| { | |
| "best_metric": 0.15468917787075043, | |
| "best_model_checkpoint": "/home/paperspace/microsoft_deberta-v3-small_cpt_f6d5daab/checkpoint-14730", | |
| "epoch": 6.0, | |
| "eval_steps": 500, | |
| "global_step": 14730, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.709048588885654e-07, | |
| "loss": 0.5815, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.418097177771311e-07, | |
| "loss": 0.2425, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.127145766656967e-07, | |
| "loss": 0.193, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.836194355542625e-07, | |
| "loss": 0.175, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9521287431248727, | |
| "eval_f1": 0.952353470322894, | |
| "eval_loss": 0.1743677854537964, | |
| "eval_precision": 0.9532059236996064, | |
| "eval_recall": 0.9521287431248727, | |
| "eval_runtime": 3.5334, | |
| "eval_samples_per_second": 1389.316, | |
| "eval_steps_per_second": 86.885, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.54524294442828e-07, | |
| "loss": 0.1522, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 8.254291533313936e-07, | |
| "loss": 0.16, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.963340122199592e-07, | |
| "loss": 0.1608, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 7.672388711085248e-07, | |
| "loss": 0.1629, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.381437299970904e-07, | |
| "loss": 0.1528, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9543695253615808, | |
| "eval_f1": 0.9546171796156795, | |
| "eval_loss": 0.1643744260072708, | |
| "eval_precision": 0.9557276168664036, | |
| "eval_recall": 0.9543695253615808, | |
| "eval_runtime": 3.5744, | |
| "eval_samples_per_second": 1373.394, | |
| "eval_steps_per_second": 85.89, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.090485888856561e-07, | |
| "loss": 0.1699, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 6.799534477742217e-07, | |
| "loss": 0.1587, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 6.508583066627873e-07, | |
| "loss": 0.1488, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 6.217631655513529e-07, | |
| "loss": 0.1473, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 5.926680244399185e-07, | |
| "loss": 0.1475, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.955591770217967, | |
| "eval_f1": 0.9558024411984405, | |
| "eval_loss": 0.1615038961172104, | |
| "eval_precision": 0.9566578816450286, | |
| "eval_recall": 0.955591770217967, | |
| "eval_runtime": 3.5432, | |
| "eval_samples_per_second": 1385.46, | |
| "eval_steps_per_second": 86.644, | |
| "step": 7365 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.63572883328484e-07, | |
| "loss": 0.137, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 5.344777422170497e-07, | |
| "loss": 0.1511, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 5.053826011056153e-07, | |
| "loss": 0.1385, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 4.7628745999418097e-07, | |
| "loss": 0.1321, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 4.4719231888274657e-07, | |
| "loss": 0.142, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9562028926461601, | |
| "eval_f1": 0.9564171496336807, | |
| "eval_loss": 0.15942582488059998, | |
| "eval_precision": 0.9573210042548447, | |
| "eval_recall": 0.9562028926461601, | |
| "eval_runtime": 3.5212, | |
| "eval_samples_per_second": 1394.129, | |
| "eval_steps_per_second": 87.186, | |
| "step": 9820 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.1809717777131217e-07, | |
| "loss": 0.1431, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.890020366598778e-07, | |
| "loss": 0.1224, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 3.599068955484434e-07, | |
| "loss": 0.1391, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.3081175443700896e-07, | |
| "loss": 0.1275, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 3.017166133255746e-07, | |
| "loss": 0.1517, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9553880627419026, | |
| "eval_f1": 0.9556193988382246, | |
| "eval_loss": 0.15658119320869446, | |
| "eval_precision": 0.9566326518090988, | |
| "eval_recall": 0.9553880627419026, | |
| "eval_runtime": 3.5602, | |
| "eval_samples_per_second": 1378.845, | |
| "eval_steps_per_second": 86.23, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.726214722141402e-07, | |
| "loss": 0.1302, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 2.4352633110270586e-07, | |
| "loss": 0.1413, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 2.1443118999127143e-07, | |
| "loss": 0.1318, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 1.8533604887983705e-07, | |
| "loss": 0.1293, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 1.5624090776840268e-07, | |
| "loss": 0.1453, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9570177225504176, | |
| "eval_f1": 0.9572109364992316, | |
| "eval_loss": 0.15468917787075043, | |
| "eval_precision": 0.9579792384583072, | |
| "eval_recall": 0.9570177225504176, | |
| "eval_runtime": 3.5009, | |
| "eval_samples_per_second": 1402.22, | |
| "eval_steps_per_second": 87.692, | |
| "step": 14730 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 17185, | |
| "num_train_epochs": 7, | |
| "save_steps": 500, | |
| "total_flos": 2877426600009204.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |