fs's picture

fs

failspy

·

failspy

AI & ML interests

None yet

Recent Activity

replied to mlabonne's post about 1 month ago

✂️ Uncensor any LLM with abliteration I wrote an article about abliteration and how NeuralDaredevil-8B was created. Beyond removing alignment, I believe it's an interesting technique with a lot of potential. It's basically fine-tuning without retraining. In this article, we see how it works, implement it in Google Colab, and heal the abliterated model to recover the performance drop due to this technique. The final model is an uncensored and high-quality model with the highest MMLU score on the Open LLM Leaderboard (8B category). https://huggingface.co/blog/mlabonne/abliteration

upvoted a collection 2 months ago

View all activity

Organizations

Collections 2

models 32

failspy/concept-vecs-3.1-8b-instruct

Updated Sep 8 • 2

failspy/dragon-plus-GGUF

Updated Jun 26 • 6

failspy/Nemotron-4-340B-Instruct-SafeTensors

Text Generation • Updated Jun 16 • 16 • 23

failspy/Codestral-22B-v0.1-abliterated-v3-GGUF

Updated Jun 3 • 895 • 8

failspy/Codestral-22B-v0.1-abliterated-v3

Text Generation • Updated Jun 3 • 9.22k • 10

failspy/Phi-3-mini-4k-geminified

Text Generation • Updated Jun 3 • 145 • 11

failspy/Llama-3-8B-Instruct-MopeyMule-GGUF

Text Generation • Updated May 30 • 46 • 6

failspy/Llama-3-8B-Instruct-MopeyMule

Text Generation • Updated May 30 • 116 • 70

failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5

Text Generation • Updated May 30 • 4.98k • 37

failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5-GGUF

Updated May 30 • 1.43k • 20

datasets

None public yet