README.md

metadata

language: en
tags:
  - obliteratus
  - abliteration
  - uncensored
  - obliterate
  - mlx
  - mlx-my-repo
base_model: usermma/My-Claude-4.6-Thinking-abliterated-failspy

usermma/My-Claude-4.6-Thinking-abliterated-failspy-mlx-6Bit

The Model usermma/My-Claude-4.6-Thinking-abliterated-failspy-mlx-6Bit was converted to MLX format from usermma/My-Claude-4.6-Thinking-abliterated-failspy using mlx-lm version 0.31.2.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("usermma/My-Claude-4.6-Thinking-abliterated-failspy-mlx-6Bit")

prompt="hello"

if hasattr(tokenizer, "apply_chat_template") and tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)