Llama-3.3-70B-Instruct-abliterated

huihui-ai/Llama-3.3-70B-Instruct-abliterated

This is an uncensored version of meta-llama/Llama-3.3-70B-Instruct created with abliteration (see remove-refusals-with-transformers to know more about it).
This is a crude, proof-of-concept implementation to remove refusals from an LLM model without using TransformerLens.

If you don't get the result you want, you can try the same question again.

Use with ollama

You can use huihui_ai/llama3.3-abliterated directly,

ollama run huihui_ai/llama3.3-abliterated

Use with transformers

Starting with transformers >= 4.43.0 onward, you can run conversational inference using the Transformers pipeline abstraction or by leveraging the Auto classes with the generate() function.

Make sure to update your transformers installation via pip install --upgrade transformers.

See the snippet below for usage with Transformers:

import transformers
import torch

model_id = "huihui-ai/Llama-3.3-70B-Instruct-abliterated"

pipeline = transformers.pipeline(
    "text-generation",
    model=model_id,
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)

messages = [
    {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"},
    {"role": "user", "content": "Who are you?"},
]

outputs = pipeline(
    messages,
    max_new_tokens=256,
)
print(outputs[0]["generated_text"][-1])