| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 18750, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 13.29955005645752, | |
| "learning_rate": 1e-5, | |
| "loss": 3.9042, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 1.7849366664886475, | |
| "learning_rate": 2e-5, | |
| "loss": 0.6647, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1.459277868270874, | |
| "learning_rate": 3e-5, | |
| "loss": 0.5227, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1.3880829811096191, | |
| "learning_rate": 4e-5, | |
| "loss": 0.5149, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 1.3338109254837036, | |
| "learning_rate": 5e-5, | |
| "loss": 0.4984, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.065587043762207, | |
| "learning_rate": 4.972602739726028e-5, | |
| "loss": 0.4847, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 1.0945565700531006, | |
| "learning_rate": 4.945205479452055e-5, | |
| "loss": 0.4615, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1.0960415601730347, | |
| "learning_rate": 4.917808219178082e-5, | |
| "loss": 0.4504, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 1.0467826128005981, | |
| "learning_rate": 4.89041095890411e-5, | |
| "loss": 0.4415, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 1.0690410137176514, | |
| "learning_rate": 4.863013698630137e-5, | |
| "loss": 0.4497, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.9305112957954407, | |
| "learning_rate": 4.835616438356165e-5, | |
| "loss": 0.4363, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.8500822186470032, | |
| "learning_rate": 4.808219178082192e-5, | |
| "loss": 0.4306, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 1.2137833833694458, | |
| "learning_rate": 4.780821917808219e-5, | |
| "loss": 0.423, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.9352328777313232, | |
| "learning_rate": 4.753424657534247e-5, | |
| "loss": 0.4161, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 1.1027398109436035, | |
| "learning_rate": 4.726027397260274e-5, | |
| "loss": 0.4083, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.924199104309082, | |
| "learning_rate": 4.698630136986302e-5, | |
| "loss": 0.4254, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.8912659883499146, | |
| "learning_rate": 4.671232876712329e-5, | |
| "loss": 0.3918, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.0025393962860107, | |
| "learning_rate": 4.643835616438356e-5, | |
| "loss": 0.4046, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.9362453818321228, | |
| "learning_rate": 4.616438356164384e-5, | |
| "loss": 0.3979, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.8841680884361267, | |
| "learning_rate": 4.589041095890411e-5, | |
| "loss": 0.3938, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.898572564125061, | |
| "learning_rate": 4.561643835616439e-5, | |
| "loss": 0.3974, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.9011989831924438, | |
| "learning_rate": 4.534246575342466e-5, | |
| "loss": 0.3955, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.913512110710144, | |
| "learning_rate": 4.506849315068493e-5, | |
| "loss": 0.3842, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.9077229499816895, | |
| "learning_rate": 4.479452054794521e-5, | |
| "loss": 0.3756, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.8107369542121887, | |
| "learning_rate": 4.452054794520548e-5, | |
| "loss": 0.3781, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.962982714176178, | |
| "learning_rate": 4.424657534246576e-5, | |
| "loss": 0.3728, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.9333651065826416, | |
| "learning_rate": 4.3972602739726035e-5, | |
| "loss": 0.3731, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.9969388246536255, | |
| "learning_rate": 4.36986301369863e-5, | |
| "loss": 0.3793, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.8959200978279114, | |
| "learning_rate": 4.342465753424658e-5, | |
| "loss": 0.3787, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.8185614943504333, | |
| "learning_rate": 4.3150684931506855e-5, | |
| "loss": 0.3626, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 1.1243007183074951, | |
| "learning_rate": 4.2876712328767126e-5, | |
| "loss": 0.3681, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 1.0251150131225586, | |
| "learning_rate": 4.2602739726027404e-5, | |
| "loss": 0.3609, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.8459119200706482, | |
| "learning_rate": 4.232876712328767e-5, | |
| "loss": 0.3608, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 0.9018300175666809, | |
| "learning_rate": 4.2054794520547946e-5, | |
| "loss": 0.3779, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.804707407951355, | |
| "learning_rate": 4.1780821917808224e-5, | |
| "loss": 0.3674, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.8781819343566895, | |
| "learning_rate": 4.1506849315068495e-5, | |
| "loss": 0.3635, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.9687257409095764, | |
| "learning_rate": 4.123287671232877e-5, | |
| "loss": 0.3617, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 0.7628118991851807, | |
| "learning_rate": 4.0958904109589044e-5, | |
| "loss": 0.3352, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.9802232980728149, | |
| "learning_rate": 4.0684931506849315e-5, | |
| "loss": 0.3244, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 0.7366902828216553, | |
| "learning_rate": 4.041095890410959e-5, | |
| "loss": 0.3237, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 0.8844860196113586, | |
| "learning_rate": 4.0136986301369864e-5, | |
| "loss": 0.3296, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.692650556564331, | |
| "learning_rate": 3.9863013698630135e-5, | |
| "loss": 0.3165, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 0.8171700239181519, | |
| "learning_rate": 3.958904109589041e-5, | |
| "loss": 0.323, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 0.9350169897079468, | |
| "learning_rate": 3.9315068493150684e-5, | |
| "loss": 0.3259, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.9551327228546143, | |
| "learning_rate": 3.904109589041096e-5, | |
| "loss": 0.3252, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 0.8646096587181091, | |
| "learning_rate": 3.8767123287671233e-5, | |
| "loss": 0.3267, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 0.8012389540672302, | |
| "learning_rate": 3.8493150684931505e-5, | |
| "loss": 0.3149, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 0.833848774433136, | |
| "learning_rate": 3.821917808219178e-5, | |
| "loss": 0.3164, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 0.7836089730262756, | |
| "learning_rate": 3.7945205479452054e-5, | |
| "loss": 0.3206, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 0.8694811463356018, | |
| "learning_rate": 3.767123287671233e-5, | |
| "loss": 0.3187, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 0.8749567866325378, | |
| "learning_rate": 3.739726027397261e-5, | |
| "loss": 0.3165, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 0.8689484596252441, | |
| "learning_rate": 3.7123287671232874e-5, | |
| "loss": 0.3154, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 0.8809706568717957, | |
| "learning_rate": 3.684931506849315e-5, | |
| "loss": 0.3301, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 0.8677769899368286, | |
| "learning_rate": 3.657534246575342e-5, | |
| "loss": 0.3184, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 0.8212382793426514, | |
| "learning_rate": 3.63013698630137e-5, | |
| "loss": 0.3181, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "grad_norm": 0.8636347651481628, | |
| "learning_rate": 3.602739726027398e-5, | |
| "loss": 0.3138, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 0.8136293292045593, | |
| "learning_rate": 3.575342465753424e-5, | |
| "loss": 0.3156, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "grad_norm": 0.7700251936912537, | |
| "learning_rate": 3.547945205479452e-5, | |
| "loss": 0.3179, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 0.7282480597496033, | |
| "learning_rate": 3.52054794520548e-5, | |
| "loss": 0.3188, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 0.7657186388969421, | |
| "learning_rate": 3.493150684931507e-5, | |
| "loss": 0.3137, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 0.8558144569396973, | |
| "learning_rate": 3.465753424657535e-5, | |
| "loss": 0.3192, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 0.7496147751808167, | |
| "learning_rate": 3.438356164383562e-5, | |
| "loss": 0.3175, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "grad_norm": 0.9365683794021606, | |
| "learning_rate": 3.410958904109589e-5, | |
| "loss": 0.3124, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.8127835392951965, | |
| "learning_rate": 3.383561643835617e-5, | |
| "loss": 0.3056, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 0.819684624671936, | |
| "learning_rate": 3.356164383561644e-5, | |
| "loss": 0.3144, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 0.7603724598884583, | |
| "learning_rate": 3.328767123287672e-5, | |
| "loss": 0.315, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "grad_norm": 0.8054817318916321, | |
| "learning_rate": 3.301369863013699e-5, | |
| "loss": 0.3073, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "grad_norm": 0.758423924446106, | |
| "learning_rate": 3.273972602739726e-5, | |
| "loss": 0.312, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 0.8245046138763428, | |
| "learning_rate": 3.246575342465754e-5, | |
| "loss": 0.3125, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "grad_norm": 0.7906696796417236, | |
| "learning_rate": 3.219178082191781e-5, | |
| "loss": 0.3009, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 0.8566040992736816, | |
| "learning_rate": 3.1917808219178086e-5, | |
| "loss": 0.3043, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 0.7341597080230713, | |
| "learning_rate": 3.164383561643836e-5, | |
| "loss": 0.309, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "grad_norm": 0.7561280131340027, | |
| "learning_rate": 3.136986301369863e-5, | |
| "loss": 0.3051, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 0.7900431156158447, | |
| "learning_rate": 3.1095890410958906e-5, | |
| "loss": 0.3093, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.880424976348877, | |
| "learning_rate": 3.082191780821918e-5, | |
| "loss": 0.3058, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "grad_norm": 0.8830358982086182, | |
| "learning_rate": 3.0547945205479455e-5, | |
| "loss": 0.2673, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.6983394026756287, | |
| "learning_rate": 3.0273972602739726e-5, | |
| "loss": 0.2739, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 0.8467246890068054, | |
| "learning_rate": 3e-5, | |
| "loss": 0.2694, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "grad_norm": 0.8425388932228088, | |
| "learning_rate": 2.9726027397260275e-5, | |
| "loss": 0.2698, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "grad_norm": 0.6956115365028381, | |
| "learning_rate": 2.945205479452055e-5, | |
| "loss": 0.2616, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 0.9649244546890259, | |
| "learning_rate": 2.9178082191780824e-5, | |
| "loss": 0.2763, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "grad_norm": 0.7081593871116638, | |
| "learning_rate": 2.8904109589041095e-5, | |
| "loss": 0.2683, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "grad_norm": 0.9411781430244446, | |
| "learning_rate": 2.863013698630137e-5, | |
| "loss": 0.2621, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 0.8201924562454224, | |
| "learning_rate": 2.8356164383561644e-5, | |
| "loss": 0.2701, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 0.8518856167793274, | |
| "learning_rate": 2.808219178082192e-5, | |
| "loss": 0.272, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 0.8004194498062134, | |
| "learning_rate": 2.7808219178082197e-5, | |
| "loss": 0.267, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 0.9312605857849121, | |
| "learning_rate": 2.7534246575342465e-5, | |
| "loss": 0.2632, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "grad_norm": 0.8414776921272278, | |
| "learning_rate": 2.726027397260274e-5, | |
| "loss": 0.2681, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "grad_norm": 0.6925989985466003, | |
| "learning_rate": 2.6986301369863014e-5, | |
| "loss": 0.2668, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 0.9184579849243164, | |
| "learning_rate": 2.671232876712329e-5, | |
| "loss": 0.2673, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "grad_norm": 1.1033433675765991, | |
| "learning_rate": 2.6438356164383566e-5, | |
| "loss": 0.2684, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "grad_norm": 0.9113504886627197, | |
| "learning_rate": 2.6164383561643834e-5, | |
| "loss": 0.2644, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 0.7905146479606628, | |
| "learning_rate": 2.589041095890411e-5, | |
| "loss": 0.2668, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "grad_norm": 0.6717493534088135, | |
| "learning_rate": 2.5616438356164386e-5, | |
| "loss": 0.271, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 0.8438414335250854, | |
| "learning_rate": 2.534246575342466e-5, | |
| "loss": 0.2706, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.8165556192398071, | |
| "learning_rate": 2.5068493150684935e-5, | |
| "loss": 0.2603, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "grad_norm": 0.8030436038970947, | |
| "learning_rate": 2.4794520547945206e-5, | |
| "loss": 0.2587, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 0.8518214225769043, | |
| "learning_rate": 2.452054794520548e-5, | |
| "loss": 0.2533, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.9882023930549622, | |
| "learning_rate": 2.4246575342465755e-5, | |
| "loss": 0.2561, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 0.8175749182701111, | |
| "learning_rate": 2.3972602739726026e-5, | |
| "loss": 0.2572, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 0.897048830986023, | |
| "learning_rate": 2.36986301369863e-5, | |
| "loss": 0.2587, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "grad_norm": 0.8218054175376892, | |
| "learning_rate": 2.342465753424658e-5, | |
| "loss": 0.2654, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 0.7128798961639404, | |
| "learning_rate": 2.315068493150685e-5, | |
| "loss": 0.2642, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "grad_norm": 0.7982375621795654, | |
| "learning_rate": 2.2876712328767124e-5, | |
| "loss": 0.2537, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 0.790105938911438, | |
| "learning_rate": 2.2602739726027396e-5, | |
| "loss": 0.2713, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "grad_norm": 0.7734562158584595, | |
| "learning_rate": 2.2328767123287673e-5, | |
| "loss": 0.2616, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "grad_norm": 0.8464659452438354, | |
| "learning_rate": 2.2054794520547948e-5, | |
| "loss": 0.2584, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 0.7386855483055115, | |
| "learning_rate": 2.178082191780822e-5, | |
| "loss": 0.257, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 0.7122279405593872, | |
| "learning_rate": 2.1506849315068494e-5, | |
| "loss": 0.2667, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "grad_norm": 0.8505749106407166, | |
| "learning_rate": 2.1232876712328768e-5, | |
| "loss": 0.2661, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.8915577530860901, | |
| "learning_rate": 2.0958904109589043e-5, | |
| "loss": 0.2567, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "grad_norm": 0.9431042671203613, | |
| "learning_rate": 2.0684931506849317e-5, | |
| "loss": 0.2578, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "grad_norm": 0.7943726181983948, | |
| "learning_rate": 2.0410958904109588e-5, | |
| "loss": 0.2393, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "grad_norm": 0.8244442939758301, | |
| "learning_rate": 2.0136986301369866e-5, | |
| "loss": 0.2175, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.7802647948265076, | |
| "learning_rate": 1.9863013698630137e-5, | |
| "loss": 0.2161, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "grad_norm": 1.1162070035934448, | |
| "learning_rate": 1.9589041095890412e-5, | |
| "loss": 0.2211, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "grad_norm": 1.0273113250732422, | |
| "learning_rate": 1.9315068493150686e-5, | |
| "loss": 0.2253, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "grad_norm": 1.0477781295776367, | |
| "learning_rate": 1.904109589041096e-5, | |
| "loss": 0.2213, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "grad_norm": 0.9134103655815125, | |
| "learning_rate": 1.8767123287671235e-5, | |
| "loss": 0.2269, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "grad_norm": 0.8156262636184692, | |
| "learning_rate": 1.8493150684931506e-5, | |
| "loss": 0.2245, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "grad_norm": 0.9004743695259094, | |
| "learning_rate": 1.821917808219178e-5, | |
| "loss": 0.2254, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "grad_norm": 0.8386040925979614, | |
| "learning_rate": 1.7945205479452055e-5, | |
| "loss": 0.2292, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "grad_norm": 0.9777556657791138, | |
| "learning_rate": 1.767123287671233e-5, | |
| "loss": 0.2213, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "grad_norm": 0.7827901244163513, | |
| "learning_rate": 1.7397260273972604e-5, | |
| "loss": 0.2174, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "grad_norm": 0.7424948811531067, | |
| "learning_rate": 1.7123287671232875e-5, | |
| "loss": 0.2199, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "grad_norm": 0.8807641267776489, | |
| "learning_rate": 1.684931506849315e-5, | |
| "loss": 0.2204, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "grad_norm": 0.8479088544845581, | |
| "learning_rate": 1.6575342465753428e-5, | |
| "loss": 0.2241, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "grad_norm": 0.9211342334747314, | |
| "learning_rate": 1.63013698630137e-5, | |
| "loss": 0.2237, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "grad_norm": 0.8683446645736694, | |
| "learning_rate": 1.6027397260273974e-5, | |
| "loss": 0.2248, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "grad_norm": 0.8828756213188171, | |
| "learning_rate": 1.5753424657534248e-5, | |
| "loss": 0.233, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "grad_norm": 0.9421214461326599, | |
| "learning_rate": 1.5479452054794523e-5, | |
| "loss": 0.2294, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "grad_norm": 0.765132486820221, | |
| "learning_rate": 1.5205479452054797e-5, | |
| "loss": 0.2277, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "grad_norm": 0.9406650066375732, | |
| "learning_rate": 1.4931506849315068e-5, | |
| "loss": 0.217, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "grad_norm": 1.0174639225006104, | |
| "learning_rate": 1.4657534246575344e-5, | |
| "loss": 0.2265, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "grad_norm": 0.826392412185669, | |
| "learning_rate": 1.4383561643835617e-5, | |
| "loss": 0.222, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "grad_norm": 0.9821271300315857, | |
| "learning_rate": 1.4109589041095892e-5, | |
| "loss": 0.2186, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "grad_norm": 0.8172212839126587, | |
| "learning_rate": 1.3835616438356164e-5, | |
| "loss": 0.2238, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "grad_norm": 0.8128436207771301, | |
| "learning_rate": 1.3561643835616439e-5, | |
| "loss": 0.2168, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "grad_norm": 0.8061575293540955, | |
| "learning_rate": 1.3287671232876714e-5, | |
| "loss": 0.2244, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "grad_norm": 0.8976914882659912, | |
| "learning_rate": 1.3013698630136986e-5, | |
| "loss": 0.2212, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.9973928332328796, | |
| "learning_rate": 1.273972602739726e-5, | |
| "loss": 0.2248, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "grad_norm": 0.8042004108428955, | |
| "learning_rate": 1.2465753424657535e-5, | |
| "loss": 0.2178, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "grad_norm": 0.8282990455627441, | |
| "learning_rate": 1.2191780821917808e-5, | |
| "loss": 0.2227, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "grad_norm": 0.6668768525123596, | |
| "learning_rate": 1.1917808219178083e-5, | |
| "loss": 0.2226, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "grad_norm": 0.7972692847251892, | |
| "learning_rate": 1.1643835616438355e-5, | |
| "loss": 0.2193, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "grad_norm": 0.7637550830841064, | |
| "learning_rate": 1.1369863013698632e-5, | |
| "loss": 0.2157, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "grad_norm": 0.8487162590026855, | |
| "learning_rate": 1.1095890410958904e-5, | |
| "loss": 0.2251, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "grad_norm": 0.8710606694221497, | |
| "learning_rate": 1.0821917808219179e-5, | |
| "loss": 0.2153, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "grad_norm": 0.8085966110229492, | |
| "learning_rate": 1.0547945205479452e-5, | |
| "loss": 0.2191, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.94338059425354, | |
| "learning_rate": 1.0273972602739726e-5, | |
| "loss": 0.2184, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "grad_norm": 1.4945096969604492, | |
| "learning_rate": 1e-5, | |
| "loss": 0.1863, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "grad_norm": 0.9178032279014587, | |
| "learning_rate": 9.726027397260275e-6, | |
| "loss": 0.1854, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "grad_norm": 0.8616482615470886, | |
| "learning_rate": 9.452054794520548e-6, | |
| "loss": 0.1843, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "grad_norm": 0.9844592213630676, | |
| "learning_rate": 9.178082191780823e-6, | |
| "loss": 0.1909, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "grad_norm": 0.7312936186790466, | |
| "learning_rate": 8.904109589041095e-6, | |
| "loss": 0.1899, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "grad_norm": 0.9658412933349609, | |
| "learning_rate": 8.630136986301372e-6, | |
| "loss": 0.1878, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "grad_norm": 1.0498002767562866, | |
| "learning_rate": 8.356164383561644e-6, | |
| "loss": 0.1825, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "grad_norm": 0.7098029255867004, | |
| "learning_rate": 8.082191780821919e-6, | |
| "loss": 0.1864, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "grad_norm": 0.9946851134300232, | |
| "learning_rate": 7.808219178082192e-6, | |
| "loss": 0.1852, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.9338549375534058, | |
| "learning_rate": 7.5342465753424655e-6, | |
| "loss": 0.1865, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "grad_norm": 0.8193784952163696, | |
| "learning_rate": 7.260273972602739e-6, | |
| "loss": 0.184, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "grad_norm": 0.9323195815086365, | |
| "learning_rate": 6.9863013698630145e-6, | |
| "loss": 0.1845, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "grad_norm": 0.9668224453926086, | |
| "learning_rate": 6.712328767123288e-6, | |
| "loss": 0.1911, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "grad_norm": 0.9941351413726807, | |
| "learning_rate": 6.438356164383562e-6, | |
| "loss": 0.1859, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "grad_norm": 0.9229924082756042, | |
| "learning_rate": 6.1643835616438354e-6, | |
| "loss": 0.1861, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "grad_norm": 0.8792287111282349, | |
| "learning_rate": 5.89041095890411e-6, | |
| "loss": 0.1903, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "grad_norm": 0.682725191116333, | |
| "learning_rate": 5.616438356164384e-6, | |
| "loss": 0.1822, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "grad_norm": 0.8012785315513611, | |
| "learning_rate": 5.342465753424658e-6, | |
| "loss": 0.1888, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "grad_norm": 0.7928184270858765, | |
| "learning_rate": 5.068493150684932e-6, | |
| "loss": 0.1869, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "grad_norm": 1.2073571681976318, | |
| "learning_rate": 4.7945205479452054e-6, | |
| "loss": 0.184, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "grad_norm": 0.763810396194458, | |
| "learning_rate": 4.52054794520548e-6, | |
| "loss": 0.1824, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "grad_norm": 0.8932220935821533, | |
| "learning_rate": 4.246575342465754e-6, | |
| "loss": 0.1898, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.7250128984451294, | |
| "learning_rate": 3.972602739726028e-6, | |
| "loss": 0.1886, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "grad_norm": 1.0617702007293701, | |
| "learning_rate": 3.6986301369863018e-6, | |
| "loss": 0.1889, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "grad_norm": 0.983672022819519, | |
| "learning_rate": 3.4246575342465754e-6, | |
| "loss": 0.1871, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "grad_norm": 0.9392043352127075, | |
| "learning_rate": 3.1506849315068495e-6, | |
| "loss": 0.1869, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "grad_norm": 0.8135913014411926, | |
| "learning_rate": 2.8767123287671236e-6, | |
| "loss": 0.1861, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "grad_norm": 0.7956686615943909, | |
| "learning_rate": 2.6027397260273973e-6, | |
| "loss": 0.1864, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "grad_norm": 0.8956461548805237, | |
| "learning_rate": 2.3287671232876713e-6, | |
| "loss": 0.1889, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "grad_norm": 0.9515472054481506, | |
| "learning_rate": 2.054794520547945e-6, | |
| "loss": 0.1871, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "grad_norm": 0.8886680006980896, | |
| "learning_rate": 1.7808219178082193e-6, | |
| "loss": 0.187, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "grad_norm": 0.8525242805480957, | |
| "learning_rate": 1.5068493150684932e-6, | |
| "loss": 0.1832, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "grad_norm": 0.9522444009780884, | |
| "learning_rate": 1.232876712328767e-6, | |
| "loss": 0.186, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "grad_norm": 0.8611086010932922, | |
| "learning_rate": 9.589041095890411e-7, | |
| "loss": 0.1855, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "grad_norm": 0.9658819437026978, | |
| "learning_rate": 6.849315068493151e-7, | |
| "loss": 0.177, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "grad_norm": 0.9198510646820068, | |
| "learning_rate": 4.1095890410958903e-7, | |
| "loss": 0.178, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "grad_norm": 0.8326091766357422, | |
| "learning_rate": 1.36986301369863e-7, | |
| "loss": 0.1865, | |
| "step": 18700 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 18750, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 6250, | |
| "total_flos": 8.12664225792e16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |