Question on Perceiver Resampler and Vision Decoder.

[guangyu-zhao]

I am sorry that I am only a beginner in AI. I wonder what the difference is between Perceiver Resampler and Vision Decoder. Do they both use the same idea of "​​class token", adding identical learnable token vectors after the transformer decoder, and then using the output of the corresponding positions as the module output?

[StarCycle]

Hi @guangyu-zhao, I guess your understanding is correct ^^