learning_rate: 0.0001
train_batch_size: 4
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
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
num_epochs: 20
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss
7.9043	1.0	19	4.0642
6.9174	2.0	38	4.0662
5.1566	3.0	57	4.1224
5.4814	4.0	76	4.2889
3.8926	5.0	95	4.4369
4.2646	6.0	114	4.6888
3.2123	7.0	133	5.0440
3.3893	8.0	152	5.1727
2.9938	9.0	171	5.4164
2.1881	10.0	190	5.5966
2.4999	11.0	209	5.8308
1.9486	12.0	228	5.8855
1.7914	13.0	247	6.0837
1.5970	14.0	266	6.1669
1.3365	15.0	285	6.2004
1.4736	16.0	304	6.2658
1.4250	17.0	323	6.2945
1.3730	18.0	342	6.3520
1.4026	19.0	361	6.3683
1.2102	20.0	380	6.3889

Lilin004/luxun-gpt2-finetuned

luxun-gpt2-finetuned