| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 27.692307692307693, | |
| "global_step": 90, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 3.768, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 12.368, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2e-05, | |
| "loss": 11.537, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.999348095389677e-05, | |
| "loss": 7.8117, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.9973932315179502e-05, | |
| "loss": 5.6659, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9941379571543597e-05, | |
| "loss": 4.2924, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.9895865165556375e-05, | |
| "loss": 2.1566, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1.9837448439320027e-05, | |
| "loss": 2.607, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.976620555710087e-05, | |
| "loss": 2.6508, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 1.9682229406025635e-05, | |
| "loss": 3.4377, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.9585629474974413e-05, | |
| "loss": 2.3918, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 1.9476531711828027e-05, | |
| "loss": 0.9541, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.935507835925601e-05, | |
| "loss": 0.2939, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 1.9221427769259333e-05, | |
| "loss": 0.717, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 1.9075754196709574e-05, | |
| "loss": 0.3357, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 1.8918247572153822e-05, | |
| "loss": 0.2568, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 1.8749113254181498e-05, | |
| "loss": 0.1306, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 1.8568571761675893e-05, | |
| "loss": 0.3485, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 1.837685848629965e-05, | |
| "loss": 0.1659, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 1.817422338558892e-05, | |
| "loss": 0.6525, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 1.796093065705644e-05, | |
| "loss": 0.1003, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 1.7737258393728363e-05, | |
| "loss": 0.3485, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 1.7503498221564026e-05, | |
| "loss": 0.2097, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "learning_rate": 1.725995491923131e-05, | |
| "loss": 0.1424, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 1.7006946020733426e-05, | |
| "loss": 0.1512, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 1.6744801401405138e-05, | |
| "loss": 0.0815, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 1.647386284781828e-05, | |
| "loss": 0.0202, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 1.6194483612157232e-05, | |
| "loss": 0.0909, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 1.590702795164551e-05, | |
| "loss": 0.1048, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 1.5611870653623826e-05, | |
| "loss": 0.1186, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 1.530939654689887e-05, | |
| "loss": 0.0219, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.0718, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 1.4684084406997903e-05, | |
| "loss": 0.0788, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 10.46, | |
| "learning_rate": 1.4362061661555675e-05, | |
| "loss": 0.0534, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 10.77, | |
| "learning_rate": 1.4034351619898088e-05, | |
| "loss": 0.0225, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "learning_rate": 1.3701381553399147e-05, | |
| "loss": 0.0345, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "learning_rate": 1.3363585591501751e-05, | |
| "loss": 0.0125, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 11.69, | |
| "learning_rate": 1.3021404155695728e-05, | |
| "loss": 0.0211, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 1.2675283385292212e-05, | |
| "loss": 0.0998, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 1.2325674555743106e-05, | |
| "loss": 0.0139, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 12.62, | |
| "learning_rate": 1.1973033490264e-05, | |
| "loss": 0.0052, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 12.92, | |
| "learning_rate": 1.161781996552765e-05, | |
| "loss": 0.0069, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 13.23, | |
| "learning_rate": 1.1260497112202895e-05, | |
| "loss": 0.1117, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 13.54, | |
| "learning_rate": 1.0901530811120655e-05, | |
| "loss": 0.0081, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 1.0541389085854177e-05, | |
| "loss": 0.009, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 1.0180541492505605e-05, | |
| "loss": 0.0019, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 14.46, | |
| "learning_rate": 9.819458507494395e-06, | |
| "loss": 0.0309, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "learning_rate": 9.458610914145826e-06, | |
| "loss": 0.0016, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 15.08, | |
| "learning_rate": 9.098469188879348e-06, | |
| "loss": 0.0158, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 15.38, | |
| "learning_rate": 8.739502887797108e-06, | |
| "loss": 0.0013, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 15.69, | |
| "learning_rate": 8.382180034472353e-06, | |
| "loss": 0.0056, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 8.026966509736001e-06, | |
| "loss": 0.0311, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 16.31, | |
| "learning_rate": 7.674325444256899e-06, | |
| "loss": 0.0198, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 16.62, | |
| "learning_rate": 7.324716614707794e-06, | |
| "loss": 0.0032, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 16.92, | |
| "learning_rate": 6.978595844304272e-06, | |
| "loss": 0.0011, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 6.636414408498249e-06, | |
| "loss": 0.0038, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 17.54, | |
| "learning_rate": 6.298618446600856e-06, | |
| "loss": 0.011, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 5.965648380101916e-06, | |
| "loss": 0.023, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 18.15, | |
| "learning_rate": 5.637938338444325e-06, | |
| "loss": 0.0005, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 18.46, | |
| "learning_rate": 5.3159155930021e-06, | |
| "loss": 0.0007, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 18.77, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.0063, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 19.08, | |
| "learning_rate": 4.690603453101134e-06, | |
| "loss": 0.0009, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 19.38, | |
| "learning_rate": 4.388129346376177e-06, | |
| "loss": 0.0021, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 19.69, | |
| "learning_rate": 4.092972048354491e-06, | |
| "loss": 0.02, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 3.8055163878427703e-06, | |
| "loss": 0.001, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 20.31, | |
| "learning_rate": 3.5261371521817247e-06, | |
| "loss": 0.0008, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 20.62, | |
| "learning_rate": 3.255198598594862e-06, | |
| "loss": 0.0098, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 20.92, | |
| "learning_rate": 2.9930539792665767e-06, | |
| "loss": 0.0014, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 21.23, | |
| "learning_rate": 2.740045080768694e-06, | |
| "loss": 0.0015, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 21.54, | |
| "learning_rate": 2.496501778435977e-06, | |
| "loss": 0.0006, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 21.85, | |
| "learning_rate": 2.2627416062716366e-06, | |
| "loss": 0.0064, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 22.15, | |
| "learning_rate": 2.0390693429435626e-06, | |
| "loss": 0.0007, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 22.46, | |
| "learning_rate": 1.8257766144110823e-06, | |
| "loss": 0.0015, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 22.77, | |
| "learning_rate": 1.6231415137003536e-06, | |
| "loss": 0.0089, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 23.08, | |
| "learning_rate": 1.4314282383241097e-06, | |
| "loss": 0.0022, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 23.38, | |
| "learning_rate": 1.2508867458185037e-06, | |
| "loss": 0.0076, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 23.69, | |
| "learning_rate": 1.0817524278461777e-06, | |
| "loss": 0.0004, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 9.242458032904311e-07, | |
| "loss": 0.0005, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 24.31, | |
| "learning_rate": 7.785722307406685e-07, | |
| "loss": 0.0004, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 24.62, | |
| "learning_rate": 6.449216407439906e-07, | |
| "loss": 0.0049, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 24.92, | |
| "learning_rate": 5.234682881719766e-07, | |
| "loss": 0.004, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 25.23, | |
| "learning_rate": 4.1437052502558693e-07, | |
| "loss": 0.0037, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 25.54, | |
| "learning_rate": 3.1777059397436693e-07, | |
| "loss": 0.0099, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 25.85, | |
| "learning_rate": 2.3379444289913344e-07, | |
| "loss": 0.001, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 26.15, | |
| "learning_rate": 1.6255156067997325e-07, | |
| "loss": 0.0004, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 26.46, | |
| "learning_rate": 1.041348344436277e-07, | |
| "loss": 0.0021, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 26.77, | |
| "learning_rate": 5.862042845640403e-08, | |
| "loss": 0.0052, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 27.08, | |
| "learning_rate": 2.606768482050215e-08, | |
| "loss": 0.0047, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 27.38, | |
| "learning_rate": 6.5190461032305085e-09, | |
| "loss": 0.0004, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 27.69, | |
| "learning_rate": 0.0, | |
| "loss": 0.0003, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 27.69, | |
| "step": 90, | |
| "total_flos": 9963642317438976.0, | |
| "train_loss": 0.7193527906240585, | |
| "train_runtime": 2724.9841, | |
| "train_samples_per_second": 1.134, | |
| "train_steps_per_second": 0.033 | |
| } | |
| ], | |
| "max_steps": 90, | |
| "num_train_epochs": 30, | |
| "total_flos": 9963642317438976.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |