| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 1000, | |
| "global_step": 1900, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.1578947368421055e-07, | |
| "loss": 0.7568, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.157894736842105e-06, | |
| "loss": 0.7081, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.31578947368421e-06, | |
| "loss": 0.6806, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.473684210526315e-06, | |
| "loss": 0.6123, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.263157894736842e-05, | |
| "loss": 0.4982, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.5789473684210526e-05, | |
| "loss": 0.4448, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.894736842105263e-05, | |
| "loss": 0.4243, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.2105263157894736e-05, | |
| "loss": 0.3961, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.526315789473684e-05, | |
| "loss": 0.3963, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.8421052631578946e-05, | |
| "loss": 0.3795, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.9999432005848255e-05, | |
| "loss": 0.3786, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.99948883107249e-05, | |
| "loss": 0.3962, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.9985802296874666e-05, | |
| "loss": 0.3566, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.9972176716673562e-05, | |
| "loss": 0.3817, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.9954015697643372e-05, | |
| "loss": 0.3644, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.9931324741201325e-05, | |
| "loss": 0.3497, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.9904110720993565e-05, | |
| "loss": 0.3601, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.987238188081299e-05, | |
| "loss": 0.3877, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.983614783210197e-05, | |
| "loss": 0.3585, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.9795419551040836e-05, | |
| "loss": 0.3498, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.9750209375222893e-05, | |
| "loss": 0.3618, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.97005309999171e-05, | |
| "loss": 0.3496, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.964639947391939e-05, | |
| "loss": 0.3483, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.958783119499408e-05, | |
| "loss": 0.3488, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.9524843904906528e-05, | |
| "loss": 0.3405, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.9457456684048772e-05, | |
| "loss": 0.3441, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.938568994565956e-05, | |
| "loss": 0.3638, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.9309565429640724e-05, | |
| "loss": 0.3695, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.9229106195971603e-05, | |
| "loss": 0.3372, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.9144336617723625e-05, | |
| "loss": 0.354, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.90552823736771e-05, | |
| "loss": 0.358, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.8961970440542496e-05, | |
| "loss": 0.3473, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.8864429084788534e-05, | |
| "loss": 0.3379, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.8762687854079563e-05, | |
| "loss": 0.3794, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.8656777568324878e-05, | |
| "loss": 0.3475, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.8546730310342593e-05, | |
| "loss": 0.3655, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.8432579416140984e-05, | |
| "loss": 0.3354, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.8314359464820184e-05, | |
| "loss": 0.3448, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.8192106268097336e-05, | |
| "loss": 0.3565, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.8065856859458346e-05, | |
| "loss": 0.3436, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.7935649482939533e-05, | |
| "loss": 0.3509, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.7801523581542563e-05, | |
| "loss": 0.3312, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.766351978528622e-05, | |
| "loss": 0.3445, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7521679898898567e-05, | |
| "loss": 0.3374, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.737604688915327e-05, | |
| "loss": 0.3438, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.72266648718539e-05, | |
| "loss": 0.3307, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.7073579098470196e-05, | |
| "loss": 0.3344, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.6916835942430292e-05, | |
| "loss": 0.325, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.6756482885073032e-05, | |
| "loss": 0.3296, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.6592568501264746e-05, | |
| "loss": 0.3536, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6425142444684735e-05, | |
| "loss": 0.3272, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.6254255432783933e-05, | |
| "loss": 0.3547, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.6079959231421347e-05, | |
| "loss": 0.3313, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.5902306639182952e-05, | |
| "loss": 0.343, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.5721351471387666e-05, | |
| "loss": 0.3129, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.5537148543785385e-05, | |
| "loss": 0.3505, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.534975365595196e-05, | |
| "loss": 0.3523, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.5159223574386117e-05, | |
| "loss": 0.3699, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.496561601531353e-05, | |
| "loss": 0.315, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.4768989627203123e-05, | |
| "loss": 0.3382, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.4569403973001045e-05, | |
| "loss": 0.3387, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.436691951208758e-05, | |
| "loss": 0.3286, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.4161597581962526e-05, | |
| "loss": 0.3096, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.395350037966456e-05, | |
| "loss": 0.3384, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.3742690942930235e-05, | |
| "loss": 0.3284, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.3529233131098313e-05, | |
| "loss": 0.3295, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.33131916057652e-05, | |
| "loss": 0.3164, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.309463181119736e-05, | |
| "loss": 0.3597, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.287361995450667e-05, | |
| "loss": 0.3239, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.2650222985594634e-05, | |
| "loss": 0.3298, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.2424508576871623e-05, | |
| "loss": 0.3176, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.219654510275728e-05, | |
| "loss": 0.3344, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.1966401618968194e-05, | |
| "loss": 0.3518, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.173414784159925e-05, | |
| "loss": 0.3323, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.149985412600492e-05, | |
| "loss": 0.3343, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1263591445486895e-05, | |
| "loss": 0.3383, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1025431369794546e-05, | |
| "loss": 0.3135, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0785446043444677e-05, | |
| "loss": 0.3278, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.0543708163867204e-05, | |
| "loss": 0.3148, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.0300290959383318e-05, | |
| "loss": 0.3511, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.0055268167022835e-05, | |
| "loss": 0.3146, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.9808714010187425e-05, | |
| "loss": 0.3321, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.9560703176166565e-05, | |
| "loss": 0.3373, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.931131079351289e-05, | |
| "loss": 0.32, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.9060612409283946e-05, | |
| "loss": 0.3106, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8808683966157132e-05, | |
| "loss": 0.3167, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8555601779424778e-05, | |
| "loss": 0.2993, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.8301442513876406e-05, | |
| "loss": 0.3405, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.804628316057508e-05, | |
| "loss": 0.3346, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.779020101353492e-05, | |
| "loss": 0.3319, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.7533273646306857e-05, | |
| "loss": 0.3087, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.7275578888479714e-05, | |
| "loss": 0.316, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7017194802103705e-05, | |
| "loss": 0.3054, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.6758199658043538e-05, | |
| "loss": 0.3255, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.6498671912268256e-05, | |
| "loss": 0.3175, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.623869018208499e-05, | |
| "loss": 0.3424, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.5978333222323858e-05, | |
| "loss": 0.2825, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.571767990148122e-05, | |
| "loss": 0.2886, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.5456809177828444e-05, | |
| "loss": 0.3196, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.5195800075493542e-05, | |
| "loss": 0.3178, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.4934731660522817e-05, | |
| "loss": 0.2676, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_loss": 0.508576512336731, | |
| "eval_runtime": 12.2399, | |
| "eval_samples_per_second": 3.595, | |
| "eval_steps_per_second": 0.899, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.4673683016929805e-05, | |
| "loss": 0.2803, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.441273322273884e-05, | |
| "loss": 0.319, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.4151961326030314e-05, | |
| "loss": 0.2965, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.3891446320995143e-05, | |
| "loss": 0.3067, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.3631267124005453e-05, | |
| "loss": 0.3072, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.337150254970891e-05, | |
| "loss": 0.2963, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.3112231287153798e-05, | |
| "loss": 0.3063, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.28535318759522e-05, | |
| "loss": 0.2732, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.2595482682488443e-05, | |
| "loss": 0.2874, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.2338161876179964e-05, | |
| "loss": 0.2998, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.2081647405797923e-05, | |
| "loss": 0.2849, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.1826016975854563e-05, | |
| "loss": 0.2735, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.1571348023064662e-05, | |
| "loss": 0.2937, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.1317717692888014e-05, | |
| "loss": 0.2993, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.1065202816160213e-05, | |
| "loss": 0.2933, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.081387988581869e-05, | |
| "loss": 0.285, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.0563825033731146e-05, | |
| "loss": 0.2896, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.031511400763332e-05, | |
| "loss": 0.2874, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.0067822148183194e-05, | |
| "loss": 0.2851, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.822024366138397e-06, | |
| "loss": 0.2925, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.577795119663966e-06, | |
| "loss": 0.2843, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.335208391777106e-06, | |
| "loss": 0.2884, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 9.094337667935942e-06, | |
| "loss": 0.269, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 8.855255913778949e-06, | |
| "loss": 0.2849, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.618035553021925e-06, | |
| "loss": 0.304, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 8.382748445519008e-06, | |
| "loss": 0.3049, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 8.149465865494633e-06, | |
| "loss": 0.2999, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.918258479952763e-06, | |
| "loss": 0.2835, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 7.689196327270171e-06, | |
| "loss": 0.3167, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.462348795980088e-06, | |
| "loss": 0.2842, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.237784603752705e-06, | |
| "loss": 0.2909, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 7.015571776578922e-06, | |
| "loss": 0.2881, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.795777628163599e-06, | |
| "loss": 0.2796, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.578468739534602e-06, | |
| "loss": 0.3056, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.363710938873759e-06, | |
| "loss": 0.2987, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 6.151569281575925e-06, | |
| "loss": 0.288, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.942108030542074e-06, | |
| "loss": 0.2954, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.735390636712514e-06, | |
| "loss": 0.2837, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.531479719846038e-06, | |
| "loss": 0.3055, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.330437049550868e-06, | |
| "loss": 0.313, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.132323526573126e-06, | |
| "loss": 0.2966, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.937199164348521e-06, | |
| "loss": 0.2741, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.745123070822786e-06, | |
| "loss": 0.2973, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.556153430546451e-06, | |
| "loss": 0.281, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.370347487049313e-06, | |
| "loss": 0.2905, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.187761525499973e-06, | |
| "loss": 0.2806, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.008450855655675e-06, | |
| "loss": 0.2716, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.83246979510764e-06, | |
| "loss": 0.2834, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.676977737529078e-06, | |
| "loss": 0.2809, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 3.5074689542164895e-06, | |
| "loss": 0.2845, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.341441539881574e-06, | |
| "loss": 0.3001, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.1789457882922753e-06, | |
| "loss": 0.2941, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.020030923389471e-06, | |
| "loss": 0.2917, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.86474508437579e-06, | |
| "loss": 0.293, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.7131353111330843e-06, | |
| "loss": 0.2941, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.565247529972901e-06, | |
| "loss": 0.2787, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.4211265397242854e-06, | |
| "loss": 0.2899, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.280815998163083e-06, | |
| "loss": 0.2814, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.144358408786986e-06, | |
| "loss": 0.2786, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.011795107940138e-06, | |
| "loss": 0.2916, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.8831662522913594e-06, | |
| "loss": 0.2834, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.7585108066697136e-06, | |
| "loss": 0.2735, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.6378665322611002e-06, | |
| "loss": 0.3039, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.521269975169471e-06, | |
| "loss": 0.2872, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.408756455346114e-06, | |
| "loss": 0.2863, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.3003600558903927e-06, | |
| "loss": 0.2854, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.196113612725116e-06, | |
| "loss": 0.2727, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.0960487046497524e-06, | |
| "loss": 0.2763, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.000195643774431e-06, | |
| "loss": 0.2962, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.085834663376629e-07, | |
| "loss": 0.2888, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.212399239105534e-07, | |
| "loss": 0.2839, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.381914749901752e-07, | |
| "loss": 0.2789, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.594632769846353e-07, | |
| "loss": 0.2772, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 5.850791785922849e-07, | |
| "loss": 0.278, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.150617125773633e-07, | |
| "loss": 0.2878, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.494320889442749e-07, | |
| "loss": 0.2734, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.882101885125539e-07, | |
| "loss": 0.2826, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.3141455689448266e-07, | |
| "loss": 0.2875, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.790623988771712e-07, | |
| "loss": 0.2898, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.3116957321080102e-07, | |
| "loss": 0.2919, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8775058780463094e-07, | |
| "loss": 0.2778, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.4881859533218466e-07, | |
| "loss": 0.3179, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.1438538924699094e-07, | |
| "loss": 0.2771, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.446140021006132e-08, | |
| "loss": 0.2817, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.9055692930179426e-08, | |
| "loss": 0.2903, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.8175963417980685e-08, | |
| "loss": 0.2934, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.1828536654647235e-08, | |
| "loss": 0.2952, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.0018364675912217e-08, | |
| "loss": 0.2881, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.7490250719663933e-09, | |
| "loss": 0.2701, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.2719903721712954e-11, | |
| "loss": 0.2955, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 1900, | |
| "total_flos": 3.4203213408659046e+17, | |
| "train_loss": 0.3227183536165639, | |
| "train_runtime": 6710.5312, | |
| "train_samples_per_second": 1.132, | |
| "train_steps_per_second": 0.283 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1900, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 1000, | |
| "total_flos": 3.4203213408659046e+17, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |