fs's picture

fs

failspy

·

failspy

AI & ML interests

None yet

Recent Activity

replied to mlabonne's post about 2 months 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

View all activity

Organizations

failspy's activity

upvoted a collection 3 months ago

Molmo

Artifacts for open multimodal language models. • 5 items • Updated 25 days ago • 289

upvoted an article 4 months ago

Article

Probabilistic Fractal Activation Function (P-FAF) and Its Advantages Over Traditional Word Vectorization

By

•

Feb 8

• 5

upvoted a collection 6 months ago

Florence

9 items • Updated Jul 11 • 160

upvoted an article 7 months ago

Article

Uncensor any LLM with abliteration

By

•

Jun 13

• 383

upvoted 6 collections 7 months ago

Experimental

[Unstable] Collection of experimental models that may or may not fit in a specific collection. Expect instability and varying performances. • 8 items • Updated May 29 • 1

Centaurus

Series of uncensored models based on Llama-3. • 5 items • Updated May 27 • 2

Orion

Series of uncensored models based on WizardLM-2. • 3 items • Updated May 27 • 1

Ophiuchus

Series of uncensored models based on Phi-3. • 5 items • Updated May 27 • 1

Orthogonalized

This is a collection of orthogonalized/abliterated models by me. • 2 items • Updated May 27 • 1

abliterated-v3

Latest gen of the abliterated models I've produced • 17 items • Updated Jun 3 • 98