Mistral-Nemo-Instruct-2407-Heretic-v2/README.md

language, base_model, library_name, license, tags

language	base_model	library_name	license	tags
en	unsloth/Mistral-Nemo-Instruct-2407	transformers	apache-2.0	mistral unsloth transformers heretic uncensored decensored abliterated

This is a decensored version of unsloth/Mistral-Nemo-Instruct-2407, made using Heretic v1.2.0

Abliteration parameters

Parameter	Value
direction_index	18.46
attn.o_proj.max_weight	1.29
attn.o_proj.max_weight_position	29.40
attn.o_proj.min_weight	1.08
attn.o_proj.min_weight_distance	18.39
mlp.down_proj.max_weight	1.47
mlp.down_proj.max_weight_position	24.17
mlp.down_proj.min_weight	1.41
mlp.down_proj.min_weight_distance	19.35

Performance

Metric	This model	Original model (unsloth/Mistral-Nemo-Instruct-2407)
KL divergence	0.0574	0 (by definition)
Refusals	4/100	88/100

---

Finetune Mistral, Gemma, Llama 2-5x faster with 70% less memory via Unsloth!

We have a free Google Colab Tesla T4 notebook for Mistral Nemo 12b here: https://colab.research.google.com/drive/17d3U-CAIwzmbDRqbZ9NnpHxCkmXB6LZ0?usp=sharing

✨ Finetune for Free

All notebooks are beginner friendly! Add your dataset, click "Run All", and you'll get a 2x faster finetuned model which can be exported to GGUF, vLLM or uploaded to Hugging Face.

Unsloth supports	Free Notebooks	Performance	Memory use
Llama-3 8b	▶️ Start on Colab	2.4x faster	58% less
Gemma 7b	▶️ Start on Colab	2.4x faster	58% less
Mistral 7b	▶️ Start on Colab	2.2x faster	62% less
Llama-2 7b	▶️ Start on Colab	2.2x faster	43% less
TinyLlama	▶️ Start on Colab	3.9x faster	74% less
CodeLlama 34b A100	▶️ Start on Colab	1.9x faster	27% less
Mistral 7b 1xT4	▶️ Start on Kaggle	5x faster*	62% less
DPO - Zephyr	▶️ Start on Colab	1.9x faster	19% less

• This conversational notebook is useful for ShareGPT ChatML / Vicuna templates.
• This text completion notebook is for raw text. This DPO notebook replicates Zephyr.
• * Kaggle has 2x T4s, but we use 1. Due to overhead, 1x T4 is 5x faster.