| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.6451612903225806, | |
| "global_step": 80000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.97983870967742e-05, | |
| "loss": 6.0836, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.959677419354839e-05, | |
| "loss": 5.1751, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.939516129032259e-05, | |
| "loss": 4.7458, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9193548387096775e-05, | |
| "loss": 4.4974, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.899193548387097e-05, | |
| "loss": 4.2937, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8790322580645164e-05, | |
| "loss": 4.097, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.858870967741936e-05, | |
| "loss": 3.9528, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.8387096774193554e-05, | |
| "loss": 3.833, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.818548387096775e-05, | |
| "loss": 3.7057, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.7983870967741937e-05, | |
| "loss": 3.6355, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.778225806451613e-05, | |
| "loss": 3.6956, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.7580645161290326e-05, | |
| "loss": 3.5527, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.737903225806452e-05, | |
| "loss": 3.4586, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.7177419354838716e-05, | |
| "loss": 3.3946, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.697580645161291e-05, | |
| "loss": 3.3624, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.67741935483871e-05, | |
| "loss": 3.3184, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.657258064516129e-05, | |
| "loss": 3.2752, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.637096774193548e-05, | |
| "loss": 3.2428, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.6169354838709676e-05, | |
| "loss": 3.1976, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.596774193548387e-05, | |
| "loss": 3.1766, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.5766129032258065e-05, | |
| "loss": 3.1756, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.556451612903226e-05, | |
| "loss": 3.1177, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5362903225806455e-05, | |
| "loss": 3.1327, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.516129032258064e-05, | |
| "loss": 3.08, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.495967741935484e-05, | |
| "loss": 3.0653, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.475806451612903e-05, | |
| "loss": 3.0551, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.455645161290323e-05, | |
| "loss": 3.0448, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.435483870967742e-05, | |
| "loss": 3.0185, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.415322580645162e-05, | |
| "loss": 3.016, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.395161290322581e-05, | |
| "loss": 3.0085, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.375e-05, | |
| "loss": 2.9812, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.3548387096774194e-05, | |
| "loss": 2.9776, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.334677419354839e-05, | |
| "loss": 2.9659, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.3145161290322584e-05, | |
| "loss": 2.9508, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.294354838709678e-05, | |
| "loss": 2.9336, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.2741935483870973e-05, | |
| "loss": 2.9322, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.254032258064516e-05, | |
| "loss": 2.9694, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.2338709677419356e-05, | |
| "loss": 3.0508, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.213709677419355e-05, | |
| "loss": 2.9583, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.1935483870967746e-05, | |
| "loss": 2.9331, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.173387096774194e-05, | |
| "loss": 2.9427, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.1532258064516135e-05, | |
| "loss": 2.91, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.133064516129033e-05, | |
| "loss": 2.9023, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.112903225806452e-05, | |
| "loss": 2.8931, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.092741935483871e-05, | |
| "loss": 2.8872, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.072580645161291e-05, | |
| "loss": 2.8756, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.0524193548387096e-05, | |
| "loss": 2.7849, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.032258064516129e-05, | |
| "loss": 2.8703, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.0120967741935485e-05, | |
| "loss": 2.8381, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.991935483870968e-05, | |
| "loss": 2.8329, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.971774193548387e-05, | |
| "loss": 2.802, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.951612903225806e-05, | |
| "loss": 2.8102, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.931451612903226e-05, | |
| "loss": 2.7902, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.911290322580645e-05, | |
| "loss": 2.7715, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.891129032258065e-05, | |
| "loss": 2.7801, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.870967741935484e-05, | |
| "loss": 2.8037, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.8508064516129036e-05, | |
| "loss": 2.7247, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.8306451612903224e-05, | |
| "loss": 2.6845, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.810483870967742e-05, | |
| "loss": 2.7579, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.7903225806451614e-05, | |
| "loss": 2.7589, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.770161290322581e-05, | |
| "loss": 2.7403, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 2.8492, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.72983870967742e-05, | |
| "loss": 2.8556, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.7096774193548386e-05, | |
| "loss": 2.8233, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.689516129032258e-05, | |
| "loss": 2.8242, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.6693548387096776e-05, | |
| "loss": 2.7993, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.649193548387097e-05, | |
| "loss": 2.7882, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.6290322580645165e-05, | |
| "loss": 2.8004, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.608870967741936e-05, | |
| "loss": 2.7491, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.5887096774193555e-05, | |
| "loss": 2.7915, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.568548387096774e-05, | |
| "loss": 2.7612, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.548387096774194e-05, | |
| "loss": 2.7913, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.528225806451613e-05, | |
| "loss": 2.7564, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.508064516129033e-05, | |
| "loss": 2.7211, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.487903225806452e-05, | |
| "loss": 2.7578, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.467741935483872e-05, | |
| "loss": 2.7616, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4475806451612905e-05, | |
| "loss": 2.7308, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.427419354838709e-05, | |
| "loss": 2.7324, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.407258064516129e-05, | |
| "loss": 2.6723, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.387096774193548e-05, | |
| "loss": 2.7095, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.366935483870968e-05, | |
| "loss": 2.6884, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.346774193548387e-05, | |
| "loss": 2.6571, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.3266129032258067e-05, | |
| "loss": 2.6433, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.306451612903226e-05, | |
| "loss": 2.6354, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.286290322580645e-05, | |
| "loss": 2.6071, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.2661290322580644e-05, | |
| "loss": 2.6559, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.245967741935484e-05, | |
| "loss": 2.6328, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.2258064516129034e-05, | |
| "loss": 2.6289, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.205645161290323e-05, | |
| "loss": 2.5489, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.185483870967742e-05, | |
| "loss": 2.6088, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.165322580645161e-05, | |
| "loss": 2.5721, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1451612903225806e-05, | |
| "loss": 2.6424, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.125e-05, | |
| "loss": 2.5594, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.1048387096774195e-05, | |
| "loss": 2.5989, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.084677419354839e-05, | |
| "loss": 2.6342, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0645161290322585e-05, | |
| "loss": 2.5973, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0443548387096776e-05, | |
| "loss": 2.5929, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.024193548387097e-05, | |
| "loss": 2.6013, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.0040322580645162e-05, | |
| "loss": 2.6019, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.9838709677419357e-05, | |
| "loss": 2.577, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9637096774193552e-05, | |
| "loss": 2.592, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9435483870967743e-05, | |
| "loss": 2.5797, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9233870967741938e-05, | |
| "loss": 2.6054, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9032258064516133e-05, | |
| "loss": 2.5631, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.8830645161290328e-05, | |
| "loss": 2.5619, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.862903225806452e-05, | |
| "loss": 2.2817, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8427419354838714e-05, | |
| "loss": 2.1577, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.822580645161291e-05, | |
| "loss": 2.3711, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.8024193548387097e-05, | |
| "loss": 2.1908, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.7822580645161288e-05, | |
| "loss": 2.2163, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7620967741935483e-05, | |
| "loss": 2.2611, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7419354838709678e-05, | |
| "loss": 2.2939, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.721774193548387e-05, | |
| "loss": 2.0934, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.7016129032258064e-05, | |
| "loss": 2.1656, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.681451612903226e-05, | |
| "loss": 2.2859, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.661290322580645e-05, | |
| "loss": 2.152, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.6411290322580645e-05, | |
| "loss": 2.244, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.620967741935484e-05, | |
| "loss": 2.1485, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.6008064516129034e-05, | |
| "loss": 2.1701, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.5806451612903226e-05, | |
| "loss": 2.2302, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.560483870967742e-05, | |
| "loss": 2.1719, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.5403225806451615e-05, | |
| "loss": 2.189, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.5201612903225806e-05, | |
| "loss": 2.2415, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.5e-05, | |
| "loss": 2.2305, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.4798387096774196e-05, | |
| "loss": 2.2169, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4596774193548387e-05, | |
| "loss": 2.2279, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4395161290322582e-05, | |
| "loss": 2.2056, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.4193548387096777e-05, | |
| "loss": 2.1959, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.3991935483870968e-05, | |
| "loss": 2.2024, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.3790322580645163e-05, | |
| "loss": 2.1846, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.3588709677419358e-05, | |
| "loss": 2.1509, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.338709677419355e-05, | |
| "loss": 2.1856, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.318548387096774e-05, | |
| "loss": 2.2516, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.2983870967741935e-05, | |
| "loss": 2.2475, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.278225806451613e-05, | |
| "loss": 2.2131, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.258064516129032e-05, | |
| "loss": 2.1794, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2379032258064516e-05, | |
| "loss": 2.1927, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.217741935483871e-05, | |
| "loss": 2.1165, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.1975806451612906e-05, | |
| "loss": 2.1751, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.1774193548387097e-05, | |
| "loss": 2.2166, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1572580645161292e-05, | |
| "loss": 2.1228, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1370967741935487e-05, | |
| "loss": 2.2166, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.1169354838709678e-05, | |
| "loss": 2.1631, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0967741935483873e-05, | |
| "loss": 2.1822, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0766129032258068e-05, | |
| "loss": 2.1899, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.056451612903226e-05, | |
| "loss": 2.1951, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0362903225806454e-05, | |
| "loss": 2.1796, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0161290322580645e-05, | |
| "loss": 2.1674, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.995967741935484e-05, | |
| "loss": 2.1218, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.975806451612903e-05, | |
| "loss": 2.14, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9556451612903226e-05, | |
| "loss": 2.1605, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.935483870967742e-05, | |
| "loss": 2.1331, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9153225806451612e-05, | |
| "loss": 2.1514, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8951612903225807e-05, | |
| "loss": 2.1411, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 2.1651, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8548387096774193e-05, | |
| "loss": 2.0862, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8346774193548388e-05, | |
| "loss": 2.1951, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.8145161290322583e-05, | |
| "loss": 2.0717, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7943548387096777e-05, | |
| "loss": 2.1325, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.774193548387097e-05, | |
| "loss": 2.1493, | |
| "step": 80000 | |
| } | |
| ], | |
| "max_steps": 124000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.18874494205952e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |