Training in progress, epoch 5

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:74941ca16584c4e830337f4d61bba8e0255f9a3ba78617ab571e17ae88283191
+oid sha256:2d4ad180bbf2a41ac7e4ecd10d30b32006ec55d3abcab4835b84eef4721f2e15
 size 267832560

wandb/run-20240924_172630-x9iddikd/files/output.log

@@ -35,3 +35,4 @@ You should probably TRAIN this model on a down-stream task to be able to use it
 {'eval_loss': 0.24476991593837738, 'eval_accuracy': 0.926, 'eval_runtime': 38.171, 'eval_samples_per_second': 26.198, 'eval_steps_per_second': 0.838, 'epoch': 2.0}
 {'eval_loss': 0.6838799715042114, 'eval_accuracy': 0.656, 'eval_runtime': 214.6826, 'eval_samples_per_second': 4.658, 'eval_steps_per_second': 0.149, 'epoch': 3.0}
 {'loss': 0.2956, 'grad_norm': 2.695140838623047, 'learning_rate': 9.200000000000002e-06, 'epoch': 4.0}
+{'eval_loss': 0.31772053241729736, 'eval_accuracy': 0.87, 'eval_runtime': 37.1806, 'eval_samples_per_second': 26.896, 'eval_steps_per_second': 0.861, 'epoch': 4.0}

wandb/run-20240924_172630-x9iddikd/files/wandb-summary.json

@@ -1 +1 @@
-{"eval/loss": 0.31772053241729736, "eval/accuracy": 0.87, "eval/runtime": 37.1806, "eval/samples_per_second": 26.896, "eval/steps_per_second": 0.861, "train/epoch": 4.0, "train/global_step": 500, "_timestamp": 1727227212.9603, "_runtime": 17622.08739089966, "_step": 12, "train/loss": 0.2956, "train/grad_norm": 2.695140838623047, "train/learning_rate": 9.200000000000002e-06, "train_runtime": 8026.8642, "train_samples_per_second": 2.492, "train_steps_per_second": 0.156, "total_flos": 2396475988298112.0, "train_loss": 0.11480112991333008}
+{"eval/loss": 0.2808445990085602, "eval/accuracy": 0.932, "eval/runtime": 37.3397, "eval/samples_per_second": 26.781, "eval/steps_per_second": 0.857, "train/epoch": 5.0, "train/global_step": 625, "_timestamp": 1727228900.823416, "_runtime": 19309.950506925583, "_step": 13, "train/loss": 0.2956, "train/grad_norm": 2.695140838623047, "train/learning_rate": 9.200000000000002e-06, "train_runtime": 8026.8642, "train_samples_per_second": 2.492, "train_steps_per_second": 0.156, "total_flos": 2396475988298112.0, "train_loss": 0.11480112991333008}