README.md

---
license: llama3.1
---
# <a href="https://huggingface.co/rombodawg/Rombos-LLM-V2.6-Nemotron-70b">Rombos-LLM-V2.6-Nemotron-70b</a> by Rombodawg

---
<p><h2>ExLlamaV2 Quantization</h2></p>
<p>Quantized with <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.2.2">ExLlamaV2 v0.2.3</a></p>

[2.2 Bits Per Weight](https://huggingface.co/UnstableLlama/Rombos-LLM-V2.6-Nemotron-70b-exl2/tree/2_2)

[4.65 Bits Per Weight](https://huggingface.co/UnstableLlama/Rombos-LLM-V2.6-Nemotron-70b-exl2/tree/4_65)

---

![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/642cc1c253e76b4c2286c58e/lLAQ7FGTYDz0xT8rs9-ti.jpeg)

I applied the last step of my continuous finetuning method to the Nemotron-70b model from Nvidia. More details bellow:

- https://docs.google.com/document/d/1OjbjU5AOz4Ftn9xHQrX3oFQGhQ6RDUuXQipnQ9gn6tU/edit?usp=sharing

Quants: (Coming Soon)

Open-LLM-Leaderboard scores: (Coming soon)