Denis641/CodeGen-Supervised
014
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 2.4922118380062304,5 "eval_steps": 500,6 "global_step": 800,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.1557632398753894,13 "grad_norm": 5.09375,14 "learning_rate": 3.3333333333333335e-05,15 "loss": 2.6073,16 "step": 5017 },18 {19 "epoch": 0.3115264797507788,20 "grad_norm": 3.53125,21 "learning_rate": 6.666666666666667e-05,22 "loss": 0.8752,23 "step": 10024 },25 {26 "epoch": 0.4672897196261682,27 "grad_norm": 1.546875,28 "learning_rate": 0.0001,29 "loss": 0.5161,30 "step": 15031 },32 {33 "epoch": 0.6230529595015576,34 "grad_norm": 2.3125,35 "learning_rate": 0.00013333333333333334,36 "loss": 0.3785,37 "step": 20038 },39 {40 "epoch": 0.778816199376947,41 "grad_norm": 2.125,42 "learning_rate": 0.0001666666666666667,43 "loss": 0.293,44 "step": 25045 },46 {47 "epoch": 0.9345794392523364,48 "grad_norm": 2.0,49 "learning_rate": 0.0002,50 "loss": 0.252,51 "step": 30052 },53 {54 "epoch": 1.0903426791277258,55 "grad_norm": 1.9921875,56 "learning_rate": 0.00018491704374057317,57 "loss": 0.2341,58 "step": 35059 },60 {61 "epoch": 1.2461059190031152,62 "grad_norm": 2.015625,63 "learning_rate": 0.0001698340874811463,64 "loss": 0.183,65 "step": 40066 },67 {68 "epoch": 1.4018691588785046,69 "grad_norm": 1.8046875,70 "learning_rate": 0.00015475113122171948,71 "loss": 0.1659,72 "step": 45073 },74 {75 "epoch": 1.557632398753894,76 "grad_norm": 1.5390625,77 "learning_rate": 0.0001396681749622926,78 "loss": 0.1368,79 "step": 50080 },81 {82 "epoch": 1.7133956386292835,83 "grad_norm": 0.408203125,84 "learning_rate": 0.00012458521870286577,85 "loss": 0.104,86 "step": 55087 },88 {89 "epoch": 1.8691588785046729,90 "grad_norm": 2.671875,91 "learning_rate": 0.00010950226244343893,92 "loss": 0.088,93 "step": 60094 },95 {96 "epoch": 2.0249221183800623,97 "grad_norm": 0.91015625,98 "learning_rate": 9.441930618401207e-05,99 "loss": 0.0903,100 "step": 650101 },102 {103 "epoch": 2.1806853582554515,104 "grad_norm": 1.8046875,105 "learning_rate": 7.933634992458521e-05,106 "loss": 0.0713,107 "step": 700108 },109 {110 "epoch": 2.336448598130841,111 "grad_norm": 0.2392578125,112 "learning_rate": 6.425339366515838e-05,113 "loss": 0.061,114 "step": 750115 },116 {117 "epoch": 2.4922118380062304,118 "grad_norm": 1.0,119 "learning_rate": 4.917043740573152e-05,120 "loss": 0.0581,121 "step": 800122 }123 ],124 "logging_steps": 50,125 "max_steps": 963,126 "num_input_tokens_seen": 0,127 "num_train_epochs": 3,128 "save_steps": 200,129 "total_flos": 0.0,130 "train_batch_size": 64,131 "trial_name": null,132 "trial_params": null133}134 