Method : QLORA
Dataset : yahma/alpaca-cleaned
Base model : huggyllama/llama-30b
Compute dtype : bfloat16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 50.12 |
| ARC (25-shot) | 61.77 |
| HellaSwag (10-shot) | 85.06 |
| MMLU (5-shot) | 57.52 |
| TruthfulQA (0-shot) | 51.49 |
| Winogrande (5-shot) | 77.35 |
| GSM8K (5-shot) | 7.73 |
| DROP (3-shot) | 9.91 |