Pinkstack/Superthoughts-lite-v1-Q8-mlx

1---2library_name: transformers3tags:4- trl5- grpo6- rl7- superthoughts8- reasoning9- cot10- mlx11- mlx-my-repo12license: apache-2.013datasets:14- openai/gsm8k15- Pinkstack/intructions-sft-sharegpt16language:17- en18base_model: Pinkstack/Superthoughts-lite-v119pipeline_tag: text-generation20---21 22# Pinkstack/Superthoughts-lite-v1-Q8-mlx23 24The Model [Pinkstack/Superthoughts-lite-v1-Q8-mlx](https://huggingface.co/Pinkstack/Superthoughts-lite-v1-Q8-mlx) was converted to MLX format from [Pinkstack/Superthoughts-lite-v1](https://huggingface.co/Pinkstack/Superthoughts-lite-v1) using mlx-lm version **0.20.5**.25 26## Use with mlx27 28```bash29pip install mlx-lm30```31 32```python33from mlx_lm import load, generate34 35model, tokenizer = load("Pinkstack/Superthoughts-lite-v1-Q8-mlx")36 37prompt="hello"38 39if hasattr(tokenizer, "apply_chat_template") and tokenizer.chat_template is not None:40    messages = [{"role": "user", "content": prompt}]41    prompt = tokenizer.apply_chat_template(42        messages, tokenize=False, add_generation_prompt=True43    )44 45response = generate(model, tokenizer, prompt=prompt, verbose=True)46```47