32 10 32

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

New activity in failspy/Nemotron-4-340B-Instruct-SafeTensors 5 months ago

Can you share how you converted this?

#1 opened 6 months ago by

bjoernp

New activity in failspy/Nemotron-4-340B-Instruct-SafeTensors 6 months ago

Status on inference?

#2 opened 6 months ago by

ZQ-Dev

New activity in nvidia/Nemotron-4-340B-Instruct 6 months ago

Add to HF Inference APIs

#4 opened 6 months ago by

mrfakename

New activity in failspy/Phi-3-mini-4k-geminified 7 months ago

tokenizer.model 0 bytes?

#2 opened 7 months ago by

beijinghouse

New activity in failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5-GGUF 7 months ago

unknown pre-tokenizer type: 'smaug-bpe'

#1 opened 7 months ago by

kep359

New activity in failspy/Phi-3-vision-128k-instruct-abliterated-alpha 7 months ago

Still get refusals.

#1 opened 7 months ago by

jackboot

New activity in failspy/Phi-3-mini-4k-geminified 7 months ago

The model card unclear xD

#1 opened 7 months ago by

Erilaz

New activity in failspy/Llama-3-70B-Instruct-abliterated-v3-GGUF 7 months ago

Splitting ggufs

#1 opened 7 months ago by

Jobaar

New activity in failspy/kappa-3-phi-abliterated 7 months ago

Ablation Code release

#1 opened 8 months ago by

failspy

New activity in failspy/Phi-3-medium-4k-instruct-abliterated-v3-GGUF 7 months ago

WizardLM-2-8x22B abliteration inquiry

#1 opened 7 months ago by

OrangeApples

New activity in failspy/Phi-3-medium-4k-instruct-abliterated-v3 7 months ago

Upload 2 files

#1 opened 7 months ago by

TheDrummer

New activity in failspy/Llama-3-70B-Instruct-abliterated-v3 7 months ago

What is new in V3 version?

#1 opened 7 months ago by

zypcastles

New activity in cognitivecomputations/Dolphin-2.9.1-Phi-3-Kensho-4.5B 7 months ago

Is this "dolphin" uncensored?

#2 opened 7 months ago by

scubadrv

New activity in cognitivecomputations/Llama-3-8B-Instruct-abliterated-v2 7 months ago

Hallucinations?

#1 opened 8 months ago by deleted

New activity in failspy/Meta-Llama-3-8B-Instruct-abliterated-v3 7 months ago

Possibly missing files

#2 opened 7 months ago by

ayyylol

New activity in failspy/Smaug-Llama-3-70B-Instruct-abliterated-v3 7 months ago

missing tokenizer?

#1 opened 7 months ago by

mradermacher

New activity in failspy/Meta-Llama-3-8B-Instruct-abliterated-v3 7 months ago

GGUF version

#1 opened 7 months ago by

epigene

New activity in failspy/Llama-3-8B-Instruct-abliterated 7 months ago

Update eos_token

#2 opened 7 months ago by

ayyylol

New activity in cognitivecomputations/Llama-3-8B-Instruct-abliterated-v2 7 months ago

Update tokenizer_config.json

#3 opened 7 months ago by

ayyylol

New activity in failspy/llama-3-70B-Instruct-abliterated 7 months ago

Trying to reproduce results

#9 opened 8 months ago by

noisefloordev