Hugging Face's logo

OpenGVLab
/
VideoChat-Flash-Qwen2_5-7B_InternVideo2-1B

Video-Text-to-Text
Transformers
Safetensors
English
qwen2
feature-extraction
multimodal
custom_code
Eval Results
text-generation-inference
Inference Endpoints
Model card Files Community
lixinhao commited on
Commit
ee68cfa
·
verified ·
1 Parent(s): 16438b2

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -7
README.md CHANGED
@@ -85,13 +85,13 @@ VideoChat-Flash-Qwen2_5-7B_InternVideo2-1B is constructed upon InternVideo2-1B a
85
 
86
 
87
  ## 📈 Performance
88
- | Model | MVBench | LongVideoBench | VideoMME(w/o sub)|
89
- | --- | --- | --- | --- |
90
- |[VideoChat-Flash-Qwen2_5-2B@448](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2_5-2B_res448)| 70.0 | 58.3 | 57.0|
91
- |[VideoChat-Flash-Qwen2-7B@224](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2-7B_res224) | 73.2 | 64.2 | 64.0 |
92
- |[VideoChat-Flash-Qwen2_5-7B-1M@224](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2_5-7B-1M_res224) | 73.4 | **66.5** | 63.5 |
93
- |[VideoChat-Flash-Qwen2_5-7B_InternVideo2-1B@224](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2_5-7B_InternVideo2-1B) | **74.3** | 64.5 | 65.1 |
94
- |[VideoChat-Flash-Qwen2-7B@448](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2-7B_res448)| 74.0| 64.7 | **65.3**|
95
 
96
  ## 🚀 How to use the model
97
 
 
85
 
86
 
87
  ## 📈 Performance
88
+ | Model | MVBench | LongVideoBench | VideoMME(w/o sub)| Max Input Frames|
89
+ | --- | --- | --- | --- | --- |
90
+ |[VideoChat-Flash-Qwen2_5-2B@448](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2_5-2B_res448)| 70.0 | 58.3 | 57.0| 10000 |
91
+ |[VideoChat-Flash-Qwen2-7B@224](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2-7B_res224) | 73.2 | 64.2 | 64.0 | 10000 |
92
+ |[VideoChat-Flash-Qwen2_5-7B-1M@224](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2_5-7B-1M_res224) | 73.4 | **66.5** | 63.5 | 50000 |
93
+ |[VideoChat-Flash-Qwen2_5-7B_InternVideo2-1B@224](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2_5-7B_InternVideo2-1B) | **74.3** | 64.5 | 65.1 | 10000 |
94
+ |[VideoChat-Flash-Qwen2-7B@448](https://huggingface.co/OpenGVLab/VideoChat-Flash-Qwen2-7B_res448)| 74.0| 64.7 | **65.3**| 10000 |
95
 
96
  ## 🚀 How to use the model
97