learning_rate: 0.0001
train_batch_size: 1
eval_batch_size: 1
seed: 14
gradient_accumulation_steps: 8
total_train_batch_size: 8
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: linear
lr_scheduler_warmup_ratio: 0.03
num_epochs: 1

Training Loss	Epoch	Step	Validation Loss
0.2362	0.0808	200	0.1888
0.1939	0.1616	400	0.1836
0.2071	0.2424	600	0.1864
0.2015	0.3232	800	0.1816
0.1942	0.4041	1000	0.1790
0.1769	0.4849	1200	0.1790
0.2015	0.5657	1400	0.1759
0.2056	0.6465	1600	0.1742
0.1906	0.7273	1800	0.1741
0.1892	0.8081	2000	0.1723
0.183	0.8889	2200	0.1707
0.1853	0.9697	2400	0.1706

paulo037/Qwen3-VL-2B-Instruct-legal-extraction-constrained

Qwen3-VL-2B-Instruct-legal-extraction-constrained