---
library_name: peft
base_model: meta-llama/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 0.4.0

# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_dhmeltzer__Llama-2-13b-hf-ds_eli5_1024_r_64_alpha_16)

| Metric                | Value                     |
|-----------------------|---------------------------|
| Avg.                  | 48.08   |
| ARC (25-shot)         | 60.41          |
| HellaSwag (10-shot)   | 82.58    |
| MMLU (5-shot)         | 55.86         |
| TruthfulQA (0-shot)   | 43.61   |
| Winogrande (5-shot)   | 76.72   |
| GSM8K (5-shot)        | 8.49        |
| DROP (3-shot)         | 8.92         |