README.md

---
license: apache-2.0
language:
- en
- fr
- de
- es
- it
- pt
base_model:
- alamios/Mistral-Small-3.1-DRAFT-0.5B
datasets:
- alamios/Mistral-Small-24B-Instruct-2501-Conversations
pipeline_tag: text-generation
library_name: exllamav2
tags:
- qwen
- qwen2.5
- mistral
- mistral-small
- mistral-small-3.1
---
# Mistral-Small-3.1-DRAFT-0.5B-exl2
Original model: [Mistral-Small-3.1-DRAFT-0.5B](https://huggingface.co/alamios/Mistral-Small-3.1-DRAFT-0.5B) by [alamios](https://huggingface.co/alamios)  
Based on: [Qwen2.5-0.5B](https://huggingface.co/Qwen/Qwen2.5-0.5B) by [Qwen](https://huggingface.co/Qwen/Qwen2.5-0.5B)

## Quants
[4bpw h6 (main)](https://huggingface.co/cgus/Mistral-Small-3.1-DRAFT-0.5B-exl2/tree/main)  
[5bpw h6](https://huggingface.co/cgus/Mistral-Small-3.1-DRAFT-0.5B-exl2/tree/5bpw-h6)
[6bpw h6](https://huggingface.co/cgus/Mistral-Small-3.1-DRAFT-0.5B-exl2/tree/6bpw-h6)
[8bpw h8](https://huggingface.co/cgus/Mistral-Small-3.1-DRAFT-0.5B-exl2/tree/8bpw-h8)

## Quantization notes
Made with Exllamav2 with default dataset.  
These quants are meant to be used as a draft model for TabbyAPI.  
8bpw version with FP16 cache probably might be the most reliable option for this purpose.

## Original model card
# Mistral-Small-3.1-DRAFT-0.5B

This model is meant to be used as draft model for speculative decoding with [mistralai/Mistral-Small-3.1-24B-Instruct-2503](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503) or [mistralai/Mistral-Small-24B-Instruct-2501](https://huggingface.co/mistralai/Mistral-Small-24B-Instruct-2501)

# Data info

The data are Mistral's outputs and includes all kind of tasks from various datasets in English, French, German, Spanish, Italian and Portuguese. It has been trained for 2 epochs on 20k unique examples, for a total of 12 million tokens per epoch.