learning_rate: 2e-05
train_batch_size: 32
eval_batch_size: 8
seed: 42
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: 0.1
num_epochs: 5

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.5836	0.2336	50	1.5637	0.3185	0.1539
1.5205	0.4673	100	nan	0.3260	0.1728
1.0952	0.7009	150	1.0534	0.6260	0.6174
0.9557	0.9346	200	0.9562	0.6491	0.6493
0.8553	1.1682	250	0.8699	0.6994	0.6987
0.7937	1.4019	300	0.8307	0.7116	0.7046
0.7229	1.6355	350	0.7923	0.7266	0.7231
0.6819	1.8692	400	0.7399	0.7532	0.7527
0.5343	2.1028	450	0.7315	0.7636	0.7636
0.5278	2.3364	500	0.7162	0.7665	0.7633
0.5152	2.5701	550	0.6859	0.7769	0.7731
0.5521	2.8037	600	0.6741	0.7827	0.7826
0.5051	3.0374	650	0.6499	0.7780	0.7784
0.357	3.2710	700	0.6639	0.7896	0.7866
0.3958	3.5047	750	0.6831	0.8012	0.7979
0.4774	3.7383	800	0.6565	0.8023	0.7998
0.44	3.9720	850	0.6658	0.8	0.7974
0.337	4.2056	900	0.6507	0.8046	0.8025
0.3425	4.4393	950	0.6648	0.8017	0.7990
0.3214	4.6729	1000	0.6502	0.8052	0.8031
0.3519	4.9065	1050	0.6476	0.8069	0.8052

ejarbe/manus-llm-router

manus-llm-router