README.md

---
datasets:
- emozilla/yarn-train-tokenized-16k-mistral
metrics:
- perplexity
library_name: transformers
quantized_by: bartowski
---

## Exllama v2 Quantizations of Yarn-Mistral-7b-128k

Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.7">turboderp's ExLlamaV2 v0.0.7</a> for quantization.

Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.

Conversion was done using wikitext-103-raw-v1-test.parquet as calibration dataset.

Original model: https://huggingface.co/NousResearch/Yarn-Mistral-7b-128k
 
<a href="https://huggingface.co/bartowski/Yarn-Mistral-7b-128k-exl2/tree/4.0">4.0 bits per weight</a>
 
<a href="https://huggingface.co/bartowski/Yarn-Mistral-7b-128k-exl2/tree/6.0">6.0 bits per weight</a>
 
<a href="https://huggingface.co/bartowski/Yarn-Mistral-7b-128k-exl2/tree/8.0">8.0 bits per weight</a>

## Download instructions

With git:

```shell
git clone --single-branch --branch 4.0 https://huggingface.co/bartowski/Yarn-Mistral-7b-128k-exl2
```

With huggingface hub (credit to TheBloke for instructions):

```shell
pip3 install huggingface-hub
```

To download the `main` (only useful if you only care about measurement.json) branch to a folder called `Yarn-Mistral-7b-128k-exl2`:

```shell
mkdir Yarn-Mistral-7b-128k-exl2
huggingface-cli download bartowski/Yarn-Mistral-7b-128k-exl2 --local-dir Yarn-Mistral-7b-128k-exl2 --local-dir-use-symlinks False
```

To download from a different branch, add the `--revision` parameter:

```shell
mkdir Yarn-Mistral-7b-128k-exl2
huggingface-cli download bartowski/Yarn-Mistral-7b-128k-exl2 --revision 4.0 --local-dir Yarn-Mistral-7b-128k-exl2 --local-dir-use-symlinks False
```