---
base_model: meta-llama/Meta-Llama-3-8B-Instruct
inference: false
pipeline_tag: text-generation
language:
- en
license: other
license_name: llama3
license_link: https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct/blob/main/LICENSE
model_creator: meta-llama
model_name: Meta-Llama-3-8B-Instruct
model_type: llama
tags:
- facebook
- meta
- pytorch
- llama
- llama-3
quantized_by: brittlewis12

---

# Meta-Llama-3-8B-Instruct GGUF

**Original model**: [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct)

**Model creator**: [Meta](https://huggingface.co/meta-llama)

This repo contains GGUF format model files for Meta’s Llama-3-8B-Instruct,
**updated as of 2024-04-20** to handle the `<|eot_id|>` special token as EOS token.

Learn more on Meta’s [Llama 3 page](https://llama.meta.com/llama3).

### What is GGUF?

GGUF is a file format for representing AI models. It is the third version of the format, 
introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp. 
Converted with llama.cpp build 2700 (revision [aed82f6](https://github.com/ggerganov/llama.cpp/commit/aed82f6837a3ea515f4d50201cfc77effc7d41b4)),
using [autogguf](https://github.com/brittlewis12/autogguf).

### Prompt template

```
<|start_header_id|>system<|end_header_id|>

{{system_prompt}}<|eot_id|><|start_header_id|>user<|end_header_id|>

{{prompt}}<|eot_id|><|start_header_id|>assistant<|end_header_id|>


```

---

## Download & run with [cnvrs](https://twitter.com/cnvrsai) on iPhone, iPad, and Mac!

![cnvrs.ai](https://pbs.twimg.com/profile_images/1744049151241797632/0mIP-P9e_400x400.jpg)

[cnvrs](https://testflight.apple.com/join/sFWReS7K) is the best app for private, local AI on your device:
- create & save **Characters** with custom system prompts & temperature settings
- download and experiment with any **GGUF model** you can [find on HuggingFace](https://huggingface.co/models?library=gguf)!
- make it your own with custom **Theme colors**
- powered by Metal ⚡️ & [Llama.cpp](https://github.com/ggerganov/llama.cpp), with **haptics** during response streaming!
- **try it out** yourself today, on [Testflight](https://testflight.apple.com/join/sFWReS7K)!
- follow [cnvrs on twitter](https://twitter.com/cnvrsai) to stay up to date

---

## Original Model Evaluation

<table>
  <tr>
   <td><strong>Benchmark</strong>
   </td>
   <td><strong>Llama 3 8B</strong>
   </td>
   <td><strong>Llama 2 7B</strong>
   </td>
   <td><strong>Llama 2 13B</strong>
   </td>
  </tr>
  <tr>
   <td>MMLU (5-shot)
   </td>
   <td><b>68.4</b>
   </td>
   <td>34.1
   </td>
   <td>47.8
   </td>
  </tr>
  <tr>
   <td>GPQA (0-shot)
   </td>
   <td><b>34.2</b>
   </td>
   <td>21.7
   </td>
   <td>22.3
   </td>
  </tr>
  <tr>
   <td>HumanEval (0-shot)
   </td>
   <td><b>62.2</b>
   </td>
   <td>7.9
   </td>
   <td>14.0
   </td>
  </tr>
  <tr>
   <td>GSM-8K (8-shot, CoT)
   </td>
   <td><b>79.6</b>
   </td>
   <td>25.7
   </td>
   <td>77.4
   </td>
  </tr>
  <tr>
   <td>MATH (4-shot, CoT)
   </td>
   <td><b>30.0</b>
   </td>
   <td>3.8
   </td>
   <td>6.7
   </td>
  </tr>
</table>