gpt2-gpt2-mc-weight0-epoch15

This model is a fine-tuned version of on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.9633
• Cls loss: 6.8154
• Lm loss: 3.9632
• Cls Accuracy: 0.1337
• Cls F1: 0.0531
• Cls Precision: 0.0331
• Cls Recall: 0.1337
• Perplexity: 52.63

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: 2e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Cls loss	Lm loss	Cls Accuracy	Cls F1	Cls Precision	Cls Recall	Perplexity
4.1973	1.0	3470	4.0341	6.8497	4.0341	0.1331	0.0529	0.0330	0.1331	56.49
4.0446	2.0	6940	3.9948	6.8450	3.9947	0.1337	0.0531	0.0331	0.1337	54.31
3.9714	3.0	10410	3.9795	6.8404	3.9794	0.1337	0.0531	0.0331	0.1337	53.48
3.9176	4.0	13880	3.9686	6.8359	3.9686	0.1337	0.0531	0.0331	0.1337	52.91
3.8739	5.0	17350	3.9580	6.8317	3.9579	0.1331	0.0529	0.0330	0.1331	52.35
3.8359	6.0	20820	3.9591	6.8286	3.9590	0.1331	0.0529	0.0330	0.1331	52.40
3.8035	7.0	24290	3.9585	6.8263	3.9585	0.1331	0.0529	0.0330	0.1331	52.38
3.7762	8.0	27760	3.9585	6.8240	3.9585	0.1331	0.0529	0.0330	0.1331	52.38
3.7517	9.0	31230	3.9567	6.8216	3.9567	0.1337	0.0531	0.0331	0.1337	52.28
3.7313	10.0	34700	3.9599	6.8193	3.9598	0.1337	0.0531	0.0331	0.1337	52.45
3.7131	11.0	38170	3.9606	6.8169	3.9605	0.1337	0.0531	0.0331	0.1337	52.48
3.6982	12.0	41640	3.9614	6.8154	3.9614	0.1337	0.0531	0.0331	0.1337	52.53
3.6862	13.0	45110	3.9623	6.8154	3.9622	0.1337	0.0531	0.0331	0.1337	52.57
3.6767	14.0	48580	3.9621	6.8154	3.9620	0.1337	0.0531	0.0331	0.1337	52.56
3.6711	15.0	52050	3.9633	6.8154	3.9632	0.1337	0.0531	0.0331	0.1337	52.63

Framework versions

• Transformers 4.21.2
• Pytorch 1.12.1
• Datasets 2.4.0
• Tokenizers 0.12.1