rus_gpt2_continued

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.5007

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: 0.005
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 50
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
3.1129	0.1698	1000	3.0504
3.0676	0.3396	2000	3.0015
3.0157	0.5094	3000	2.9528
2.9646	0.6792	4000	2.9015
2.9108	0.8490	5000	2.8427
2.8472	1.0188	6000	2.7835
2.7629	1.1886	7000	2.7155
2.6931	1.3583	8000	2.6456
2.6264	1.5281	9000	2.5798
2.5615	1.6979	10000	2.5268
2.5206	1.8677	11000	2.5007

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1