss": 1.0893, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.02032708121592904, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 1.3476923076923078e-05, | |
| "loss": 0.7646, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.02078906033447288, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 1.3784615384615386e-05, | |
| "loss": 0.8373, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.021251039453016724, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 1.4092307692307693e-05, | |
| "loss": 0.8159, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.021713018571560567, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.4400000000000001e-05, | |
| "loss": 0.8434, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.022174997690104406, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 1.4707692307692309e-05, | |
| "loss": 0.8048, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.02263697680864825, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.5015384615384617e-05, | |
| "loss": 0.8276, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.02309895592719209, | |
| "grad_norm": 6.0625, | |
| "learning_rate": 1.5323076923076926e-05, | |
| "loss": 0.9429, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.023560935045735934, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.5630769230769232e-05, | |
| "loss": 0.8995, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.024022914164279773, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.593846153846154e-05, | |
| "loss": 0.8191, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.024484893282823616, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 1.6246153846153848e-05, | |
| "loss": 0.9168, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.02494687240136746, | |
| "grad_norm": 4.21875, | |
| "learning_rate": 1.6553846153846157e-05, | |
| "loss": 0.8813, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.0254088515199113, | |
| "grad_norm": 4.3125, | |
| "learning_rate": 1.6861538461538463e-05, | |
| "loss": 0.8089, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.02587083063845514, | |
| "grad_norm": 7.625, | |
| "learning_rate": 1.7169230769230772e-05, | |
| "loss": 0.7353, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.026332809756998983, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.747692307692308e-05, | |
| "loss": 0.7367, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.026794788875542826, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.7784615384615388e-05, | |
| "loss": 0.7641, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.02725676799408667, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 1.8092307692307694e-05, | |
| "loss": 0.7952, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.027718747112630508, | |
| "grad_norm": 4.75, | |
| "learning_rate": 1.8400000000000003e-05, | |
| "loss": 0.8539, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02818072623117435, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 1.870769230769231e-05, | |
| "loss": 0.8364, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.028642705349718193, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 1.901538461538462e-05, | |
| "loss": 0.9327, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.029104684468262036, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 1.9323076923076925e-05, | |
| "loss": 0.7912, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.029566663586805875, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.9630769230769234e-05, | |
| "loss": 1.0159, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.030028642705349717, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.993846153846154e-05, | |
| "loss": 0.8499, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.03049062182389356, | |
| "grad_norm": 6.96875, | |
| "learning_rate": 1.9999992835654137e-05, | |
| "loss": 0.8496, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.030952600942437403, | |
| "grad_norm": 3.90625, | |
| "learning_rate": 1.999996373051665e-05, | |
| "loss": 0.8142, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.031414580060981245, | |
| "grad_norm": 4.0, | |
| "learning_rate": 1.9999912236881042e-05, | |
| "loss": 0.8672, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.03187655917952509, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.99998383548626e-05, | |
| "loss": 0.9359, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.03233853829806893, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9999742084626726e-05, | |
| "loss": 0.9701, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.032800517416612766, | |
| "grad_norm": 3.703125, | |
| "learning_rate": 1.9999623426388963e-05, | |
| "loss": 0.7881, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.03326249653515661, | |
| "grad_norm": 3.671875, | |
| "learning_rate": 1.9999482380414973e-05, | |
| "loss": 0.7965, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.03372447565370045, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 1.9999318947020527e-05, | |
| "loss": 0.7915, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.034186454772244294, | |
| "grad_norm": 6.03125, | |
| "learning_rate": 1.999913312657154e-05, | |
| "loss": 0.9149, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.03464843389078814, | |
| "grad_norm": 5.53125, | |
| "learning_rate": 1.9998924919484034e-05, | |
| "loss": 0.7844, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03511041300933198, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.9998694326224154e-05, | |
| "loss": 0.7716, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.03557239212787582, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 1.9998441347308164e-05, | |
| "loss": 0.7758, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.036034371246419665, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.999816598330245e-05, | |
| "loss": 0.8478, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.0364963503649635, | |
| "grad_norm": 5.0, | |
| "learning_rate": 1.9997868234823513e-05, | |
| "loss": 0.7166, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.03695832948350734, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 1.999754810253797e-05, | |
| "loss": 0.8396, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.037420308602051186, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 1.999720558716255e-05, | |
| "loss": 0.9007, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.03788228772059503, | |
| "grad_norm": 4.78125, | |
| "learning_rate": 1.99968406894641e-05, | |
| "loss": 0.8649, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.03834426683913887, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.999645341025957e-05, | |
| "loss": 0.8454, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.038806245957682714, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.9996043750416026e-05, | |
| "loss": 0.8146, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.03926822507622656, | |
| "grad_norm": 4.25, | |
| "learning_rate": 1.9995611710850634e-05, | |
| "loss": 0.7879, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.0397302041947704, | |
| "grad_norm": 3.5625, | |
| "learning_rate": 1.9995157292530672e-05, | |
| "loss": 0.8767, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.040192183313314235, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.999468049647352e-05, | |
| "loss": 0.7827, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.04065416243185808, | |
| "grad_norm": 5.34375, | |
| "learning_rate": 1.9994181323746652e-05, | |
| "loss": 0.9704, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.04111614155040192, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.9993659775467648e-05, | |
| "loss": 0.7169, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.04157812066894576, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 1.999311585280418e-05, | |
| "loss": 1.0719, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.042040099787489606, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 1.9992549556974015e-05, | |
| "loss": 0.8138, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.04250207890603345, | |
| "grad_norm": 3.890625, | |
| "learning_rate": 1.9991960889245005e-05, | |
| "loss": 0.724, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.04296405802457729, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.99913498509351e-05, | |
| "loss": 0.7338, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.043426037143121134, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.999071644341232e-05, | |
| "loss": 0.7862, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.04388801626166497, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.9990060668094778e-05, | |
| "loss": 1.0113, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04434999538020881, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.998938252645066e-05, | |
| "loss": 0.7534, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.044811974498752655, | |
| "grad_norm": 5.5, | |
| "learning_rate": 1.9988682019998236e-05, | |
| "loss": 0.8548, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.0452739536172965, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.9987959150305834e-05, | |
| "loss": 0.9163, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.04573593273584034, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 1.9987213918991855e-05, | |
| "loss": 0.7051, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.04619791185438418, | |
| "grad_norm": 7.71875, | |
| "learning_rate": 1.998644632772477e-05, | |
| "loss": 0.7463, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04619791185438418, | |
| "eval_loss": 0.8494759798049927, | |
| "eval_runtime": 442.9436, | |
| "eval_samples_per_second": 20.578, | |
| "eval_steps_per_second": 2.574, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.046659890972928025, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.998565637822311e-05, | |
| "loss": 0.8329, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.04712187009147187, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.998484407225545e-05, | |
| "loss": 0.7516, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.04758384921001571, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 1.9984009411640433e-05, | |
| "loss": 0.6933, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.048045828328559546, | |
| "grad_norm": 4.125, | |
| "learning_rate": 1.9983152398246747e-05, | |
| "loss": 0.6367, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.04850780744710339, | |
| "grad_norm": 4.25, | |
| "learning_rate": 1.998227303399312e-05, | |
| "loss": 0.8222, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.04896978656564723, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.9981371320848327e-05, | |
| "loss": 0.7939, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.049431765684191074, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.9980447260831177e-05, | |
| "loss": 0.8247, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.04989374480273492, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.99795008560105e-05, | |
| "loss": 0.7723, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.05035572392127876, | |
| "grad_norm": 3.828125, | |
| "learning_rate": 1.997853210850517e-05, | |
| "loss": 0.7777, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.0508177030398226, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 1.9977541020484078e-05, | |
| "loss": 0.9256, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.051279682158366445, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.9976527594166116e-05, | |
| "loss": 0.841, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.05174166127691028, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.9975491831820216e-05, | |
| "loss": 0.8144, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.05220364039545412, | |
| "grad_norm": 6.125, | |
| "learning_rate": 1.9974433735765297e-05, | |
| "loss": 0.8988, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.052665619513997966, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 1.9973353308370282e-05, | |
| "loss": 0.9735, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.05312759863254181, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.99722505520541e-05, | |
| "loss": 0.7992, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05358957775108565, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.9971125469285663e-05, | |
| "loss": 0.7968, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.054051556869629494, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.996997806258387e-05, | |
| "loss": 0.7315, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.05451353598817334, | |
| "grad_norm": 6.59375, | |
| "learning_rate": 1.9968808334517607e-05, | |
| "loss": 0.8133, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.05497551510671718, | |
| "grad_norm": 3.828125, | |
| "learning_rate": 1.9967616287705724e-05, | |
| "loss": 0.8487, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.055437494225261015, | |
| "grad_norm": 6.59375, | |
| "learning_rate": 1.9966401924817042e-05, | |
| "loss": 0.8386, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05589947334380486, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9965165248570357e-05, | |
| "loss": 0.911, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.0563614524623487, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 1.9963906261734404e-05, | |
| "loss": 0.8836, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.05682343158089254, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.9962624967127877e-05, | |
| "loss": 0.7977, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.057285410699436386, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.9961321367619413e-05, | |
| "loss": 0.8145, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.05774738981798023, | |
| "grad_norm": 3.84375, | |
| "learning_rate": 1.9959995466127582e-05, | |
| "loss": 0.7705, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.05820936893652407, | |
| "grad_norm": 4.78125, | |
| "learning_rate": 1.99586472656209e-05, | |
| "loss": 0.7896, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.058671348055067914, | |
| "grad_norm": 6.625, | |
| "learning_rate": 1.9957276769117785e-05, | |
| "loss": 0.8064, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.05913332717361175, | |
| "grad_norm": 4.21875, | |
| "learning_rate": 1.9955883979686587e-05, | |
| "loss": 0.7744, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.05959530629215559, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 1.9954468900445567e-05, | |
| "loss": 0.8338, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.060057285410699435, | |
| "grad_norm": 3.96875, | |
| "learning_rate": 1.9953031534562884e-05, | |
| "loss": 0.7888, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06051926452924328, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.9951571885256594e-05, | |
| "loss": 0.9061, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.06098124364778712, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 1.995008995579465e-05, | |
| "loss": 0.8619, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.06144322276633096, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 1.9948585749494877e-05, | |
| "loss": 0.7711, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.061905201884874805, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.9947059269724983e-05, | |
| "loss": 0.7418, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.06236718100341865, | |
| "grad_norm": 6.34375, | |
| "learning_rate": 1.9945510519902533e-05, | |
| "loss": 0.9288, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.06282916012196249, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.994393950349496e-05, | |
| "loss": 0.8096, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.06329113924050633, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.9942346224019557e-05, | |
| "loss": 0.852, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.06375311835905018, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 1.9940730685043435e-05, | |
| "loss": 0.942, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.06421509747759402, | |
| "grad_norm": 4.09375, | |
| "learning_rate": 1.9939092890183562e-05, | |
| "loss": 0.8851, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.06467707659613786, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.9937432843106733e-05, | |
| "loss": 0.7966, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06513905571468169, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.9935750547529547e-05, | |
| "loss": 0.9728, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.06560103483322553, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9934046007218437e-05, | |
| "loss": 0.8777, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.06606301395176938, | |
| "grad_norm": 3.546875, | |
| "learning_rate": 1.993231922598962e-05, | |
| "loss": 0.8269, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.06652499307031322, | |
| "grad_norm": 4.25, | |
| "learning_rate": 1.993057020770911e-05, | |
| "loss": 0.7792, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.06698697218885706, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 1.9928798956292722e-05, | |
| "loss": 0.8521, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.0674489513074009, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.9927005475706024e-05, | |
| "loss": 0.7401, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.06791093042594475, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 1.9925189769964374e-05, | |
| "loss": 0.7713, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.06837290954448859, | |
| "grad_norm": 8.375, | |
| "learning_rate": 1.992335184313287e-05, | |
| "loss": 0.8873, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.06883488866303243, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 1.992149169932638e-05, | |
| "loss": 0.7554, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.06929686778157627, | |
| "grad_norm": 6.1875, | |
| "learning_rate": 1.9919609342709493e-05, | |
| "loss": 0.8749, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.06975884690012012, | |
| "grad_norm": 4.75, | |
| "learning_rate": 1.991770477749654e-05, | |
| "loss": 0.8489, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.07022082601866396, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.9915778007951572e-05, | |
| "loss": 0.7187, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.0706828051372078, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.9913829038388355e-05, | |
| "loss": 0.8325, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.07114478425575164, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.9911857873170352e-05, | |
| "loss": 0.8896, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.07160676337429549, | |
| "grad_norm": 7.1875, | |
| "learning_rate": 1.9909864516710724e-05, | |
| "loss": 0.7469, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.07206874249283933, | |
| "grad_norm": 4.21875, | |
| "learning_rate": 1.9907848973472307e-05, | |
| "loss": 0.9805, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.07253072161138316, | |
| "grad_norm": 3.703125, | |
| "learning_rate": 1.9905811247967623e-05, | |
| "loss": 0.806, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.072992700729927, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.990375134475885e-05, | |
| "loss": 0.8426, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.07345467984847084, | |
| "grad_norm": 4.0625, | |
| "learning_rate": 1.9901669268457814e-05, | |
| "loss": 0.7435, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.07391665896701469, | |
| "grad_norm": 4.78125, | |
| "learning_rate": 1.9899565023725992e-05, | |
| "loss": 0.8943, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07437863808555853, | |
| "grad_norm": 4.15625, | |
| "learning_rate": 1.989743861527448e-05, | |
| "loss": 0.9149, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.07484061720410237, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.989529004786402e-05, | |
| "loss": 0.8323, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.07530259632264621, | |
| "grad_norm": 3.421875, | |
| "learning_rate": 1.9893119326304938e-05, | |
| "loss": 0.7721, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.07576457544119006, | |
| "grad_norm": 4.25, | |
| "learning_rate": 1.9890926455457172e-05, | |
| "loss": 0.7716, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.0762265545597339, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 1.9888711440230258e-05, | |
| "loss": 0.7043, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.07668853367827774, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.9886474285583283e-05, | |
| "loss": 0.777, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.07715051279682159, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 1.9884214996524935e-05, | |
| "loss": 0.9524, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.07761249191536543, | |
| "grad_norm": 4.78125, | |
| "learning_rate": 1.988193357811343e-05, | |
| "loss": 0.7875, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.07807447103390927, | |
| "grad_norm": 5.34375, | |
| "learning_rate": 1.987963003545655e-05, | |
| "loss": 0.8051, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.07853645015245311, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.9877304373711588e-05, | |
| "loss": 0.8002, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.07899842927099696, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.9874956598085378e-05, | |
| "loss": 0.7691, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.0794604083895408, | |
| "grad_norm": 5.625, | |
| "learning_rate": 1.9872586713834253e-05, | |
| "loss": 0.8483, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.07992238750808464, | |
| "grad_norm": 4.71875, | |
| "learning_rate": 1.987019472626405e-05, | |
| "loss": 0.8258, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.08038436662662847, | |
| "grad_norm": 6.75, | |
| "learning_rate": 1.986778064073009e-05, | |
| "loss": 0.9104, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.08084634574517231, | |
| "grad_norm": 5.28125, | |
| "learning_rate": 1.9865344462637163e-05, | |
| "loss": 0.765, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.08130832486371616, | |
| "grad_norm": 3.984375, | |
| "learning_rate": 1.9862886197439525e-05, | |
| "loss": 0.8766, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.08177030398226, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 1.9860405850640888e-05, | |
| "loss": 0.7801, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.08223228310080384, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.9857903427794393e-05, | |
| "loss": 0.8898, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.08269426221934768, | |
| "grad_norm": 3.859375, | |
| "learning_rate": 1.985537893450261e-05, | |
| "loss": 0.7725, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.08315624133789153, | |
| "grad_norm": 4.71875, | |
| "learning_rate": 1.985283237641752e-05, | |
| "loss": 0.8464, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.08361822045643537, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.9850263759240507e-05, | |
| "loss": 0.7433, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.08408019957497921, | |
| "grad_norm": 3.71875, | |
| "learning_rate": 1.9847673088722337e-05, | |
| "loss": 0.7768, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.08454217869352305, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.9845060370663157e-05, | |
| "loss": 0.9128, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.0850041578120669, | |
| "grad_norm": 6.46875, | |
| "learning_rate": 1.9842425610912467e-05, | |
| "loss": 0.7274, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.08546613693061074, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.983976881536912e-05, | |
| "loss": 0.7639, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.08592811604915458, | |
| "grad_norm": 3.921875, | |
| "learning_rate": 1.9837089989981307e-05, | |
| "loss": 0.7655, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.08639009516769842, | |
| "grad_norm": 3.796875, | |
| "learning_rate": 1.983438914074654e-05, | |
| "loss": 0.7869, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.08685207428624227, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 1.9831666273711628e-05, | |
| "loss": 0.9739, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.08731405340478611, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 1.982892139497269e-05, | |
| "loss": 1.0108, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.08777603252332994, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.9826154510675118e-05, | |
| "loss": 0.8337, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08823801164187378, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 1.9823365627013573e-05, | |
| "loss": 0.8139, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.08869999076041762, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 1.9820554750231968e-05, | |
| "loss": 0.7524, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.08916196987896147, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.981772188662346e-05, | |
| "loss": 0.8152, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.08962394899750531, | |
| "grad_norm": 4.25, | |
| "learning_rate": 1.981486704253042e-05, | |
| "loss": 0.8497, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.09008592811604915, | |
| "grad_norm": 7.1875, | |
| "learning_rate": 1.981199022434445e-05, | |
| "loss": 0.8954, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.090547907234593, | |
| "grad_norm": 5.25, | |
| "learning_rate": 1.9809091438506333e-05, | |
| "loss": 0.8133, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.09100988635313684, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.980617069150603e-05, | |
| "loss": 0.7652, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.09147186547168068, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 1.9803227989882693e-05, | |
| "loss": 0.7811, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.09193384459022452, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9800263340224603e-05, | |
| "loss": 0.8253, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.09239582370876837, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9797276749169192e-05, | |
| "loss": 0.8832, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09239582370876837, | |
| "eval_loss": 0.8367779850959778, | |
| "eval_runtime": 442.761, | |
| "eval_samples_per_second": 20.587, | |
| "eval_steps_per_second": 2.575, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09285780282731221, | |
| "grad_norm": 4.375, | |
| "learning_rate": 1.9794268223403012e-05, | |
| "loss": 1.0233, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.09331978194585605, | |
| "grad_norm": 3.84375, | |
| "learning_rate": 1.9791237769661728e-05, | |
| "loss": 0.8371, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.0937817610643999, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 1.978818539473009e-05, | |
| "loss": 1.0483, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.09424374018294374, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9785111105441942e-05, | |
| "loss": 0.7854, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.09470571930148758, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 1.9782014908680167e-05, | |
| "loss": 0.6771, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.09516769842003142, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 1.977889681137672e-05, | |
| "loss": 0.7874, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.09562967753857525, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 1.9775756820512574e-05, | |
| "loss": 0.9448, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.09609165665711909, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.9772594943117723e-05, | |
| "loss": 0.8937, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.09655363577566294, | |
| "grad_norm": 6.03125, | |
| "learning_rate": 1.9769411186271162e-05, | |
| "loss": 0.9173, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.09701561489420678, | |
| "grad_norm": 7.84375, | |
| "learning_rate": 1.976620555710087e-05, | |
| "loss": 0.8327, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.09747759401275062, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 1.9762978062783793e-05, | |
| "loss": 0.7447, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.09793957313129446, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 1.9759728710545836e-05, | |
| "loss": 0.7932, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.0984015522498383, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 1.9756457507661833e-05, | |
| "loss": 0.8749, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.09886353136838215, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 1.9753164461455548e-05, | |
| "loss": 0.7447, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.09932551048692599, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.974984957929964e-05, | |
| "loss": 0.7091, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.09978748960546983, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 1.9746512868615656e-05, | |
| "loss": 0.9072, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.10024946872401368, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.9743154336874024e-05, | |
| "loss": 0.7947, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.10071144784255752, | |
| "grad_norm": 4.75, | |
| "learning_rate": 1.9739773991594017e-05, | |
| "loss": 0.9017, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.10117342696110136, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 1.9736371840343745e-05, | |
| "loss": 0.7268, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.1016354060796452, | |
| "grad_norm": 5.34375, | |
| "learning_rate": 1.9732947890740143e-05, | |
| "loss": 0.7386, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.10209738519818905, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 1.972950215044895e-05, | |
| "loss": 0.7445, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.10255936431673289, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 1.9726034627184685e-05, | |
| "loss": 0.8159, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.10302134343527672, | |
| "grad_norm": 4.15625, | |
| "learning_rate": 1.9722545328710643e-05, | |
| "loss": 0.7683, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.10348332255382056, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.971903426283887e-05, | |
| "loss": 0.9848, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.1039453016723644, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.971550143743014e-05, | |
| "loss": 0.7591, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.10440728079090825, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.971194686039394e-05, | |
| "loss": 0.7353, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.10486925990945209, | |
| "grad_norm": 5.25, | |
| "learning_rate": 1.9708370539688476e-05, | |
| "loss": 0.8047, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.10533123902799593, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9704772483320616e-05, | |
| "loss": 0.8734, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.10579321814653977, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 1.9701152699345898e-05, | |
| "loss": 1.0071, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.10625519726508362, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 1.9697511195868504e-05, | |
| "loss": 0.9231, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.10671717638362746, | |
| "grad_norm": 4.375, | |
| "learning_rate": 1.969384798104124e-05, | |
| "loss": 0.7658, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.1071791555021713, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 1.9690163063065532e-05, | |
| "loss": 0.8602, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.10764113462071515, | |
| "grad_norm": 6.03125, | |
| "learning_rate": 1.9686456450191372e-05, | |
| "loss": 0.8779, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.10810311373925899, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 1.968272815071736e-05, | |
| "loss": 0.8708, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.10856509285780283, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 1.9678978172990612e-05, | |
| "loss": 0.8517, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.10902707197634667, | |
| "grad_norm": 6.59375, | |
| "learning_rate": 1.9675206525406803e-05, | |
| "loss": 0.8727, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.10948905109489052, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.9671413216410116e-05, | |
| "loss": 0.7547, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.10995103021343436, | |
| "grad_norm": 4.03125, | |
| "learning_rate": 1.966759825449323e-05, | |
| "loss": 0.7722, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.1104130093319782, | |
| "grad_norm": 3.78125, | |
| "learning_rate": 1.9663761648197302e-05, | |
| "loss": 0.9483, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.11087498845052203, | |
| "grad_norm": 3.84375, | |
| "learning_rate": 1.965990340611195e-05, | |
| "loss": 0.846, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.11133696756906587, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.9656023536875227e-05, | |
| "loss": 0.8802, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.11179894668760972, | |
| "grad_norm": 3.921875, | |
| "learning_rate": 1.965212204917361e-05, | |
| "loss": 0.8081, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.11226092580615356, | |
| "grad_norm": 5.0, | |
| "learning_rate": 1.964819895174198e-05, | |
| "loss": 0.8051, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.1127229049246974, | |
| "grad_norm": 4.15625, | |
| "learning_rate": 1.964425425336359e-05, | |
| "loss": 0.8133, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.11318488404324124, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 1.9640287962870063e-05, | |
| "loss": 0.8061, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.11364686316178509, | |
| "grad_norm": 3.984375, | |
| "learning_rate": 1.9636300089141355e-05, | |
| "loss": 0.7362, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.11410884228032893, | |
| "grad_norm": 7.625, | |
| "learning_rate": 1.9632290641105754e-05, | |
| "loss": 1.0105, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.11457082139887277, | |
| "grad_norm": 4.75, | |
| "learning_rate": 1.962825962773984e-05, | |
| "loss": 0.6403, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.11503280051741661, | |
| "grad_norm": 5.75, | |
| "learning_rate": 1.962420705806848e-05, | |
| "loss": 0.8769, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.11549477963596046, | |
| "grad_norm": 4.21875, | |
| "learning_rate": 1.9620132941164806e-05, | |
| "loss": 0.8211, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.1159567587545043, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.961603728615018e-05, | |
| "loss": 0.8711, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.11641873787304814, | |
| "grad_norm": 6.375, | |
| "learning_rate": 1.96119201021942e-05, | |
| "loss": 0.8252, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.11688071699159198, | |
| "grad_norm": 4.375, | |
| "learning_rate": 1.9607781398514646e-05, | |
| "loss": 0.8147, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.11734269611013583, | |
| "grad_norm": 4.21875, | |
| "learning_rate": 1.9603621184377498e-05, | |
| "loss": 0.7746, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.11780467522867967, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.9599439469096876e-05, | |
| "loss": 0.965, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.1182666543472235, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 1.9595236262035057e-05, | |
| "loss": 0.711, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.11872863346576734, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 1.959101157260241e-05, | |
| "loss": 0.7464, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.11919061258431118, | |
| "grad_norm": 5.125, | |
| "learning_rate": 1.9586765410257424e-05, | |
| "loss": 0.7712, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.11965259170285503, | |
| "grad_norm": 6.0625, | |
| "learning_rate": 1.958249778450665e-05, | |
| "loss": 0.8344, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.12011457082139887, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 1.95782087049047e-05, | |
| "loss": 0.7442, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.12057654993994271, | |
| "grad_norm": 7.8125, | |
| "learning_rate": 1.957389818105421e-05, | |
| "loss": 0.818, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.12103852905848655, | |
| "grad_norm": 4.125, | |
| "learning_rate": 1.9569566222605832e-05, | |
| "loss": 0.6755, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.1215005081770304, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.9565212839258204e-05, | |
| "loss": 0.8302, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.12196248729557424, | |
| "grad_norm": 4.3125, | |
| "learning_rate": 1.9560838040757933e-05, | |
| "loss": 0.7821, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.12242446641411808, | |
| "grad_norm": 5.75, | |
| "learning_rate": 1.955644183689957e-05, | |
| "loss": 0.8591, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.12288644553266193, | |
| "grad_norm": 4.0, | |
| "learning_rate": 1.9552024237525597e-05, | |
| "loss": 0.7422, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.12334842465120577, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 1.9547585252526388e-05, | |
| "loss": 0.7762, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.12381040376974961, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.9543124891840196e-05, | |
| "loss": 0.8118, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.12427238288829345, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.9538643165453138e-05, | |
| "loss": 0.7443, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.1247343620068373, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.9534140083399165e-05, | |
| "loss": 0.864, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.12519634112538114, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 1.9529615655760034e-05, | |
| "loss": 0.7972, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.12565832024392498, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 1.9525069892665295e-05, | |
| "loss": 0.8191, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.12612029936246882, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.952050280429227e-05, | |
| "loss": 0.7575, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.12658227848101267, | |
| "grad_norm": 5.3125, | |
| "learning_rate": 1.9515914400866022e-05, | |
| "loss": 0.817, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.1270442575995565, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.951130469265933e-05, | |
| "loss": 0.9624, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.12750623671810035, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.9506673689992673e-05, | |
| "loss": 0.8555, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.1279682158366442, | |
| "grad_norm": 4.03125, | |
| "learning_rate": 1.950202140323422e-05, | |
| "loss": 0.803, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.12843019495518804, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.9497347842799767e-05, | |
| "loss": 0.8625, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.12889217407373188, | |
| "grad_norm": 5.90625, | |
| "learning_rate": 1.9492653019152762e-05, | |
| "loss": 0.7429, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.12935415319227572, | |
| "grad_norm": 6.71875, | |
| "learning_rate": 1.9487936942804237e-05, | |
| "loss": 0.8089, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.12981613231081954, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 1.948319962431283e-05, | |
| "loss": 0.8077, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.13027811142936338, | |
| "grad_norm": 4.0625, | |
| "learning_rate": 1.9478441074284713e-05, | |
| "loss": 0.8612, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.13074009054790722, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 1.947366130337361e-05, | |
| "loss": 0.7953, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.13120206966645107, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.9468860322280746e-05, | |
| "loss": 0.8029, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.1316640487849949, | |
| "grad_norm": 6.0, | |
| "learning_rate": 1.946403814175484e-05, | |
| "loss": 0.9268, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.13212602790353875, | |
| "grad_norm": 3.921875, | |
| "learning_rate": 1.9459194772592062e-05, | |
| "loss": 0.8112, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.1325880070220826, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 1.9454330225636035e-05, | |
| "loss": 0.6757, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.13304998614062644, | |
| "grad_norm": 5.25, | |
| "learning_rate": 1.944944451177778e-05, | |
| "loss": 0.784, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.13351196525917028, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.9444537641955725e-05, | |
| "loss": 0.9202, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.13397394437771412, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.943960962715565e-05, | |
| "loss": 0.8239, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.13443592349625796, | |
| "grad_norm": 5.125, | |
| "learning_rate": 1.9434660478410676e-05, | |
| "loss": 0.8025, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.1348979026148018, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 1.9429690206801255e-05, | |
| "loss": 0.8056, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.13535988173334565, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.942469882345511e-05, | |
| "loss": 0.9176, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.1358218608518895, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.941968633954724e-05, | |
| "loss": 0.8589, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.13628383997043333, | |
| "grad_norm": 4.375, | |
| "learning_rate": 1.9414652766299887e-05, | |
| "loss": 0.9654, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.13674581908897718, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.9409598114982503e-05, | |
| "loss": 0.7178, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.13720779820752102, | |
| "grad_norm": 5.875, | |
| "learning_rate": 1.9404522396911742e-05, | |
| "loss": 0.8832, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.13766977732606486, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.9399425623451405e-05, | |
| "loss": 0.8601, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.1381317564446087, | |
| "grad_norm": 4.0, | |
| "learning_rate": 1.9394307806012454e-05, | |
| "loss": 0.6866, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.13859373556315255, | |
| "grad_norm": 5.125, | |
| "learning_rate": 1.9389168956052945e-05, | |
| "loss": 0.7372, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.13859373556315255, | |
| "eval_loss": 0.8288407325744629, | |
| "eval_runtime": 443.0014, | |
| "eval_samples_per_second": 20.576, | |
| "eval_steps_per_second": 2.573, | |
| "step": 1500 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 10823, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 6.4949569191936e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |