| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 156, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02564102564102564, | |
| "grad_norm": 4.75, | |
| "learning_rate": 0.0, | |
| "loss": 2.8786, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.05128205128205128, | |
| "grad_norm": 8.0625, | |
| "learning_rate": 2.5e-06, | |
| "loss": 2.9525, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.07692307692307693, | |
| "grad_norm": 11.6875, | |
| "learning_rate": 5e-06, | |
| "loss": 2.7153, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.10256410256410256, | |
| "grad_norm": 12.4375, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 2.5826, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.1282051282051282, | |
| "grad_norm": 10.375, | |
| "learning_rate": 1e-05, | |
| "loss": 2.5254, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.15384615384615385, | |
| "grad_norm": 8.875, | |
| "learning_rate": 9.998932083939657e-06, | |
| "loss": 2.3523, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.1794871794871795, | |
| "grad_norm": 8.8125, | |
| "learning_rate": 9.995728791936505e-06, | |
| "loss": 2.2738, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.20512820512820512, | |
| "grad_norm": 7.28125, | |
| "learning_rate": 9.990391492329341e-06, | |
| "loss": 2.2553, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.23076923076923078, | |
| "grad_norm": 6.90625, | |
| "learning_rate": 9.98292246503335e-06, | |
| "loss": 2.1269, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.2564102564102564, | |
| "grad_norm": 6.875, | |
| "learning_rate": 9.973324900566214e-06, | |
| "loss": 2.1726, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.28205128205128205, | |
| "grad_norm": 7.25, | |
| "learning_rate": 9.961602898685225e-06, | |
| "loss": 2.087, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 6.8125, | |
| "learning_rate": 9.947761466636014e-06, | |
| "loss": 2.0203, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 9.931806517013612e-06, | |
| "loss": 2.0242, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.358974358974359, | |
| "grad_norm": 6.53125, | |
| "learning_rate": 9.913744865236798e-06, | |
| "loss": 1.9997, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 6.34375, | |
| "learning_rate": 9.893584226636773e-06, | |
| "loss": 1.8334, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.41025641025641024, | |
| "grad_norm": 6.40625, | |
| "learning_rate": 9.871333213161438e-06, | |
| "loss": 1.9528, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.4358974358974359, | |
| "grad_norm": 6.0, | |
| "learning_rate": 9.847001329696653e-06, | |
| "loss": 1.8608, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.46153846153846156, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 9.820598970006068e-06, | |
| "loss": 1.8053, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.48717948717948717, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 9.792137412291265e-06, | |
| "loss": 1.8232, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.5128205128205128, | |
| "grad_norm": 5.34375, | |
| "learning_rate": 9.761628814374074e-06, | |
| "loss": 1.7059, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.5384615384615384, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 9.729086208503174e-06, | |
| "loss": 1.9746, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.5641025641025641, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 9.694523495787149e-06, | |
| "loss": 1.6262, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.5897435897435898, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 9.657955440256396e-06, | |
| "loss": 1.8171, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 9.619397662556434e-06, | |
| "loss": 1.7618, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.6410256410256411, | |
| "grad_norm": 5.25, | |
| "learning_rate": 9.578866633275289e-06, | |
| "loss": 1.711, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 9.536379665907801e-06, | |
| "loss": 1.5066, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.6923076923076923, | |
| "grad_norm": 5.625, | |
| "learning_rate": 9.491954909459895e-06, | |
| "loss": 1.4406, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.717948717948718, | |
| "grad_norm": 5.5, | |
| "learning_rate": 9.445611340695926e-06, | |
| "loss": 1.558, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.7435897435897436, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 9.397368756032445e-06, | |
| "loss": 1.4596, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 9.347247763081834e-06, | |
| "loss": 1.6387, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.7948717948717948, | |
| "grad_norm": 8.75, | |
| "learning_rate": 9.295269771849426e-06, | |
| "loss": 1.5038, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.8205128205128205, | |
| "grad_norm": 7.0, | |
| "learning_rate": 9.241456985587868e-06, | |
| "loss": 2.154, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.8461538461538461, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 9.185832391312644e-06, | |
| "loss": 1.9146, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.8717948717948718, | |
| "grad_norm": 6.8125, | |
| "learning_rate": 9.12841974998278e-06, | |
| "loss": 2.3066, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.8974358974358975, | |
| "grad_norm": 7.0, | |
| "learning_rate": 9.069243586350976e-06, | |
| "loss": 2.2234, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 9.008329178487442e-06, | |
| "loss": 2.9886, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.9487179487179487, | |
| "grad_norm": 9.5, | |
| "learning_rate": 8.94570254698197e-06, | |
| "loss": 4.0233, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.9743589743589743, | |
| "grad_norm": 13.1875, | |
| "learning_rate": 8.881390443828788e-06, | |
| "loss": 4.4944, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 45.75, | |
| "learning_rate": 8.815420340999034e-06, | |
| "loss": 6.7771, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 1.0256410256410255, | |
| "grad_norm": 4.03125, | |
| "learning_rate": 8.747820418705632e-06, | |
| "loss": 2.741, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.0512820512820513, | |
| "grad_norm": 23.75, | |
| "learning_rate": 8.67861955336566e-06, | |
| "loss": 2.7361, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 1.0769230769230769, | |
| "grad_norm": 41.25, | |
| "learning_rate": 8.607847305265312e-06, | |
| "loss": 2.4577, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 1.1025641025641026, | |
| "grad_norm": 36.5, | |
| "learning_rate": 8.535533905932739e-06, | |
| "loss": 2.0977, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 1.1282051282051282, | |
| "grad_norm": 11.625, | |
| "learning_rate": 8.461710245224149e-06, | |
| "loss": 2.0235, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 1.1538461538461537, | |
| "grad_norm": 12.4375, | |
| "learning_rate": 8.386407858128707e-06, | |
| "loss": 1.8987, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 1.1794871794871795, | |
| "grad_norm": 11.3125, | |
| "learning_rate": 8.309658911297833e-06, | |
| "loss": 1.7952, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 1.205128205128205, | |
| "grad_norm": 7.0, | |
| "learning_rate": 8.231496189304704e-06, | |
| "loss": 1.8796, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 1.2307692307692308, | |
| "grad_norm": 5.5, | |
| "learning_rate": 8.151953080639777e-06, | |
| "loss": 1.8016, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 1.2564102564102564, | |
| "grad_norm": 5.875, | |
| "learning_rate": 8.071063563448341e-06, | |
| "loss": 1.8173, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 1.282051282051282, | |
| "grad_norm": 5.125, | |
| "learning_rate": 7.988862191016204e-06, | |
| "loss": 1.8295, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.3076923076923077, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 7.905384077009693e-06, | |
| "loss": 1.7912, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 7.820664880476257e-06, | |
| "loss": 1.7695, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.358974358974359, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 7.734740790612137e-06, | |
| "loss": 1.7942, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.3846153846153846, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 7.647648511303545e-06, | |
| "loss": 1.7302, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.4102564102564101, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 7.559425245448006e-06, | |
| "loss": 1.6814, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.435897435897436, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 7.470108679062521e-06, | |
| "loss": 1.6977, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.4615384615384617, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 7.379736965185369e-06, | |
| "loss": 1.6699, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.4871794871794872, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 7.288348707578409e-06, | |
| "loss": 1.6252, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.5128205128205128, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 7.195982944236853e-06, | |
| "loss": 1.5826, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 7.102679130713538e-06, | |
| "loss": 1.7134, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.564102564102564, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 7.008477123264849e-06, | |
| "loss": 1.5422, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.5897435897435899, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 6.913417161825449e-06, | |
| "loss": 1.6654, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.6153846153846154, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 6.817539852819149e-06, | |
| "loss": 1.7428, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.641025641025641, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 6.720886151813194e-06, | |
| "loss": 1.5221, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 6.6234973460234184e-06, | |
| "loss": 1.4163, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.6923076923076923, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 6.525415036677745e-06, | |
| "loss": 1.3288, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.717948717948718, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 6.426681121245527e-06, | |
| "loss": 1.3884, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.7435897435897436, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 6.327337775540362e-06, | |
| "loss": 1.4421, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.7692307692307692, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 6.227427435703997e-06, | |
| "loss": 1.5454, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.7948717948717947, | |
| "grad_norm": 4.875, | |
| "learning_rate": 6.126992780079032e-06, | |
| "loss": 1.5869, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.8205128205128205, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 6.026076710978172e-06, | |
| "loss": 2.0383, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.8461538461538463, | |
| "grad_norm": 5.5, | |
| "learning_rate": 5.924722336357793e-06, | |
| "loss": 1.7922, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.8717948717948718, | |
| "grad_norm": 6.75, | |
| "learning_rate": 5.82297295140367e-06, | |
| "loss": 2.1761, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.8974358974358974, | |
| "grad_norm": 6.375, | |
| "learning_rate": 5.720872020036734e-06, | |
| "loss": 2.1503, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.9230769230769231, | |
| "grad_norm": 6.96875, | |
| "learning_rate": 5.61846315634674e-06, | |
| "loss": 2.8915, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.9487179487179487, | |
| "grad_norm": 7.25, | |
| "learning_rate": 5.515790105961785e-06, | |
| "loss": 3.8878, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.9743589743589745, | |
| "grad_norm": 11.125, | |
| "learning_rate": 5.412896727361663e-06, | |
| "loss": 4.3157, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 38.5, | |
| "learning_rate": 5.309826973142974e-06, | |
| "loss": 6.2624, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 2.0256410256410255, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 5.206624871244066e-06, | |
| "loss": 2.6988, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 2.051282051282051, | |
| "grad_norm": 16.125, | |
| "learning_rate": 5.103334506137773e-06, | |
| "loss": 2.6872, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.076923076923077, | |
| "grad_norm": 27.0, | |
| "learning_rate": 5e-06, | |
| "loss": 2.3881, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 2.1025641025641026, | |
| "grad_norm": 29.0, | |
| "learning_rate": 4.89666549386223e-06, | |
| "loss": 2.1485, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 2.128205128205128, | |
| "grad_norm": 23.75, | |
| "learning_rate": 4.793375128755934e-06, | |
| "loss": 2.1047, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 2.1538461538461537, | |
| "grad_norm": 19.375, | |
| "learning_rate": 4.690173026857028e-06, | |
| "loss": 1.9177, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 2.1794871794871793, | |
| "grad_norm": 14.0, | |
| "learning_rate": 4.587103272638339e-06, | |
| "loss": 1.7993, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 2.2051282051282053, | |
| "grad_norm": 7.625, | |
| "learning_rate": 4.4842098940382155e-06, | |
| "loss": 1.8896, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 2.230769230769231, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 4.381536843653262e-06, | |
| "loss": 1.761, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 2.2564102564102564, | |
| "grad_norm": 7.9375, | |
| "learning_rate": 4.279127979963266e-06, | |
| "loss": 1.8078, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 2.282051282051282, | |
| "grad_norm": 7.84375, | |
| "learning_rate": 4.17702704859633e-06, | |
| "loss": 1.7515, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 2.3076923076923075, | |
| "grad_norm": 7.0625, | |
| "learning_rate": 4.075277663642208e-06, | |
| "loss": 1.7459, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.3333333333333335, | |
| "grad_norm": 6.34375, | |
| "learning_rate": 3.973923289021829e-06, | |
| "loss": 1.6839, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 2.358974358974359, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 3.8730072199209705e-06, | |
| "loss": 1.7688, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 2.3846153846153846, | |
| "grad_norm": 5.53125, | |
| "learning_rate": 3.7725725642960047e-06, | |
| "loss": 1.6122, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 2.41025641025641, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 3.67266222445964e-06, | |
| "loss": 1.7193, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 2.435897435897436, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 3.573318878754475e-06, | |
| "loss": 1.6538, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 2.4615384615384617, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 3.4745849633222566e-06, | |
| "loss": 1.622, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 2.4871794871794872, | |
| "grad_norm": 4.875, | |
| "learning_rate": 3.3765026539765832e-06, | |
| "loss": 1.5624, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 2.5128205128205128, | |
| "grad_norm": 4.625, | |
| "learning_rate": 3.2791138481868084e-06, | |
| "loss": 1.552, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 2.5384615384615383, | |
| "grad_norm": 4.75, | |
| "learning_rate": 3.1824601471808504e-06, | |
| "loss": 1.6924, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 2.564102564102564, | |
| "grad_norm": 5.0, | |
| "learning_rate": 3.0865828381745515e-06, | |
| "loss": 1.4784, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.58974358974359, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 2.991522876735154e-06, | |
| "loss": 1.6101, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 2.6153846153846154, | |
| "grad_norm": 4.875, | |
| "learning_rate": 2.8973208692864623e-06, | |
| "loss": 1.6149, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 2.641025641025641, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 2.804017055763149e-06, | |
| "loss": 1.5607, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 2.711651292421593e-06, | |
| "loss": 1.3517, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 2.6923076923076925, | |
| "grad_norm": 4.78125, | |
| "learning_rate": 2.6202630348146323e-06, | |
| "loss": 1.362, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 2.717948717948718, | |
| "grad_norm": 5.125, | |
| "learning_rate": 2.529891320937481e-06, | |
| "loss": 1.232, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 2.7435897435897436, | |
| "grad_norm": 5.25, | |
| "learning_rate": 2.4405747545519966e-06, | |
| "loss": 1.5141, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 2.769230769230769, | |
| "grad_norm": 5.28125, | |
| "learning_rate": 2.352351488696457e-06, | |
| "loss": 1.4774, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 2.7948717948717947, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 2.265259209387867e-06, | |
| "loss": 1.6229, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 2.8205128205128203, | |
| "grad_norm": 5.875, | |
| "learning_rate": 2.179335119523745e-06, | |
| "loss": 1.9325, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.8461538461538463, | |
| "grad_norm": 6.6875, | |
| "learning_rate": 2.094615922990309e-06, | |
| "loss": 1.8139, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 2.871794871794872, | |
| "grad_norm": 8.3125, | |
| "learning_rate": 2.0111378089837958e-06, | |
| "loss": 2.1465, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 2.8974358974358974, | |
| "grad_norm": 9.6875, | |
| "learning_rate": 1.928936436551661e-06, | |
| "loss": 2.1325, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.9230769230769234, | |
| "grad_norm": 9.1875, | |
| "learning_rate": 1.848046919360225e-06, | |
| "loss": 2.9035, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.948717948717949, | |
| "grad_norm": 7.3125, | |
| "learning_rate": 1.7685038106952952e-06, | |
| "loss": 3.8152, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.9743589743589745, | |
| "grad_norm": 11.1875, | |
| "learning_rate": 1.6903410887021676e-06, | |
| "loss": 4.2236, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 38.25, | |
| "learning_rate": 1.6135921418712959e-06, | |
| "loss": 6.2164, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 3.0256410256410255, | |
| "grad_norm": 3.875, | |
| "learning_rate": 1.5382897547758513e-06, | |
| "loss": 2.6836, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 3.051282051282051, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 1.4644660940672628e-06, | |
| "loss": 2.5084, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 3.076923076923077, | |
| "grad_norm": 8.125, | |
| "learning_rate": 1.3921526947346902e-06, | |
| "loss": 2.0748, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 3.1025641025641026, | |
| "grad_norm": 6.9375, | |
| "learning_rate": 1.321380446634342e-06, | |
| "loss": 1.8859, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 3.128205128205128, | |
| "grad_norm": 7.25, | |
| "learning_rate": 1.2521795812943704e-06, | |
| "loss": 1.9025, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 3.1538461538461537, | |
| "grad_norm": 6.375, | |
| "learning_rate": 1.1845796590009684e-06, | |
| "loss": 1.7788, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 3.1794871794871793, | |
| "grad_norm": 7.28125, | |
| "learning_rate": 1.118609556171213e-06, | |
| "loss": 1.727, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 3.2051282051282053, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.0542974530180327e-06, | |
| "loss": 1.7812, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 3.230769230769231, | |
| "grad_norm": 7.375, | |
| "learning_rate": 9.916708215125586e-07, | |
| "loss": 1.7469, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 3.2564102564102564, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 9.307564136490255e-07, | |
| "loss": 1.7702, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 3.282051282051282, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 8.715802500172215e-07, | |
| "loss": 1.7148, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 3.3076923076923075, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 8.141676086873574e-07, | |
| "loss": 1.7481, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 7.585430144121319e-07, | |
| "loss": 1.6527, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 3.358974358974359, | |
| "grad_norm": 4.375, | |
| "learning_rate": 7.047302281505735e-07, | |
| "loss": 1.7278, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 3.3846153846153846, | |
| "grad_norm": 4.625, | |
| "learning_rate": 6.527522369181655e-07, | |
| "loss": 1.6841, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 3.41025641025641, | |
| "grad_norm": 4.875, | |
| "learning_rate": 6.026312439675553e-07, | |
| "loss": 1.6241, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 3.435897435897436, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 5.543886593040737e-07, | |
| "loss": 1.6332, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 3.4615384615384617, | |
| "grad_norm": 4.375, | |
| "learning_rate": 5.080450905401057e-07, | |
| "loss": 1.587, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 3.4871794871794872, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 4.6362033409220077e-07, | |
| "loss": 1.6122, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 3.5128205128205128, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 4.211333667247125e-07, | |
| "loss": 1.5133, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 3.5384615384615383, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 3.8060233744356634e-07, | |
| "loss": 1.7055, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 3.564102564102564, | |
| "grad_norm": 4.75, | |
| "learning_rate": 3.420445597436056e-07, | |
| "loss": 1.4768, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 3.58974358974359, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 3.0547650421285216e-07, | |
| "loss": 1.6144, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 3.6153846153846154, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 2.7091379149682683e-07, | |
| "loss": 1.6258, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 3.641025641025641, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 2.3837118562592799e-07, | |
| "loss": 1.47, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 3.6666666666666665, | |
| "grad_norm": 4.5, | |
| "learning_rate": 2.0786258770873647e-07, | |
| "loss": 1.3829, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 3.6923076923076925, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.7940102999393194e-07, | |
| "loss": 1.3299, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 3.717948717948718, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.5299867030334815e-07, | |
| "loss": 1.4538, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 3.7435897435897436, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.286667868385627e-07, | |
| "loss": 1.269, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 3.769230769230769, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.0641577336322761e-07, | |
| "loss": 1.4599, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 3.7948717948717947, | |
| "grad_norm": 5.25, | |
| "learning_rate": 8.625513476320291e-08, | |
| "loss": 1.5192, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 3.8205128205128203, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 6.819348298638839e-08, | |
| "loss": 1.986, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 3.8461538461538463, | |
| "grad_norm": 6.0, | |
| "learning_rate": 5.223853336398632e-08, | |
| "loss": 1.8131, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.871794871794872, | |
| "grad_norm": 7.0625, | |
| "learning_rate": 3.839710131477492e-08, | |
| "loss": 2.138, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 3.8974358974358974, | |
| "grad_norm": 8.5, | |
| "learning_rate": 2.6675099433787212e-08, | |
| "loss": 2.1238, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 3.9230769230769234, | |
| "grad_norm": 8.5, | |
| "learning_rate": 1.7077534966650767e-08, | |
| "loss": 2.887, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 3.948717948717949, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 9.608507670659239e-09, | |
| "loss": 3.8071, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 3.9743589743589745, | |
| "grad_norm": 11.1875, | |
| "learning_rate": 4.2712080634949024e-09, | |
| "loss": 4.2345, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 38.0, | |
| "learning_rate": 1.0679160603449533e-09, | |
| "loss": 6.1763, | |
| "step": 156 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 156, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 39, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.786072307381043e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |