gpt2-kl_01_03_hscnspecial-hs_cn

This model is a fine-tuned version of gpt2-medium on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5608

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 4
• seed: 21
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
73.5326	0.02	10	69.5739
46.0984	0.04	20	32.9644
13.5448	0.06	30	10.6411
6.7693	0.08	40	4.2565
2.6755	0.1	50	2.0724
1.4407	0.12	60	1.1299
1.082	0.14	70	0.9249
0.8847	0.16	80	0.7332
0.7248	0.18	90	0.6492
0.7073	0.2	100	0.6516
0.6199	0.22	110	0.6608
0.6244	0.24	120	0.6069
0.6289	0.26	130	0.6602
0.6198	0.28	140	0.5919
0.5988	0.3	150	0.6068
0.6156	0.32	160	0.5664
0.5252	0.34	170	0.5626
0.6351	0.36	180	0.5663
0.6246	0.38	190	0.5599
0.5791	0.4	200	0.5609
0.6755	0.42	210	0.5627
0.6367	0.44	220	0.5608

Framework versions

• Transformers 4.29.0.dev0
• Pytorch 1.12.0a0+bd13bc6
• Datasets 2.12.0
• Tokenizers 0.13.3