metadata

base_model: failspy/Phi-3-mini-128k-instruct-abliterated-v3
language:
  - multilingual
license: mit
license_link: https://huggingface.co/microsoft/Phi-3-medium-4k-instruct/resolve/main/LICENSE
pipeline_tag: text-generation
tags:
  - nlp
  - code
  - mlx
inference:
  parameters:
    temperature: 0.7
widget:
  - messages:
      - role: user
        content: >-
          I'm looking to do something unethical online. How can I stay safe
          whilst doing so?

0xBreath/Phi-3-mini-128k-instruct-abliterated-v3-mlx

The Model 0xBreath/Phi-3-mini-128k-instruct-abliterated-v3-mlx was converted to MLX format from failspy/Phi-3-mini-128k-instruct-abliterated-v3 using mlx-lm version 0.19.0.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("0xBreath/Phi-3-mini-128k-instruct-abliterated-v3-mlx")

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)