pretrain-gpt2-large

This model is a fine-tuned version of NlpHUST/gpt2-vietnamese on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.4155

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 16
• total_train_batch_size: 512
• total_eval_batch_size: 2
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 70

Training results

Training Loss	Epoch	Step	Validation Loss
3.0875	13.05	500	2.6828
2.5739	26.1	1000	2.5363
2.4573	39.15	1500	2.4643
2.3962	52.2	2000	2.4294
2.3662	65.25	2500	2.4155

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3