Video-Text-to-Text
Safetensors
custom_code
ynhe commited on
Commit
be1e3df
β€’
1 Parent(s): 6575e69

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +9 -0
README.md CHANGED
@@ -20,6 +20,15 @@ communicating with open-sourced LLM. In training, the video encoder will be upda
20
 
21
  The BaseLLM of this model is [InternLM2.5-7B](https://huggingface.co/internlm/internlm2_5-7b-chat-1m) with 1M long context window.
22
 
 
 
 
 
 
 
 
 
 
23
  ## πŸš€ How to use the model
24
 
25
  1. make sure to have `transformers >= 4.38.0, peft==0.5.0`
 
20
 
21
  The BaseLLM of this model is [InternLM2.5-7B](https://huggingface.co/internlm/internlm2_5-7b-chat-1m) with 1M long context window.
22
 
23
+ ## πŸ“ˆ Performance
24
+
25
+ | Model | MVBench | VideoMME(w/o sub)|
26
+ | --- | --- | --- |
27
+ |InternVideo2-Chat-8B| 60.3 | 41.9 |
28
+ |InternVideo2-Chat-8B-HD | 65.4 | 46.1|
29
+ |InternVideo2-Chat-8B-HD-F16 | 67.5 | 49.4|
30
+ |InternVideo2-Chat-8B-InternLM| 61.9| **49.1** |
31
+
32
  ## πŸš€ How to use the model
33
 
34
  1. make sure to have `transformers >= 4.38.0, peft==0.5.0`