Image-Text-to-Text
Safetensors
MLX
lance-mlx
qwen2_5_vl
lance
vqa
image-understanding
quantized
awq
apple-silicon
4-bit precision
Instructions to use mlx-community/Lance-3B-AWQ-INT4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/Lance-3B-AWQ-INT4 with MLX:
# Make sure mlx-vlm is installed # pip install --upgrade mlx-vlm from mlx_vlm import load, generate from mlx_vlm.prompt_utils import apply_chat_template from mlx_vlm.utils import load_config # Load the model model, processor = load("mlx-community/Lance-3B-AWQ-INT4") config = load_config("mlx-community/Lance-3B-AWQ-INT4") # Prepare input image = ["http://images.cocodataset.org/val2017/000000039769.jpg"] prompt = "Describe this image." # Apply chat template formatted_prompt = apply_chat_template( processor, config, prompt, num_images=1 ) # Generate output output = generate(model, processor, formatted_prompt, image) print(output) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Download NOTICE from mlx-community/Lance-3B-AWQ-INT4: direct link, hf CLI and curl.
- Browser
- Download file 942 Bytes
-
https://huggingface.co/mlx-community/Lance-3B-AWQ-INT4/resolve/main/NOTICE
- Command line
-
hf download hf://mlx-community/Lance-3B-AWQ-INT4/NOTICE
-
curl -L -o NOTICE https://huggingface.co/mlx-community/Lance-3B-AWQ-INT4/resolve/main/NOTICE
942 Bytes
| Lance-3B-AWQ-INT4 | |
| This product is a derivative of Lance (bytedance-research/Lance), originally | |
| created by ByteDance Intelligent Creation Lab and released under the Apache | |
| License 2.0. | |
| Components: | |
| Lance LLM weights (AWQ-INT4 quantized) | |
| Copyright (c) ByteDance Intelligent Creation Lab | |
| Licensed under the Apache License, Version 2.0 | |
| Quantization via AWQ alpha-search algorithm | |
| (algorithm: github.com/Reza2kn/lance-quant — Apache 2.0) | |
| MLX port + AWQ implementation: xocialize/lance-mlx | |
| Wan2.2 3D causal VAE (bf16) | |
| Copyright (c) Alibaba Group / Wan-AI team | |
| Licensed under the Apache License, Version 2.0 | |
| Qwen2.5-VL ViT (vision encoder, used as init, bf16) | |
| Copyright (c) Alibaba Group / Qwen team | |
| Licensed under the Apache License, Version 2.0 | |
| MLX conversion and packaging by MVS Collective (https://github.com/xocialize/lance-mlx) | |
| Licensed under the Apache License, Version 2.0 | |