gpt2-large-lora-sft

This model is a fine-tuned version of gpt2-large on the customized dataset.

Model description

More information needed

Intended uses & limitations

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.00013
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 6
• total_train_batch_size: 6
• total_eval_batch_size: 48
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2.5

Training results

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.1+cu117
• Datasets 2.10.1
• Tokenizers 0.13.3

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	28.05
ARC (25-shot)	26.79
HellaSwag (10-shot)	44.15
MMLU (5-shot)	25.82
TruthfulQA (0-shot)	39.06
Winogrande (5-shot)	55.09
GSM8K (5-shot)	0.0
DROP (3-shot)	5.46