---
language: 
  - en
base_model: meta-llama/Llama-2-70b-hf
---

4.65bpw exl2 quant of [ausboss/SuperCOT-70B](https://huggingface.co/ausboss/SuperCOT-70B/).  
Calibration done using [wikitext](https://huggingface.co/datasets/wikitext/blob/refs%2Fconvert%2Fparquet/wikitext-103-v1/test/0000.parquet).  
measurements.json file included in repo.  

Original model card below:
----

Special thanks to Alpin, Tav and the rest of the Pygmalion peeps involved in training this one. Its trained on the supercot dataset like my other qloras and models. I'll update the card with more info soon.

Might be a bit overbaked 🧑‍🍳🔥