{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.0, "global_step": 162500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.03, "learning_rate": 4.984615384615385e-05, "loss": 3.5497, "step": 500 }, { "epoch": 0.06, "learning_rate": 4.969230769230769e-05, "loss": 3.4573, "step": 1000 }, { "epoch": 0.09, "learning_rate": 4.953846153846154e-05, "loss": 3.3973, "step": 1500 }, { "epoch": 0.12, "learning_rate": 4.9384615384615384e-05, "loss": 3.3772, "step": 2000 }, { "epoch": 0.15, "learning_rate": 4.923076923076924e-05, "loss": 3.3331, "step": 2500 }, { "epoch": 0.18, "learning_rate": 4.907692307692308e-05, "loss": 3.3312, "step": 3000 }, { "epoch": 0.22, "learning_rate": 4.892307692307693e-05, "loss": 3.3091, "step": 3500 }, { "epoch": 0.25, "learning_rate": 4.876923076923077e-05, "loss": 3.3126, "step": 4000 }, { "epoch": 0.28, "learning_rate": 4.861538461538462e-05, "loss": 3.2995, "step": 4500 }, { "epoch": 0.31, "learning_rate": 4.846153846153846e-05, "loss": 3.274, "step": 5000 }, { "epoch": 0.34, "learning_rate": 4.830769230769231e-05, "loss": 3.2612, "step": 5500 }, { "epoch": 0.37, "learning_rate": 4.815384615384615e-05, "loss": 3.2786, "step": 6000 }, { "epoch": 0.4, "learning_rate": 4.8e-05, "loss": 3.2465, "step": 6500 }, { "epoch": 0.43, "learning_rate": 4.784615384615384e-05, "loss": 3.265, "step": 7000 }, { "epoch": 0.46, "learning_rate": 4.76923076923077e-05, "loss": 3.2207, "step": 7500 }, { "epoch": 0.49, "learning_rate": 4.753846153846154e-05, "loss": 3.2264, "step": 8000 }, { "epoch": 0.52, "learning_rate": 4.738461538461539e-05, "loss": 3.227, "step": 8500 }, { "epoch": 0.55, "learning_rate": 4.723076923076923e-05, "loss": 3.2152, "step": 9000 }, { "epoch": 0.58, "learning_rate": 4.707692307692308e-05, "loss": 3.2229, "step": 9500 }, { "epoch": 0.62, "learning_rate": 4.692307692307693e-05, "loss": 3.2352, "step": 10000 }, { "epoch": 0.65, "learning_rate": 4.676923076923077e-05, "loss": 3.2247, "step": 10500 }, { "epoch": 0.68, "learning_rate": 4.661538461538462e-05, "loss": 3.2262, "step": 11000 }, { "epoch": 0.71, "learning_rate": 4.646153846153846e-05, "loss": 3.2073, "step": 11500 }, { "epoch": 0.74, "learning_rate": 4.630769230769231e-05, "loss": 3.2027, "step": 12000 }, { "epoch": 0.77, "learning_rate": 4.615384615384616e-05, "loss": 3.1998, "step": 12500 }, { "epoch": 0.8, "learning_rate": 4.600000000000001e-05, "loss": 3.1736, "step": 13000 }, { "epoch": 0.83, "learning_rate": 4.584615384615385e-05, "loss": 3.1947, "step": 13500 }, { "epoch": 0.86, "learning_rate": 4.56923076923077e-05, "loss": 3.186, "step": 14000 }, { "epoch": 0.89, "learning_rate": 4.553846153846154e-05, "loss": 3.1766, "step": 14500 }, { "epoch": 0.92, "learning_rate": 4.538461538461539e-05, "loss": 3.1851, "step": 15000 }, { "epoch": 0.95, "learning_rate": 4.523076923076923e-05, "loss": 3.1989, "step": 15500 }, { "epoch": 0.98, "learning_rate": 4.507692307692308e-05, "loss": 3.2022, "step": 16000 }, { "epoch": 1.02, "learning_rate": 4.492307692307692e-05, "loss": 3.0511, "step": 16500 }, { "epoch": 1.05, "learning_rate": 4.476923076923077e-05, "loss": 2.9082, "step": 17000 }, { "epoch": 1.08, "learning_rate": 4.461538461538462e-05, "loss": 2.9545, "step": 17500 }, { "epoch": 1.11, "learning_rate": 4.4461538461538466e-05, "loss": 2.9195, "step": 18000 }, { "epoch": 1.14, "learning_rate": 4.430769230769231e-05, "loss": 2.9344, "step": 18500 }, { "epoch": 1.17, "learning_rate": 4.415384615384616e-05, "loss": 2.9238, "step": 19000 }, { "epoch": 1.2, "learning_rate": 4.4000000000000006e-05, "loss": 2.9501, "step": 19500 }, { "epoch": 1.23, "learning_rate": 4.384615384615385e-05, "loss": 2.943, "step": 20000 }, { "epoch": 1.26, "learning_rate": 4.3692307692307696e-05, "loss": 2.9397, "step": 20500 }, { "epoch": 1.29, "learning_rate": 4.353846153846154e-05, "loss": 2.9391, "step": 21000 }, { "epoch": 1.32, "learning_rate": 4.338461538461539e-05, "loss": 2.9594, "step": 21500 }, { "epoch": 1.35, "learning_rate": 4.323076923076923e-05, "loss": 2.9497, "step": 22000 }, { "epoch": 1.38, "learning_rate": 4.3076923076923084e-05, "loss": 2.9476, "step": 22500 }, { "epoch": 1.42, "learning_rate": 4.2923076923076926e-05, "loss": 2.9507, "step": 23000 }, { "epoch": 1.45, "learning_rate": 4.2769230769230775e-05, "loss": 2.9537, "step": 23500 }, { "epoch": 1.48, "learning_rate": 4.2615384615384617e-05, "loss": 2.9657, "step": 24000 }, { "epoch": 1.51, "learning_rate": 4.2461538461538465e-05, "loss": 2.9845, "step": 24500 }, { "epoch": 1.54, "learning_rate": 4.230769230769231e-05, "loss": 2.9733, "step": 25000 }, { "epoch": 1.57, "learning_rate": 4.2153846153846156e-05, "loss": 2.9281, "step": 25500 }, { "epoch": 1.6, "learning_rate": 4.2e-05, "loss": 2.9585, "step": 26000 }, { "epoch": 1.63, "learning_rate": 4.1846153846153846e-05, "loss": 2.9716, "step": 26500 }, { "epoch": 1.66, "learning_rate": 4.169230769230769e-05, "loss": 2.9583, "step": 27000 }, { "epoch": 1.69, "learning_rate": 4.1538461538461544e-05, "loss": 2.964, "step": 27500 }, { "epoch": 1.72, "learning_rate": 4.1384615384615386e-05, "loss": 2.9686, "step": 28000 }, { "epoch": 1.75, "learning_rate": 4.1230769230769234e-05, "loss": 2.9525, "step": 28500 }, { "epoch": 1.78, "learning_rate": 4.1076923076923076e-05, "loss": 2.9803, "step": 29000 }, { "epoch": 1.82, "learning_rate": 4.0923076923076925e-05, "loss": 2.9545, "step": 29500 }, { "epoch": 1.85, "learning_rate": 4.0769230769230773e-05, "loss": 2.9797, "step": 30000 }, { "epoch": 1.88, "learning_rate": 4.0615384615384615e-05, "loss": 2.9701, "step": 30500 }, { "epoch": 1.91, "learning_rate": 4.0461538461538464e-05, "loss": 2.939, "step": 31000 }, { "epoch": 1.94, "learning_rate": 4.0307692307692306e-05, "loss": 2.9627, "step": 31500 }, { "epoch": 1.97, "learning_rate": 4.0153846153846155e-05, "loss": 2.9526, "step": 32000 }, { "epoch": 2.0, "learning_rate": 4e-05, "loss": 2.977, "step": 32500 }, { "epoch": 2.03, "learning_rate": 3.984615384615385e-05, "loss": 2.7112, "step": 33000 }, { "epoch": 2.06, "learning_rate": 3.9692307692307694e-05, "loss": 2.694, "step": 33500 }, { "epoch": 2.09, "learning_rate": 3.953846153846154e-05, "loss": 2.6971, "step": 34000 }, { "epoch": 2.12, "learning_rate": 3.9384615384615384e-05, "loss": 2.7095, "step": 34500 }, { "epoch": 2.15, "learning_rate": 3.923076923076923e-05, "loss": 2.7013, "step": 35000 }, { "epoch": 2.18, "learning_rate": 3.9076923076923075e-05, "loss": 2.7182, "step": 35500 }, { "epoch": 2.22, "learning_rate": 3.8923076923076924e-05, "loss": 2.7253, "step": 36000 }, { "epoch": 2.25, "learning_rate": 3.8769230769230766e-05, "loss": 2.7323, "step": 36500 }, { "epoch": 2.28, "learning_rate": 3.861538461538462e-05, "loss": 2.7529, "step": 37000 }, { "epoch": 2.31, "learning_rate": 3.846153846153846e-05, "loss": 2.7394, "step": 37500 }, { "epoch": 2.34, "learning_rate": 3.830769230769231e-05, "loss": 2.7744, "step": 38000 }, { "epoch": 2.37, "learning_rate": 3.8153846153846153e-05, "loss": 2.745, "step": 38500 }, { "epoch": 2.4, "learning_rate": 3.8e-05, "loss": 2.7412, "step": 39000 }, { "epoch": 2.43, "learning_rate": 3.784615384615385e-05, "loss": 2.7509, "step": 39500 }, { "epoch": 2.46, "learning_rate": 3.769230769230769e-05, "loss": 2.7477, "step": 40000 }, { "epoch": 2.49, "learning_rate": 3.753846153846154e-05, "loss": 2.7719, "step": 40500 }, { "epoch": 2.52, "learning_rate": 3.738461538461538e-05, "loss": 2.762, "step": 41000 }, { "epoch": 2.55, "learning_rate": 3.723076923076923e-05, "loss": 2.7684, "step": 41500 }, { "epoch": 2.58, "learning_rate": 3.707692307692308e-05, "loss": 2.7693, "step": 42000 }, { "epoch": 2.62, "learning_rate": 3.692307692307693e-05, "loss": 2.7502, "step": 42500 }, { "epoch": 2.65, "learning_rate": 3.676923076923077e-05, "loss": 2.7662, "step": 43000 }, { "epoch": 2.68, "learning_rate": 3.661538461538462e-05, "loss": 2.7844, "step": 43500 }, { "epoch": 2.71, "learning_rate": 3.646153846153846e-05, "loss": 2.7539, "step": 44000 }, { "epoch": 2.74, "learning_rate": 3.630769230769231e-05, "loss": 2.7934, "step": 44500 }, { "epoch": 2.77, "learning_rate": 3.615384615384615e-05, "loss": 2.7851, "step": 45000 }, { "epoch": 2.8, "learning_rate": 3.6e-05, "loss": 2.7789, "step": 45500 }, { "epoch": 2.83, "learning_rate": 3.584615384615384e-05, "loss": 2.7653, "step": 46000 }, { "epoch": 2.86, "learning_rate": 3.569230769230769e-05, "loss": 2.7917, "step": 46500 }, { "epoch": 2.89, "learning_rate": 3.553846153846154e-05, "loss": 2.7721, "step": 47000 }, { "epoch": 2.92, "learning_rate": 3.538461538461539e-05, "loss": 2.7728, "step": 47500 }, { "epoch": 2.95, "learning_rate": 3.523076923076923e-05, "loss": 2.7738, "step": 48000 }, { "epoch": 2.98, "learning_rate": 3.507692307692308e-05, "loss": 2.8052, "step": 48500 }, { "epoch": 3.02, "learning_rate": 3.492307692307693e-05, "loss": 2.6651, "step": 49000 }, { "epoch": 3.05, "learning_rate": 3.476923076923077e-05, "loss": 2.5394, "step": 49500 }, { "epoch": 3.08, "learning_rate": 3.461538461538462e-05, "loss": 2.5482, "step": 50000 }, { "epoch": 3.11, "learning_rate": 3.446153846153846e-05, "loss": 2.5339, "step": 50500 }, { "epoch": 3.14, "learning_rate": 3.430769230769231e-05, "loss": 2.519, "step": 51000 }, { "epoch": 3.17, "learning_rate": 3.415384615384615e-05, "loss": 2.5416, "step": 51500 }, { "epoch": 3.2, "learning_rate": 3.4000000000000007e-05, "loss": 2.5611, "step": 52000 }, { "epoch": 3.23, "learning_rate": 3.384615384615385e-05, "loss": 2.5651, "step": 52500 }, { "epoch": 3.26, "learning_rate": 3.36923076923077e-05, "loss": 2.5802, "step": 53000 }, { "epoch": 3.29, "learning_rate": 3.353846153846154e-05, "loss": 2.5526, "step": 53500 }, { "epoch": 3.32, "learning_rate": 3.338461538461539e-05, "loss": 2.5521, "step": 54000 }, { "epoch": 3.35, "learning_rate": 3.323076923076923e-05, "loss": 2.5678, "step": 54500 }, { "epoch": 3.38, "learning_rate": 3.307692307692308e-05, "loss": 2.5762, "step": 55000 }, { "epoch": 3.42, "learning_rate": 3.292307692307692e-05, "loss": 2.5937, "step": 55500 }, { "epoch": 3.45, "learning_rate": 3.276923076923077e-05, "loss": 2.5692, "step": 56000 }, { "epoch": 3.48, "learning_rate": 3.261538461538462e-05, "loss": 2.5932, "step": 56500 }, { "epoch": 3.51, "learning_rate": 3.2461538461538466e-05, "loss": 2.5469, "step": 57000 }, { "epoch": 3.54, "learning_rate": 3.230769230769231e-05, "loss": 2.6019, "step": 57500 }, { "epoch": 3.57, "learning_rate": 3.215384615384616e-05, "loss": 2.5929, "step": 58000 }, { "epoch": 3.6, "learning_rate": 3.2000000000000005e-05, "loss": 2.6075, "step": 58500 }, { "epoch": 3.63, "learning_rate": 3.184615384615385e-05, "loss": 2.6013, "step": 59000 }, { "epoch": 3.66, "learning_rate": 3.1692307692307696e-05, "loss": 2.6138, "step": 59500 }, { "epoch": 3.69, "learning_rate": 3.153846153846154e-05, "loss": 2.6259, "step": 60000 }, { "epoch": 3.72, "learning_rate": 3.1384615384615386e-05, "loss": 2.5918, "step": 60500 }, { "epoch": 3.75, "learning_rate": 3.123076923076923e-05, "loss": 2.5902, "step": 61000 }, { "epoch": 3.78, "learning_rate": 3.107692307692308e-05, "loss": 2.5902, "step": 61500 }, { "epoch": 3.82, "learning_rate": 3.0923076923076926e-05, "loss": 2.6132, "step": 62000 }, { "epoch": 3.85, "learning_rate": 3.0769230769230774e-05, "loss": 2.6187, "step": 62500 }, { "epoch": 3.88, "learning_rate": 3.0615384615384616e-05, "loss": 2.5985, "step": 63000 }, { "epoch": 3.91, "learning_rate": 3.0461538461538465e-05, "loss": 2.6227, "step": 63500 }, { "epoch": 3.94, "learning_rate": 3.030769230769231e-05, "loss": 2.6189, "step": 64000 }, { "epoch": 3.97, "learning_rate": 3.0153846153846155e-05, "loss": 2.607, "step": 64500 }, { "epoch": 4.0, "learning_rate": 3e-05, "loss": 2.6284, "step": 65000 }, { "epoch": 4.03, "learning_rate": 2.9846153846153846e-05, "loss": 2.3572, "step": 65500 }, { "epoch": 4.06, "learning_rate": 2.969230769230769e-05, "loss": 2.3599, "step": 66000 }, { "epoch": 4.09, "learning_rate": 2.9538461538461543e-05, "loss": 2.3719, "step": 66500 }, { "epoch": 4.12, "learning_rate": 2.938461538461539e-05, "loss": 2.3775, "step": 67000 }, { "epoch": 4.15, "learning_rate": 2.9230769230769234e-05, "loss": 2.3772, "step": 67500 }, { "epoch": 4.18, "learning_rate": 2.907692307692308e-05, "loss": 2.4074, "step": 68000 }, { "epoch": 4.22, "learning_rate": 2.8923076923076925e-05, "loss": 2.4128, "step": 68500 }, { "epoch": 4.25, "learning_rate": 2.876923076923077e-05, "loss": 2.3943, "step": 69000 }, { "epoch": 4.28, "learning_rate": 2.8615384615384615e-05, "loss": 2.3941, "step": 69500 }, { "epoch": 4.31, "learning_rate": 2.846153846153846e-05, "loss": 2.4156, "step": 70000 }, { "epoch": 4.34, "learning_rate": 2.8307692307692306e-05, "loss": 2.4033, "step": 70500 }, { "epoch": 4.37, "learning_rate": 2.8153846153846154e-05, "loss": 2.4138, "step": 71000 }, { "epoch": 4.4, "learning_rate": 2.8000000000000003e-05, "loss": 2.4045, "step": 71500 }, { "epoch": 4.43, "learning_rate": 2.7846153846153848e-05, "loss": 2.403, "step": 72000 }, { "epoch": 4.46, "learning_rate": 2.7692307692307694e-05, "loss": 2.4146, "step": 72500 }, { "epoch": 4.49, "learning_rate": 2.7538461538461542e-05, "loss": 2.4194, "step": 73000 }, { "epoch": 4.52, "learning_rate": 2.7384615384615387e-05, "loss": 2.421, "step": 73500 }, { "epoch": 4.55, "learning_rate": 2.7230769230769233e-05, "loss": 2.4465, "step": 74000 }, { "epoch": 4.58, "learning_rate": 2.7076923076923078e-05, "loss": 2.4323, "step": 74500 }, { "epoch": 4.62, "learning_rate": 2.6923076923076923e-05, "loss": 2.4401, "step": 75000 }, { "epoch": 4.65, "learning_rate": 2.676923076923077e-05, "loss": 2.4214, "step": 75500 }, { "epoch": 4.68, "learning_rate": 2.6615384615384614e-05, "loss": 2.4235, "step": 76000 }, { "epoch": 4.71, "learning_rate": 2.6461538461538466e-05, "loss": 2.4677, "step": 76500 }, { "epoch": 4.74, "learning_rate": 2.630769230769231e-05, "loss": 2.4496, "step": 77000 }, { "epoch": 4.77, "learning_rate": 2.6153846153846157e-05, "loss": 2.4291, "step": 77500 }, { "epoch": 4.8, "learning_rate": 2.6000000000000002e-05, "loss": 2.4532, "step": 78000 }, { "epoch": 4.83, "learning_rate": 2.5846153846153847e-05, "loss": 2.4601, "step": 78500 }, { "epoch": 4.86, "learning_rate": 2.5692307692307692e-05, "loss": 2.4492, "step": 79000 }, { "epoch": 4.89, "learning_rate": 2.5538461538461538e-05, "loss": 2.4744, "step": 79500 }, { "epoch": 4.92, "learning_rate": 2.5384615384615383e-05, "loss": 2.461, "step": 80000 }, { "epoch": 4.95, "learning_rate": 2.523076923076923e-05, "loss": 2.4562, "step": 80500 }, { "epoch": 4.98, "learning_rate": 2.5076923076923077e-05, "loss": 2.4677, "step": 81000 }, { "epoch": 5.02, "learning_rate": 2.4923076923076926e-05, "loss": 2.3597, "step": 81500 }, { "epoch": 5.05, "learning_rate": 2.476923076923077e-05, "loss": 2.21, "step": 82000 }, { "epoch": 5.08, "learning_rate": 2.461538461538462e-05, "loss": 2.2361, "step": 82500 }, { "epoch": 5.11, "learning_rate": 2.4461538461538465e-05, "loss": 2.2432, "step": 83000 }, { "epoch": 5.14, "learning_rate": 2.430769230769231e-05, "loss": 2.2396, "step": 83500 }, { "epoch": 5.17, "learning_rate": 2.4153846153846155e-05, "loss": 2.243, "step": 84000 }, { "epoch": 5.2, "learning_rate": 2.4e-05, "loss": 2.2377, "step": 84500 }, { "epoch": 5.23, "learning_rate": 2.384615384615385e-05, "loss": 2.2259, "step": 85000 }, { "epoch": 5.26, "learning_rate": 2.3692307692307695e-05, "loss": 2.2558, "step": 85500 }, { "epoch": 5.29, "learning_rate": 2.353846153846154e-05, "loss": 2.2454, "step": 86000 }, { "epoch": 5.32, "learning_rate": 2.3384615384615385e-05, "loss": 2.276, "step": 86500 }, { "epoch": 5.35, "learning_rate": 2.323076923076923e-05, "loss": 2.2689, "step": 87000 }, { "epoch": 5.38, "learning_rate": 2.307692307692308e-05, "loss": 2.273, "step": 87500 }, { "epoch": 5.42, "learning_rate": 2.2923076923076924e-05, "loss": 2.267, "step": 88000 }, { "epoch": 5.45, "learning_rate": 2.276923076923077e-05, "loss": 2.2673, "step": 88500 }, { "epoch": 5.48, "learning_rate": 2.2615384615384615e-05, "loss": 2.2708, "step": 89000 }, { "epoch": 5.51, "learning_rate": 2.246153846153846e-05, "loss": 2.2909, "step": 89500 }, { "epoch": 5.54, "learning_rate": 2.230769230769231e-05, "loss": 2.2623, "step": 90000 }, { "epoch": 5.57, "learning_rate": 2.2153846153846154e-05, "loss": 2.3009, "step": 90500 }, { "epoch": 5.6, "learning_rate": 2.2000000000000003e-05, "loss": 2.2974, "step": 91000 }, { "epoch": 5.63, "learning_rate": 2.1846153846153848e-05, "loss": 2.2947, "step": 91500 }, { "epoch": 5.66, "learning_rate": 2.1692307692307693e-05, "loss": 2.2825, "step": 92000 }, { "epoch": 5.69, "learning_rate": 2.1538461538461542e-05, "loss": 2.2875, "step": 92500 }, { "epoch": 5.72, "learning_rate": 2.1384615384615387e-05, "loss": 2.3044, "step": 93000 }, { "epoch": 5.75, "learning_rate": 2.1230769230769233e-05, "loss": 2.3037, "step": 93500 }, { "epoch": 5.78, "learning_rate": 2.1076923076923078e-05, "loss": 2.3085, "step": 94000 }, { "epoch": 5.82, "learning_rate": 2.0923076923076923e-05, "loss": 2.3279, "step": 94500 }, { "epoch": 5.85, "learning_rate": 2.0769230769230772e-05, "loss": 2.3185, "step": 95000 }, { "epoch": 5.88, "learning_rate": 2.0615384615384617e-05, "loss": 2.31, "step": 95500 }, { "epoch": 5.91, "learning_rate": 2.0461538461538462e-05, "loss": 2.3214, "step": 96000 }, { "epoch": 5.94, "learning_rate": 2.0307692307692308e-05, "loss": 2.301, "step": 96500 }, { "epoch": 5.97, "learning_rate": 2.0153846153846153e-05, "loss": 2.3112, "step": 97000 }, { "epoch": 6.0, "learning_rate": 2e-05, "loss": 2.3008, "step": 97500 }, { "epoch": 6.03, "learning_rate": 1.9846153846153847e-05, "loss": 2.0927, "step": 98000 }, { "epoch": 6.06, "learning_rate": 1.9692307692307692e-05, "loss": 2.0993, "step": 98500 }, { "epoch": 6.09, "learning_rate": 1.9538461538461537e-05, "loss": 2.096, "step": 99000 }, { "epoch": 6.12, "learning_rate": 1.9384615384615383e-05, "loss": 2.1266, "step": 99500 }, { "epoch": 6.15, "learning_rate": 1.923076923076923e-05, "loss": 2.1288, "step": 100000 }, { "epoch": 6.18, "learning_rate": 1.9076923076923077e-05, "loss": 2.1283, "step": 100500 }, { "epoch": 6.22, "learning_rate": 1.8923076923076925e-05, "loss": 2.1149, "step": 101000 }, { "epoch": 6.25, "learning_rate": 1.876923076923077e-05, "loss": 2.1298, "step": 101500 }, { "epoch": 6.28, "learning_rate": 1.8615384615384616e-05, "loss": 2.1446, "step": 102000 }, { "epoch": 6.31, "learning_rate": 1.8461538461538465e-05, "loss": 2.1571, "step": 102500 }, { "epoch": 6.34, "learning_rate": 1.830769230769231e-05, "loss": 2.1238, "step": 103000 }, { "epoch": 6.37, "learning_rate": 1.8153846153846155e-05, "loss": 2.1477, "step": 103500 }, { "epoch": 6.4, "learning_rate": 1.8e-05, "loss": 2.1577, "step": 104000 }, { "epoch": 6.43, "learning_rate": 1.7846153846153846e-05, "loss": 2.1405, "step": 104500 }, { "epoch": 6.46, "learning_rate": 1.7692307692307694e-05, "loss": 2.1526, "step": 105000 }, { "epoch": 6.49, "learning_rate": 1.753846153846154e-05, "loss": 2.1475, "step": 105500 }, { "epoch": 6.52, "learning_rate": 1.7384615384615385e-05, "loss": 2.1501, "step": 106000 }, { "epoch": 6.55, "learning_rate": 1.723076923076923e-05, "loss": 2.1552, "step": 106500 }, { "epoch": 6.58, "learning_rate": 1.7076923076923076e-05, "loss": 2.1491, "step": 107000 }, { "epoch": 6.62, "learning_rate": 1.6923076923076924e-05, "loss": 2.1729, "step": 107500 }, { "epoch": 6.65, "learning_rate": 1.676923076923077e-05, "loss": 2.1683, "step": 108000 }, { "epoch": 6.68, "learning_rate": 1.6615384615384615e-05, "loss": 2.1438, "step": 108500 }, { "epoch": 6.71, "learning_rate": 1.646153846153846e-05, "loss": 2.1581, "step": 109000 }, { "epoch": 6.74, "learning_rate": 1.630769230769231e-05, "loss": 2.1641, "step": 109500 }, { "epoch": 6.77, "learning_rate": 1.6153846153846154e-05, "loss": 2.1598, "step": 110000 }, { "epoch": 6.8, "learning_rate": 1.6000000000000003e-05, "loss": 2.158, "step": 110500 }, { "epoch": 6.83, "learning_rate": 1.5846153846153848e-05, "loss": 2.1557, "step": 111000 }, { "epoch": 6.86, "learning_rate": 1.5692307692307693e-05, "loss": 2.1795, "step": 111500 }, { "epoch": 6.89, "learning_rate": 1.553846153846154e-05, "loss": 2.1682, "step": 112000 }, { "epoch": 6.92, "learning_rate": 1.5384615384615387e-05, "loss": 2.17, "step": 112500 }, { "epoch": 6.95, "learning_rate": 1.5230769230769232e-05, "loss": 2.1608, "step": 113000 }, { "epoch": 6.98, "learning_rate": 1.5076923076923078e-05, "loss": 2.1838, "step": 113500 }, { "epoch": 7.02, "learning_rate": 1.4923076923076923e-05, "loss": 2.0964, "step": 114000 }, { "epoch": 7.05, "learning_rate": 1.4769230769230772e-05, "loss": 1.9989, "step": 114500 }, { "epoch": 7.08, "learning_rate": 1.4615384615384617e-05, "loss": 1.9878, "step": 115000 }, { "epoch": 7.11, "learning_rate": 1.4461538461538462e-05, "loss": 1.9903, "step": 115500 }, { "epoch": 7.14, "learning_rate": 1.4307692307692308e-05, "loss": 2.0201, "step": 116000 }, { "epoch": 7.17, "learning_rate": 1.4153846153846153e-05, "loss": 1.9994, "step": 116500 }, { "epoch": 7.2, "learning_rate": 1.4000000000000001e-05, "loss": 2.0244, "step": 117000 }, { "epoch": 7.23, "learning_rate": 1.3846153846153847e-05, "loss": 2.0113, "step": 117500 }, { "epoch": 7.26, "learning_rate": 1.3692307692307694e-05, "loss": 2.0214, "step": 118000 }, { "epoch": 7.29, "learning_rate": 1.3538461538461539e-05, "loss": 2.0313, "step": 118500 }, { "epoch": 7.32, "learning_rate": 1.3384615384615384e-05, "loss": 2.0371, "step": 119000 }, { "epoch": 7.35, "learning_rate": 1.3230769230769233e-05, "loss": 2.0231, "step": 119500 }, { "epoch": 7.38, "learning_rate": 1.3076923076923078e-05, "loss": 2.0159, "step": 120000 }, { "epoch": 7.42, "learning_rate": 1.2923076923076924e-05, "loss": 2.0376, "step": 120500 }, { "epoch": 7.45, "learning_rate": 1.2769230769230769e-05, "loss": 2.0012, "step": 121000 }, { "epoch": 7.48, "learning_rate": 1.2615384615384616e-05, "loss": 2.0265, "step": 121500 }, { "epoch": 7.51, "learning_rate": 1.2461538461538463e-05, "loss": 2.0141, "step": 122000 }, { "epoch": 7.54, "learning_rate": 1.230769230769231e-05, "loss": 2.0411, "step": 122500 }, { "epoch": 7.57, "learning_rate": 1.2153846153846155e-05, "loss": 2.0424, "step": 123000 }, { "epoch": 7.6, "learning_rate": 1.2e-05, "loss": 2.036, "step": 123500 }, { "epoch": 7.63, "learning_rate": 1.1846153846153847e-05, "loss": 2.0416, "step": 124000 }, { "epoch": 7.66, "learning_rate": 1.1692307692307693e-05, "loss": 2.058, "step": 124500 }, { "epoch": 7.69, "learning_rate": 1.153846153846154e-05, "loss": 2.0258, "step": 125000 }, { "epoch": 7.72, "learning_rate": 1.1384615384615385e-05, "loss": 2.0486, "step": 125500 }, { "epoch": 7.75, "learning_rate": 1.123076923076923e-05, "loss": 2.0433, "step": 126000 }, { "epoch": 7.78, "learning_rate": 1.1076923076923077e-05, "loss": 2.0413, "step": 126500 }, { "epoch": 7.82, "learning_rate": 1.0923076923076924e-05, "loss": 2.0454, "step": 127000 }, { "epoch": 7.85, "learning_rate": 1.0769230769230771e-05, "loss": 2.0756, "step": 127500 }, { "epoch": 7.88, "learning_rate": 1.0615384615384616e-05, "loss": 2.055, "step": 128000 }, { "epoch": 7.91, "learning_rate": 1.0461538461538462e-05, "loss": 2.0466, "step": 128500 }, { "epoch": 7.94, "learning_rate": 1.0307692307692309e-05, "loss": 2.0635, "step": 129000 }, { "epoch": 7.97, "learning_rate": 1.0153846153846154e-05, "loss": 2.0534, "step": 129500 }, { "epoch": 8.0, "learning_rate": 1e-05, "loss": 2.0431, "step": 130000 }, { "epoch": 8.03, "learning_rate": 9.846153846153846e-06, "loss": 1.9157, "step": 130500 }, { "epoch": 8.06, "learning_rate": 9.692307692307691e-06, "loss": 1.9128, "step": 131000 }, { "epoch": 8.09, "learning_rate": 9.538461538461538e-06, "loss": 1.9017, "step": 131500 }, { "epoch": 8.12, "learning_rate": 9.384615384615385e-06, "loss": 1.9096, "step": 132000 }, { "epoch": 8.15, "learning_rate": 9.230769230769232e-06, "loss": 1.9143, "step": 132500 }, { "epoch": 8.18, "learning_rate": 9.076923076923078e-06, "loss": 1.9189, "step": 133000 }, { "epoch": 8.22, "learning_rate": 8.923076923076923e-06, "loss": 1.9204, "step": 133500 }, { "epoch": 8.25, "learning_rate": 8.76923076923077e-06, "loss": 1.9375, "step": 134000 }, { "epoch": 8.28, "learning_rate": 8.615384615384615e-06, "loss": 1.9227, "step": 134500 }, { "epoch": 8.31, "learning_rate": 8.461538461538462e-06, "loss": 1.9247, "step": 135000 }, { "epoch": 8.34, "learning_rate": 8.307692307692307e-06, "loss": 1.9315, "step": 135500 }, { "epoch": 8.37, "learning_rate": 8.153846153846154e-06, "loss": 1.9205, "step": 136000 }, { "epoch": 8.4, "learning_rate": 8.000000000000001e-06, "loss": 1.9284, "step": 136500 }, { "epoch": 8.43, "learning_rate": 7.846153846153847e-06, "loss": 1.9429, "step": 137000 }, { "epoch": 8.46, "learning_rate": 7.692307692307694e-06, "loss": 1.9331, "step": 137500 }, { "epoch": 8.49, "learning_rate": 7.538461538461539e-06, "loss": 1.9289, "step": 138000 }, { "epoch": 8.52, "learning_rate": 7.384615384615386e-06, "loss": 1.9454, "step": 138500 }, { "epoch": 8.55, "learning_rate": 7.230769230769231e-06, "loss": 1.9316, "step": 139000 }, { "epoch": 8.58, "learning_rate": 7.076923076923076e-06, "loss": 1.9423, "step": 139500 }, { "epoch": 8.62, "learning_rate": 6.923076923076923e-06, "loss": 1.9649, "step": 140000 }, { "epoch": 8.65, "learning_rate": 6.7692307692307695e-06, "loss": 1.9378, "step": 140500 }, { "epoch": 8.68, "learning_rate": 6.6153846153846165e-06, "loss": 1.9322, "step": 141000 }, { "epoch": 8.71, "learning_rate": 6.461538461538462e-06, "loss": 1.962, "step": 141500 }, { "epoch": 8.74, "learning_rate": 6.307692307692308e-06, "loss": 1.9442, "step": 142000 }, { "epoch": 8.77, "learning_rate": 6.153846153846155e-06, "loss": 1.9509, "step": 142500 }, { "epoch": 8.8, "learning_rate": 6e-06, "loss": 1.9582, "step": 143000 }, { "epoch": 8.83, "learning_rate": 5.846153846153846e-06, "loss": 1.9279, "step": 143500 }, { "epoch": 8.86, "learning_rate": 5.692307692307692e-06, "loss": 1.9428, "step": 144000 }, { "epoch": 8.89, "learning_rate": 5.5384615384615385e-06, "loss": 1.9471, "step": 144500 }, { "epoch": 8.92, "learning_rate": 5.3846153846153855e-06, "loss": 1.9205, "step": 145000 }, { "epoch": 8.95, "learning_rate": 5.230769230769231e-06, "loss": 1.9582, "step": 145500 }, { "epoch": 8.98, "learning_rate": 5.076923076923077e-06, "loss": 1.9541, "step": 146000 }, { "epoch": 9.02, "learning_rate": 4.923076923076923e-06, "loss": 1.9153, "step": 146500 }, { "epoch": 9.05, "learning_rate": 4.769230769230769e-06, "loss": 1.8547, "step": 147000 }, { "epoch": 9.08, "learning_rate": 4.615384615384616e-06, "loss": 1.863, "step": 147500 }, { "epoch": 9.11, "learning_rate": 4.4615384615384614e-06, "loss": 1.8348, "step": 148000 }, { "epoch": 9.14, "learning_rate": 4.3076923076923076e-06, "loss": 1.8621, "step": 148500 }, { "epoch": 9.17, "learning_rate": 4.153846153846154e-06, "loss": 1.8745, "step": 149000 }, { "epoch": 9.2, "learning_rate": 4.000000000000001e-06, "loss": 1.8555, "step": 149500 }, { "epoch": 9.23, "learning_rate": 3.846153846153847e-06, "loss": 1.8731, "step": 150000 }, { "epoch": 9.26, "learning_rate": 3.692307692307693e-06, "loss": 1.8495, "step": 150500 }, { "epoch": 9.29, "learning_rate": 3.538461538461538e-06, "loss": 1.8596, "step": 151000 }, { "epoch": 9.32, "learning_rate": 3.3846153846153848e-06, "loss": 1.8625, "step": 151500 }, { "epoch": 9.35, "learning_rate": 3.230769230769231e-06, "loss": 1.8601, "step": 152000 }, { "epoch": 9.38, "learning_rate": 3.0769230769230774e-06, "loss": 1.8519, "step": 152500 }, { "epoch": 9.42, "learning_rate": 2.923076923076923e-06, "loss": 1.8648, "step": 153000 }, { "epoch": 9.45, "learning_rate": 2.7692307692307693e-06, "loss": 1.859, "step": 153500 }, { "epoch": 9.48, "learning_rate": 2.6153846153846154e-06, "loss": 1.8728, "step": 154000 }, { "epoch": 9.51, "learning_rate": 2.4615384615384615e-06, "loss": 1.8536, "step": 154500 }, { "epoch": 9.54, "learning_rate": 2.307692307692308e-06, "loss": 1.8706, "step": 155000 }, { "epoch": 9.57, "learning_rate": 2.1538461538461538e-06, "loss": 1.8446, "step": 155500 }, { "epoch": 9.6, "learning_rate": 2.0000000000000003e-06, "loss": 1.8545, "step": 156000 }, { "epoch": 9.63, "learning_rate": 1.8461538461538465e-06, "loss": 1.8396, "step": 156500 }, { "epoch": 9.66, "learning_rate": 1.6923076923076924e-06, "loss": 1.8661, "step": 157000 }, { "epoch": 9.69, "learning_rate": 1.5384615384615387e-06, "loss": 1.8567, "step": 157500 }, { "epoch": 9.72, "learning_rate": 1.3846153846153846e-06, "loss": 1.8641, "step": 158000 }, { "epoch": 9.75, "learning_rate": 1.2307692307692308e-06, "loss": 1.8823, "step": 158500 }, { "epoch": 9.78, "learning_rate": 1.0769230769230769e-06, "loss": 1.8767, "step": 159000 }, { "epoch": 9.82, "learning_rate": 9.230769230769232e-07, "loss": 1.8615, "step": 159500 }, { "epoch": 9.85, "learning_rate": 7.692307692307694e-07, "loss": 1.8743, "step": 160000 }, { "epoch": 9.88, "learning_rate": 6.153846153846154e-07, "loss": 1.8741, "step": 160500 }, { "epoch": 9.91, "learning_rate": 4.615384615384616e-07, "loss": 1.8367, "step": 161000 }, { "epoch": 9.94, "learning_rate": 3.076923076923077e-07, "loss": 1.868, "step": 161500 }, { "epoch": 9.97, "learning_rate": 1.5384615384615385e-07, "loss": 1.8581, "step": 162000 }, { "epoch": 10.0, "learning_rate": 0.0, "loss": 1.8659, "step": 162500 }, { "epoch": 10.0, "step": 162500, "total_flos": 9.755546820784128e+16, "train_loss": 2.4223546108774037, "train_runtime": 17439.7175, "train_samples_per_second": 37.271, "train_steps_per_second": 9.318 } ], "max_steps": 162500, "num_train_epochs": 10, "total_flos": 9.755546820784128e+16, "trial_name": null, "trial_params": null }