metadata
library_name: peft
base_model: NousResearch/Llama-2-13b-hf
Training procedure
The following bitsandbytes quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: bfloat16
Framework versions
- PEFT 0.6.0.dev0
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 51.77 |
| ARC (25-shot) | 58.36 |
| HellaSwag (10-shot) | 81.03 |
| MMLU (5-shot) | 54.7 |
| TruthfulQA (0-shot) | 52.98 |
| Winogrande (5-shot) | 72.85 |
| GSM8K (5-shot) | 5.53 |
| DROP (3-shot) | 36.94 |