| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 170000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0009970588235294119, | |
| "loss": 5.4687, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0009941176470588235, | |
| "loss": 2.9773, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0009911764705882353, | |
| "loss": 2.803, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0009882352941176472, | |
| "loss": 2.6892, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0009852941176470588, | |
| "loss": 2.6263, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0009823529411764707, | |
| "loss": 2.5926, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0009794117647058823, | |
| "loss": 2.5264, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0009764705882352941, | |
| "loss": 2.5018, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009735294117647059, | |
| "loss": 2.4978, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009705882352941176, | |
| "loss": 2.4435, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009676470588235295, | |
| "loss": 2.3768, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009647058823529412, | |
| "loss": 2.382, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009617647058823529, | |
| "loss": 2.4059, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009588235294117648, | |
| "loss": 2.3918, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009558823529411765, | |
| "loss": 2.3325, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009529411764705882, | |
| "loss": 2.3217, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00095, | |
| "loss": 2.2997, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009470588235294117, | |
| "loss": 2.2969, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009441176470588235, | |
| "loss": 2.3028, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009411764705882353, | |
| "loss": 2.2885, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009382352941176471, | |
| "loss": 2.3032, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009352941176470589, | |
| "loss": 2.2727, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009323529411764706, | |
| "loss": 2.2315, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009294117647058824, | |
| "loss": 2.2389, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009264705882352942, | |
| "loss": 2.2585, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000923529411764706, | |
| "loss": 2.2445, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009205882352941176, | |
| "loss": 2.1763, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009176470588235294, | |
| "loss": 2.2386, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009147058823529412, | |
| "loss": 2.2169, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009117647058823529, | |
| "loss": 2.2082, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009088235294117648, | |
| "loss": 2.2337, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009058823529411765, | |
| "loss": 2.1778, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009029411764705882, | |
| "loss": 2.2025, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009000000000000001, | |
| "loss": 2.1734, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0008970588235294118, | |
| "loss": 2.1567, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0008941176470588236, | |
| "loss": 2.1574, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0008911764705882354, | |
| "loss": 2.1401, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000888235294117647, | |
| "loss": 2.1853, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0008852941176470588, | |
| "loss": 2.1417, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0008823529411764706, | |
| "loss": 2.1332, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0008794117647058824, | |
| "loss": 2.12, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0008764705882352941, | |
| "loss": 2.1348, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0008735294117647059, | |
| "loss": 2.1464, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0008705882352941177, | |
| "loss": 2.1532, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0008676470588235294, | |
| "loss": 2.1142, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0008647058823529413, | |
| "loss": 2.0977, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000861764705882353, | |
| "loss": 2.1114, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0008588235294117646, | |
| "loss": 2.079, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0008558823529411765, | |
| "loss": 2.1021, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0008529411764705882, | |
| "loss": 2.1075, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00085, | |
| "loss": 2.0557, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0008470588235294118, | |
| "loss": 2.0445, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0008441176470588235, | |
| "loss": 2.072, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0008411764705882353, | |
| "loss": 2.0684, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0008382352941176471, | |
| "loss": 2.06, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0008352941176470589, | |
| "loss": 2.0768, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008323529411764706, | |
| "loss": 2.1055, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008294117647058824, | |
| "loss": 2.0504, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008264705882352941, | |
| "loss": 2.0279, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008235294117647058, | |
| "loss": 2.0852, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008205882352941177, | |
| "loss": 2.0934, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008176470588235294, | |
| "loss": 2.0453, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0008147058823529411, | |
| "loss": 2.0469, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000811764705882353, | |
| "loss": 2.6994, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0008088235294117647, | |
| "loss": 2.0459, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0008058823529411766, | |
| "loss": 2.0367, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0008029411764705883, | |
| "loss": 2.0672, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0008, | |
| "loss": 2.0765, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007970588235294119, | |
| "loss": 2.0736, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007941176470588235, | |
| "loss": 2.0491, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007911764705882353, | |
| "loss": 2.0626, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007882352941176471, | |
| "loss": 2.034, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007852941176470588, | |
| "loss": 2.0163, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0007823529411764706, | |
| "loss": 2.0318, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0007794117647058824, | |
| "loss": 2.0477, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0007764705882352942, | |
| "loss": 2.0535, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0007735294117647059, | |
| "loss": 1.9894, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0007705882352941177, | |
| "loss": 2.0341, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0007676470588235295, | |
| "loss": 1.9852, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0007647058823529411, | |
| "loss": 1.9756, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.000761764705882353, | |
| "loss": 2.0002, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0007588235294117647, | |
| "loss": 2.0056, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0007558823529411764, | |
| "loss": 2.0109, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0007529411764705883, | |
| "loss": 2.0476, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00075, | |
| "loss": 2.0153, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0007470588235294118, | |
| "loss": 1.983, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0007441176470588236, | |
| "loss": 1.9831, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0007411764705882353, | |
| "loss": 1.9843, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0007382352941176471, | |
| "loss": 1.9926, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0007352941176470589, | |
| "loss": 1.9853, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0007323529411764706, | |
| "loss": 1.9806, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0007294117647058823, | |
| "loss": 1.9876, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0007264705882352941, | |
| "loss": 1.9705, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0007235294117647059, | |
| "loss": 1.9724, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0007205882352941176, | |
| "loss": 2.0018, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0007176470588235295, | |
| "loss": 1.9356, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0007147058823529412, | |
| "loss": 1.966, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0007117647058823529, | |
| "loss": 1.9758, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0007088235294117648, | |
| "loss": 1.9851, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0007058823529411765, | |
| "loss": 1.9648, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0007029411764705881, | |
| "loss": 1.9601, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0007, | |
| "loss": 1.9646, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006970588235294117, | |
| "loss": 1.9551, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006941176470588235, | |
| "loss": 1.9455, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006911764705882353, | |
| "loss": 1.9527, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000688235294117647, | |
| "loss": 1.9583, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006852941176470589, | |
| "loss": 1.9525, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006823529411764706, | |
| "loss": 1.9276, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006794117647058824, | |
| "loss": 1.9626, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006764705882352942, | |
| "loss": 1.8995, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000673529411764706, | |
| "loss": 1.9457, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0006705882352941176, | |
| "loss": 1.9706, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0006676470588235294, | |
| "loss": 1.9671, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0006647058823529412, | |
| "loss": 1.9613, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0006617647058823529, | |
| "loss": 1.9511, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0006588235294117648, | |
| "loss": 1.9437, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0006558823529411765, | |
| "loss": 1.9381, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0006529411764705882, | |
| "loss": 1.9205, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0006500000000000001, | |
| "loss": 1.9415, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0006470588235294118, | |
| "loss": 1.9307, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0006441176470588236, | |
| "loss": 1.9616, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0006411764705882354, | |
| "loss": 1.9267, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000638235294117647, | |
| "loss": 1.911, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0006352941176470588, | |
| "loss": 1.9316, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0006323529411764706, | |
| "loss": 1.9174, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0006294117647058824, | |
| "loss": 1.9185, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0006264705882352941, | |
| "loss": 1.9067, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0006235294117647059, | |
| "loss": 1.9071, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0006205882352941177, | |
| "loss": 1.9217, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0006176470588235294, | |
| "loss": 1.9262, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0006147058823529413, | |
| "loss": 1.9182, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.000611764705882353, | |
| "loss": 1.9291, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0006088235294117646, | |
| "loss": 1.9115, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0006058823529411765, | |
| "loss": 1.8964, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006029411764705882, | |
| "loss": 1.9267, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006, | |
| "loss": 1.8993, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005970588235294118, | |
| "loss": 1.9361, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005941176470588235, | |
| "loss": 1.8742, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005911764705882353, | |
| "loss": 1.9027, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005882352941176471, | |
| "loss": 1.9098, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005852941176470589, | |
| "loss": 1.8969, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005823529411764706, | |
| "loss": 1.9119, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005794117647058824, | |
| "loss": 1.9021, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005764705882352941, | |
| "loss": 1.8577, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005735294117647058, | |
| "loss": 1.892, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005705882352941177, | |
| "loss": 1.8874, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005676470588235294, | |
| "loss": 1.8978, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005647058823529411, | |
| "loss": 1.8824, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.000561764705882353, | |
| "loss": 1.9075, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005588235294117647, | |
| "loss": 1.891, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005558823529411766, | |
| "loss": 1.8859, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0005529411764705883, | |
| "loss": 1.8606, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00055, | |
| "loss": 1.9076, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0005470588235294119, | |
| "loss": 1.8524, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005441176470588235, | |
| "loss": 1.8389, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005411764705882352, | |
| "loss": 1.8674, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005382352941176471, | |
| "loss": 1.9254, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005352941176470588, | |
| "loss": 1.9068, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005323529411764706, | |
| "loss": 1.8746, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005294117647058824, | |
| "loss": 1.8625, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005264705882352942, | |
| "loss": 1.7992, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005235294117647059, | |
| "loss": 1.8703, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005205882352941177, | |
| "loss": 1.8678, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005176470588235295, | |
| "loss": 1.8611, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005147058823529411, | |
| "loss": 1.8053, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.000511764705882353, | |
| "loss": 1.8697, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005088235294117647, | |
| "loss": 1.8965, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005058823529411764, | |
| "loss": 1.8519, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005029411764705883, | |
| "loss": 1.8454, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005, | |
| "loss": 1.8719, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004970588235294117, | |
| "loss": 1.876, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004941176470588236, | |
| "loss": 1.8438, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004911764705882353, | |
| "loss": 1.8712, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00048823529411764707, | |
| "loss": 1.8232, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004852941176470588, | |
| "loss": 1.8492, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004823529411764706, | |
| "loss": 1.8605, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004794117647058824, | |
| "loss": 1.8502, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004764705882352941, | |
| "loss": 1.8803, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00047352941176470587, | |
| "loss": 1.8466, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00047058823529411766, | |
| "loss": 1.8823, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00046764705882352945, | |
| "loss": 1.8624, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004647058823529412, | |
| "loss": 1.8055, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.000461764705882353, | |
| "loss": 1.8456, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004588235294117647, | |
| "loss": 1.8181, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00045588235294117646, | |
| "loss": 1.8324, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00045294117647058825, | |
| "loss": 1.8439, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00045000000000000004, | |
| "loss": 1.8487, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004470588235294118, | |
| "loss": 1.8355, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0004441176470588235, | |
| "loss": 1.8536, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0004411764705882353, | |
| "loss": 1.8101, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00043823529411764705, | |
| "loss": 1.7965, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00043529411764705884, | |
| "loss": 1.8239, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00043235294117647063, | |
| "loss": 1.8155, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0004294117647058823, | |
| "loss": 1.831, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0004264705882352941, | |
| "loss": 1.8305, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0004235294117647059, | |
| "loss": 1.824, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00042058823529411764, | |
| "loss": 1.8162, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00041764705882352943, | |
| "loss": 1.8092, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0004147058823529412, | |
| "loss": 1.7905, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0004117647058823529, | |
| "loss": 1.8432, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "eval_gen_len": 85.6, | |
| "eval_loss": 1.6427801847457886, | |
| "eval_rouge1": 29.1215, | |
| "eval_rouge2": 10.0431, | |
| "eval_rougeL": 20.203, | |
| "eval_rougeLsum": 24.5992, | |
| "eval_runtime": 107.8904, | |
| "eval_samples_per_second": 0.927, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0004088235294117647, | |
| "loss": 1.8176, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0004058823529411765, | |
| "loss": 1.8058, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0004029411764705883, | |
| "loss": 1.7967, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0004, | |
| "loss": 1.7941, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00039705882352941176, | |
| "loss": 1.8045, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00039411764705882355, | |
| "loss": 1.7988, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003911764705882353, | |
| "loss": 1.8047, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003882352941176471, | |
| "loss": 1.817, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00038529411764705887, | |
| "loss": 1.8081, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00038235294117647055, | |
| "loss": 1.7954, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00037941176470588235, | |
| "loss": 1.8404, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00037647058823529414, | |
| "loss": 1.815, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0003735294117647059, | |
| "loss": 1.8023, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00037058823529411767, | |
| "loss": 1.7957, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00036764705882352946, | |
| "loss": 1.7989, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00036470588235294114, | |
| "loss": 1.8369, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00036176470588235294, | |
| "loss": 1.816, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00035882352941176473, | |
| "loss": 1.7874, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00035588235294117647, | |
| "loss": 1.7832, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00035294117647058826, | |
| "loss": 1.8078, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00035, | |
| "loss": 1.7959, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00034705882352941173, | |
| "loss": 1.7853, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0003441176470588235, | |
| "loss": 1.8219, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0003411764705882353, | |
| "loss": 1.8028, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0003382352941176471, | |
| "loss": 1.8283, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0003352941176470588, | |
| "loss": 1.7884, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0003323529411764706, | |
| "loss": 1.7835, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0003294117647058824, | |
| "loss": 1.7767, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0003264705882352941, | |
| "loss": 1.8169, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0003235294117647059, | |
| "loss": 1.8163, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0003205882352941177, | |
| "loss": 1.7781, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0003176470588235294, | |
| "loss": 1.7734, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0003147058823529412, | |
| "loss": 1.7818, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00031176470588235297, | |
| "loss": 1.7562, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0003088235294117647, | |
| "loss": 1.7839, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0003058823529411765, | |
| "loss": 1.7727, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00030294117647058824, | |
| "loss": 1.7838, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0003, | |
| "loss": 1.7936, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00029705882352941177, | |
| "loss": 1.7998, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00029411764705882356, | |
| "loss": 1.773, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002911764705882353, | |
| "loss": 1.7997, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00028823529411764703, | |
| "loss": 1.7989, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002852941176470588, | |
| "loss": 1.773, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00028235294117647056, | |
| "loss": 1.7351, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00027941176470588236, | |
| "loss": 1.7558, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00027647058823529415, | |
| "loss": 1.7618, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00027352941176470594, | |
| "loss": 1.7601, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0002705882352941176, | |
| "loss": 1.8319, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0002676470588235294, | |
| "loss": 1.7765, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0002647058823529412, | |
| "loss": 1.7838, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00026176470588235295, | |
| "loss": 1.7566, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00025882352941176474, | |
| "loss": 1.7626, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0002558823529411765, | |
| "loss": 1.7678, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0002529411764705882, | |
| "loss": 1.794, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00025, | |
| "loss": 1.747, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0002470588235294118, | |
| "loss": 1.7565, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00024411764705882354, | |
| "loss": 1.7932, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0002411764705882353, | |
| "loss": 1.7623, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00023823529411764704, | |
| "loss": 1.7712, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00023529411764705883, | |
| "loss": 1.7684, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0002323529411764706, | |
| "loss": 1.7317, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00022941176470588236, | |
| "loss": 1.7679, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00022647058823529412, | |
| "loss": 1.7735, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0002235294117647059, | |
| "loss": 1.7688, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00022058823529411765, | |
| "loss": 1.7494, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00021764705882352942, | |
| "loss": 1.7739, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00021470588235294116, | |
| "loss": 1.7537, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00021176470588235295, | |
| "loss": 1.7243, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00020882352941176471, | |
| "loss": 1.7828, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00020588235294117645, | |
| "loss": 1.783, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00020294117647058824, | |
| "loss": 1.714, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7523, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00019705882352941177, | |
| "loss": 1.7409, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00019411764705882354, | |
| "loss": 1.7641, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00019117647058823528, | |
| "loss": 1.7501, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00018823529411764707, | |
| "loss": 1.7367, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00018529411764705883, | |
| "loss": 1.7362, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00018235294117647057, | |
| "loss": 1.7564, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00017941176470588236, | |
| "loss": 1.7527, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00017647058823529413, | |
| "loss": 1.7404, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00017352941176470587, | |
| "loss": 1.7694, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00017058823529411766, | |
| "loss": 1.7443, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0001676470588235294, | |
| "loss": 1.7413, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0001647058823529412, | |
| "loss": 1.7572, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00016176470588235295, | |
| "loss": 1.7442, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0001588235294117647, | |
| "loss": 1.7608, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00015588235294117648, | |
| "loss": 1.7467, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00015294117647058825, | |
| "loss": 1.7496, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00015, | |
| "loss": 1.7396, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00014705882352941178, | |
| "loss": 1.7341, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00014411764705882352, | |
| "loss": 1.7358, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00014117647058823528, | |
| "loss": 1.7721, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00013823529411764707, | |
| "loss": 1.7224, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0001352941176470588, | |
| "loss": 1.7137, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0001323529411764706, | |
| "loss": 1.7555, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00012941176470588237, | |
| "loss": 1.7432, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0001264705882352941, | |
| "loss": 1.7319, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0001235294117647059, | |
| "loss": 1.725, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00012058823529411765, | |
| "loss": 1.7655, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00011764705882352942, | |
| "loss": 1.7358, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00011470588235294118, | |
| "loss": 1.7214, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00011176470588235294, | |
| "loss": 1.7413, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00010882352941176471, | |
| "loss": 1.7069, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00010588235294117647, | |
| "loss": 1.7282, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.00010294117647058823, | |
| "loss": 1.7121, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7227, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.705882352941177e-05, | |
| "loss": 1.7391, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.411764705882353e-05, | |
| "loss": 1.7422, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.117647058823529e-05, | |
| "loss": 1.7154, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.823529411764706e-05, | |
| "loss": 1.7419, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.529411764705883e-05, | |
| "loss": 1.7608, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.23529411764706e-05, | |
| "loss": 1.7154, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.941176470588235e-05, | |
| "loss": 1.7082, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.647058823529412e-05, | |
| "loss": 1.7347, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.352941176470589e-05, | |
| "loss": 1.7054, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.058823529411764e-05, | |
| "loss": 1.7326, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.76470588235294e-05, | |
| "loss": 1.7224, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.470588235294118e-05, | |
| "loss": 1.7362, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.176470588235295e-05, | |
| "loss": 1.7159, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.882352941176471e-05, | |
| "loss": 1.73, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.588235294117647e-05, | |
| "loss": 1.7523, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.294117647058824e-05, | |
| "loss": 1.7315, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5e-05, | |
| "loss": 1.709, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.705882352941177e-05, | |
| "loss": 1.7341, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.411764705882353e-05, | |
| "loss": 1.7186, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.11764705882353e-05, | |
| "loss": 1.719, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.823529411764706e-05, | |
| "loss": 1.7115, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.529411764705882e-05, | |
| "loss": 1.7036, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.235294117647059e-05, | |
| "loss": 1.7147, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.9411764705882354e-05, | |
| "loss": 1.7347, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.647058823529412e-05, | |
| "loss": 1.7334, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.3529411764705884e-05, | |
| "loss": 1.7337, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.058823529411765e-05, | |
| "loss": 1.7266, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.764705882352941e-05, | |
| "loss": 1.6958, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.4705882352941177e-05, | |
| "loss": 1.7052, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1764705882352942e-05, | |
| "loss": 1.6985, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.823529411764705e-06, | |
| "loss": 1.6896, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.882352941176471e-06, | |
| "loss": 1.7455, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.9411764705882355e-06, | |
| "loss": 1.7554, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.6962, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 170000, | |
| "total_flos": 3421528306437104640, | |
| "train_runtime": 228303.6105, | |
| "train_samples_per_second": 0.745 | |
| } | |
| ], | |
| "max_steps": 170000, | |
| "num_train_epochs": 1, | |
| "total_flos": 3421528306437104640, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |