Training details

[wanghao-cst]

Awesome work! Will you share the training or fintuning code?

[rese1f]

Thank you for your insterest, please see the closed issue #2
"Our method is training-free, you can implement this mechanism in any model."

[wanghao-cst]

Thank you for your insterest, please see the closed issue #2 "Our method is training-free, you can implement this mechanism in any model."

Thank you for the reply. May I know what is the baseline of model part? It seems like MiniGPT-4. In the paper it illustrates a lot of MLLM.

[rese1f]

We build our model based on video-llama, since it is a simple but strong video-based MLLM.

[wanghao-cst]

Thank you.