learning_rate: 2e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 30

Training Loss	Epoch	Step	Validation Loss
1.132	1.0	1120	1.0509
1.0022	2.0	2240	1.0456
0.879	3.0	3360	1.0670
0.8043	4.0	4480	1.1118
0.7533	5.0	5600	1.1508
0.6813	6.0	6720	1.2327
0.5886	7.0	7840	1.2912
0.5763	8.0	8960	1.3573
0.4803	9.0	10080	1.4066
0.495	10.0	11200	1.4495
0.4392	11.0	12320	1.5231
0.414	12.0	13440	1.5337
0.3899	13.0	14560	1.5533
0.3697	14.0	15680	1.5738
0.379	15.0	16800	1.6026
0.359	16.0	17920	1.6333
0.3412	17.0	19040	1.6379
0.3434	18.0	20160	1.6605
0.3042	19.0	21280	1.6659
0.3229	20.0	22400	1.6751
0.321	21.0	23520	1.6872
0.3178	22.0	24640	1.6906
0.3293	23.0	25760	1.6882
0.2988	24.0	26880	1.6916
0.3235	25.0	28000	1.6961
0.3197	26.0	29120	1.7036
0.3371	27.0	30240	1.6975
0.3042	28.0	31360	1.6991
0.3234	29.0	32480	1.6963
0.3228	30.0	33600	1.6970

g-assismoraes/Qwen3-1.7B-Base-hatebr-ep30

Qwen3-1.7B-Base-hatebr-ep30