File size: 953 Bytes
985de71 68c9fc8 985de71 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 |
---
library_name: transformers
base_model:
- Sao10K/MN-12B-Lyra-v1
datasets:
- jondurbin/gutenberg-dpo-v0.1
license: apache-2.0
---
Mostly quanting this to try it out, didn't see any other quants for EXL2 on this so here we are.
[This is the 8bpw version of this model. Find the original here.](https://huggingface.co/nbeerbower/Lyra-Gutenberg-mistral-nemo-12B)
<br>
[For the 6bpw version, go here](https://huggingface.co/Statuo/Lyra-Gutenberg-12b-EXL2-6bpw)
<br>
[For the 4bpw version, go here](https://huggingface.co/Statuo/Lyra-Gutenberg-12b-EXL2-4bpw)
<br>
# mistral-nemo-gutenberg-12B-v4
[Sao10K/MN-12B-Lyra-v1](https://huggingface.co/Sao10K/MN-12B-Lyra-v1) finetuned on [jondurbin/gutenberg-dpo-v0.1](https://huggingface.co/datasets/jondurbin/gutenberg-dpo-v0.1).
### Method
Finetuned using an A100 on Google Colab for 3 epochs.
[Fine-tune Llama 3 with ORPO](https://mlabonne.github.io/blog/posts/2024-04-19_Fine_tune_Llama_3_with_ORPO.html) |