|
--- |
|
license: llama2 |
|
tags: |
|
- vision-language model |
|
- llama |
|
- video understanding |
|
--- |
|
|
|
# Flash-VStream Model Card |
|
<a href='https://invinciblewyq.github.io/vstream-page/'><img src='https://img.shields.io/badge/Project-Page-Green'></a> |
|
<a href='https://arxiv.org/abs/2406.08085v1'><img src='https://img.shields.io/badge/Paper-Arxiv-red'></a> |
|
|
|
## Model details |
|
We proposed Flash-VStream, a video-language model that simulates the memory mechanism of human. Our model is able to process extremely long video streams in real-time and respond to user queries simultaneously. |
|
|
|
**This is the checkpoint only after stage-1 pretraining.** |
|
|
|
**Please use [the checkpoint](https://huggingface.co/IVGSZ/Flash-VStream-7b) after stage-2 finetuning for better performance.** |
|
|
|
## License |
|
Llama 2 is licensed under the LLAMA 2 Community License, |
|
Copyright (c) Meta Platforms, Inc. All Rights Reserved. |
|
|
|
## Training data |
|
This model is trained based on image data from LLaVA-1.5 dataset, and video data from WebVid and ActivityNet datasets following LLaMA-VID, including |
|
- 558K filtered image-text pairs from LAION/CC/SBU, captioned by BLIP. |
|
- 158K GPT-generated multimodal instruction-following data. |
|
- 450K academic-task-oriented VQA data mixture. |
|
- 40K ShareGPT data. |
|
- 232K video-caption pairs sampled from the WebVid 2.5M dataset. |
|
- 98K videos from ActivityNet with QA pairs from Video-ChatGPT. |
|
|
