Instructions to use CuckmeisterFuller/Cydonia-22B-v1.3-Q3-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use CuckmeisterFuller/Cydonia-22B-v1.3-Q3-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir Cydonia-22B-v1.3-Q3-mlx CuckmeisterFuller/Cydonia-22B-v1.3-Q3-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
metadata
license: other
base_model: TheDrummer/Cydonia-22B-v1.3
tags:
- mlx
CuckmeisterFuller/Cydonia-22B-v1.3-Q3-mlx
The Model CuckmeisterFuller/Cydonia-22B-v1.3-Q3-mlx was converted to MLX format from TheDrummer/Cydonia-22B-v1.3 using mlx-lm version 0.20.5.
Use with mlx
pip install mlx-lm
from mlx_lm import load, generate
model, tokenizer = load("CuckmeisterFuller/Cydonia-22B-v1.3-Q3-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)