Video-Text-to-Text
Transformers
Safetensors
English
qwen3_5
image-text-to-text
streaming-video
video-understanding
on-policy-distillation
knowledge-distillation
Instructions to use UniX-Lab/StreamOPD-4B-ST-CueGate with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use UniX-Lab/StreamOPD-4B-ST-CueGate with Transformers:
# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("UniX-Lab/StreamOPD-4B-ST-CueGate") model = AutoModelForMultimodalLM.from_pretrained("UniX-Lab/StreamOPD-4B-ST-CueGate", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Link the arXiv preprint
Browse files
README.md
CHANGED
|
@@ -9,6 +9,7 @@ tags:
|
|
| 9 |
- on-policy-distillation
|
| 10 |
- knowledge-distillation
|
| 11 |
- qwen3_5
|
|
|
|
| 12 |
language:
|
| 13 |
- en
|
| 14 |
---
|
|
@@ -20,6 +21,7 @@ distillation from a **Qwen3.5-9B** teacher under **ST-CueGate**, the spatio-temp
|
|
| 20 |
gating method from *StreamOPD: A Post-Training Recipe with Spatio-Temporal Cue Gating for
|
| 21 |
Streaming Video Understanding*.
|
| 22 |
|
|
|
|
| 23 |
- **Code:** https://github.com/UniX-AI-Lab/StreamOPD
|
| 24 |
- **Project page:** https://unix-ai-lab.github.io/StreamOPD/
|
| 25 |
|
|
@@ -126,7 +128,7 @@ data pipeline, the parquets, and the cue-generation tooling are in the code repo
|
|
| 126 |
author = {Wu, Keming and Wang, Baoyi and Zhang, Kaichen and An, Xiang and
|
| 127 |
Yang, Zuhao and Wang, Sudong and Zhu, Haowei and Huang, Tingxuan and
|
| 128 |
Gao, Hongcheng and Wang, Bin},
|
| 129 |
-
journal = {arXiv preprint arXiv:
|
| 130 |
year = {2026}
|
| 131 |
}
|
| 132 |
```
|
|
|
|
| 9 |
- on-policy-distillation
|
| 10 |
- knowledge-distillation
|
| 11 |
- qwen3_5
|
| 12 |
+
- arxiv:2608.16320
|
| 13 |
language:
|
| 14 |
- en
|
| 15 |
---
|
|
|
|
| 21 |
gating method from *StreamOPD: A Post-Training Recipe with Spatio-Temporal Cue Gating for
|
| 22 |
Streaming Video Understanding*.
|
| 23 |
|
| 24 |
+
- **Paper:** https://arxiv.org/abs/2608.16320
|
| 25 |
- **Code:** https://github.com/UniX-AI-Lab/StreamOPD
|
| 26 |
- **Project page:** https://unix-ai-lab.github.io/StreamOPD/
|
| 27 |
|
|
|
|
| 128 |
author = {Wu, Keming and Wang, Baoyi and Zhang, Kaichen and An, Xiang and
|
| 129 |
Yang, Zuhao and Wang, Sudong and Zhu, Haowei and Huang, Tingxuan and
|
| 130 |
Gao, Hongcheng and Wang, Bin},
|
| 131 |
+
journal = {arXiv preprint arXiv:2608.16320},
|
| 132 |
year = {2026}
|
| 133 |
}
|
| 134 |
```
|