learning_rate: 0.0002
train_batch_size: 64
eval_batch_size: 64
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 128
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
num_epochs: 3

Training Loss	Epoch	Step	Validation Loss	Accuracy
8.6828	0.2096	50	1.0808	0.7447
1.1761	0.4193	100	0.3601	0.8970
0.5759	0.6289	150	0.2304	0.9286
0.4480	0.8386	200	0.1634	0.9498
0.3433	1.0461	250	0.1386	0.9490
0.2805	1.2558	300	0.1368	0.9556
0.2085	1.4654	350	0.1242	0.9627
0.2029	1.6751	400	0.1033	0.9674
0.1771	1.8847	450	0.0664	0.9769
0.1202	2.0922	500	0.0654	0.9795
0.0885	2.3019	550	0.0617	0.9808
0.0818	2.5115	600	0.0548	0.9829
0.0769	2.7212	650	0.0503	0.9824
0.0606	2.9308	700	0.0491	0.9837

delpgiero/customs-classifier-model1

customs-classifier-model1