Video-Text-to-Text
Transformers
Safetensors
English
Chinese
moss_vl
feature-extraction
MOSS-VL
image-understanding
video-understanding
bitsandbytes
NF4
quantized
custom_code
4-bit precision
Instructions to use OpenMOSS-Team/MOSS-VL-Instruct-0708-NF4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use OpenMOSS-Team/MOSS-VL-Instruct-0708-NF4 with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("OpenMOSS-Team/MOSS-VL-Instruct-0708-NF4", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -26,6 +26,10 @@ tags:
|
|
| 26 |
|
| 27 |
# MOSS-VL-Instruct-0708 W4A16 NF4
|
| 28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
This is the Transformers NF4 release of
|
| 30 |
[MOSS-VL-Instruct-0708](https://huggingface.co/OpenMOSS-Team/MOSS-VL-Instruct-0708).
|
| 31 |
It supports image and video inference through the standard MOSS-VL offline
|
|
@@ -205,3 +209,27 @@ Full inputs, commands and raw results:
|
|
| 205 |
- `config.json`: model and bitsandbytes NF4 configuration.
|
| 206 |
- `generation_config.json`: standard generation settings with BF16 KV cache.
|
| 207 |
- `modeling_moss_vl.py`: checkpoint-local offline MOSS-VL code.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
# MOSS-VL-Instruct-0708 W4A16 NF4
|
| 28 |
|
| 29 |
+
MOSS-VL is an open vision-language model family from OpenMOSS, supporting image understanding, long-video understanding, and realtime streaming interaction. This repository provides the W4A16 NF4-quantized checkpoint of MOSS-VL-Instruct-0708.
|
| 30 |
+
|
| 31 |
+
**Technical Report**: [https://arxiv.org/pdf/2608.15045](https://arxiv.org/pdf/2608.15045)
|
| 32 |
+
|
| 33 |
This is the Transformers NF4 release of
|
| 34 |
[MOSS-VL-Instruct-0708](https://huggingface.co/OpenMOSS-Team/MOSS-VL-Instruct-0708).
|
| 35 |
It supports image and video inference through the standard MOSS-VL offline
|
|
|
|
| 209 |
- `config.json`: model and bitsandbytes NF4 configuration.
|
| 210 |
- `generation_config.json`: standard generation settings with BF16 KV cache.
|
| 211 |
- `modeling_moss_vl.py`: checkpoint-local offline MOSS-VL code.
|
| 212 |
+
|
| 213 |
+
## Citation
|
| 214 |
+
|
| 215 |
+
```bibtex
|
| 216 |
+
@misc{mossvl,
|
| 217 |
+
title = {MOSS-VL Technical Report},
|
| 218 |
+
author = {Wang, Pengyu and Tan, Chenkun and Zhou, Shaojun and Zhou, Qirui and Chen, Yanxin and He, Xingyang and Zeng, Huazheng and Cheng, Jijun and Wang, Chenghao and Qian, Xiaomeng and Wang, Pengfei and Huang, Zhan and Gao, Shanqing and Huang, Wei and Cao, Longjun and Ran, Wu and Liu, Jie and Zhu, Changtai and Wang, Hongkai and Tian, Yixian and Liu, Chenghao and Ye, Zhen and Wang, Xinghao and Jiang, Botian and Feng, Guoguo and Fei, Zhaoye and Li, Ruixiao and Chen, Mingshu and Gao, Yang and Cheng, Qinyuan and Li, Shimin and Qiu, Xipeng},
|
| 219 |
+
year = {2026},
|
| 220 |
+
eprint = {2608.15045},
|
| 221 |
+
archivePrefix = {arXiv},
|
| 222 |
+
primaryClass = {cs.CV},
|
| 223 |
+
url = {https://arxiv.org/abs/2608.15045}
|
| 224 |
+
}
|
| 225 |
+
|
| 226 |
+
@misc{mossvideopreview,
|
| 227 |
+
title = {{MOSS-Video-Preview: Toward Real-Time Video Understanding via Cross-Attention}},
|
| 228 |
+
author = {Pengyu Wang and Chenkun Tan and Shaojun Zhou and Wei Huang and Qirui Zhou and Zhan Huang and Zhen Ye and Jijun Cheng and Xiaomeng Qian and Yanxin Chen and Xingyang He and Huazheng Zeng and Chenghao Wang and Pengfei Wang and Hongkai Wang and Shanqing Gao and Yixian Tian and Chenghao Liu and Xinghao Wang and Botian Jiang and Xipeng Qiu},
|
| 229 |
+
year = {2026},
|
| 230 |
+
eprint = {2606.07639},
|
| 231 |
+
archivePrefix = {arXiv},
|
| 232 |
+
primaryClass = {cs.CV},
|
| 233 |
+
url = {https://arxiv.org/abs/2606.07639}
|
| 234 |
+
}
|
| 235 |
+
```
|