tinygpt_arxiv

This model is trained from scratch using AIGym/TinyGPT2-81M on article abstracts from arxiv. It achieves the following results on the evaluation set:

• Loss: 5.4123

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: 0.0005
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 8
• 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: 500
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
3.5186	17.2414	1000	4.0977
0.5985	34.4828	2000	5.4123

Framework versions

• Transformers 4.42.4
• Pytorch 2.2.2
• Datasets 2.20.0
• Tokenizers 0.19.1