Instructions to use VisualInference/P-JiT-21K-MLX-8bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use VisualInference/P-JiT-21K-MLX-8bit with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download VisualInference/P-JiT-21K-MLX-8bit --local-dir P-JiT-21K-MLX-8bit
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Download NOTICE from VisualInference/P-JiT-21K-MLX-8bit: direct link, hf CLI and curl.
- Browser
- Download file 701 Bytes
-
https://huggingface.co/VisualInference/P-JiT-21K-MLX-8bit/resolve/main/NOTICE
- Command line
-
hf download hf://VisualInference/P-JiT-21K-MLX-8bit/NOTICE
-
curl -L -o NOTICE https://huggingface.co/VisualInference/P-JiT-21K-MLX-8bit/resolve/main/NOTICE
701 Bytes
| pjit-mlx | |
| Copyright 2026 the pjit-mlx authors. Licensed under the Apache License, Version 2.0 (LICENSE). | |
| This is an MLX port of P-JiT 21K by Linum Inc. (https://github.com/Linum-AI/pjit-21k, | |
| https://huggingface.co/Linum-AI/pjit-21k), Apache License 2.0. The model architecture, sampler and | |
| seeded-noise scheme follow Linum's reference implementation; the released weights are Linum's. | |
| The caption encoder is Qwen3.5-4B by the Qwen team, Alibaba Cloud (https://huggingface.co/Qwen/Qwen3.5-4B), | |
| Apache License 2.0. | |
| pjit_mlx/upsampler.py is Linum's pjit_21k/upsampler.py, unchanged (Apache-2.0). | |
| pjit_mlx/text_encoder.py adapts the transformers-faithful Qwen3.5 layers of moss-image-mlx (Apache-2.0). | |