3.75bpw EXL2 quant of https://huggingface.co/gghfez/WizardLM-2-8x22B-Beige

This is the biggest quant we can fit into 72GB of VRAM (eg. 3x3090 cards) with a  Q4 cache

---
base_model:
- openbmb/Eurux-8x22b-nca
- alpindale/WizardLM-2-8x22B
- fireworks-ai/mixtral-8x22b-instruct-oh
- migtissera/Tess-2.0-Mixtral-8x22B
library_name: transformers
tags:
- mergekit
- merge

---
# WizardLM-2-8x22B-BigMerge

This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).

## Merge Details
### Merge Method

This model was merged using the [Model Stock](https://arxiv.org/abs/2403.19522) merge method using [alpindale/WizardLM-2-8x22B](https://huggingface.co/alpindale/WizardLM-2-8x22B) as a base.

### Models Merged

The following models were included in the merge:
* [openbmb/Eurux-8x22b-nca](https://huggingface.co/openbmb/Eurux-8x22b-nca)
* [fireworks-ai/mixtral-8x22b-instruct-oh](https://huggingface.co/fireworks-ai/mixtral-8x22b-instruct-oh)
* [migtissera/Tess-2.0-Mixtral-8x22B](https://huggingface.co/migtissera/Tess-2.0-Mixtral-8x22B)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
models:
  - model: alpindale/WizardLM-2-8x22B
  - model: openbmb/Eurux-8x22b-nca
  - model: migtissera/Tess-2.0-Mixtral-8x22B
  - model: fireworks-ai/mixtral-8x22b-instruct-oh
base_model: alpindale/WizardLM-2-8x22B
merge_method: model_stock
dtype: bfloat16

```