facebookresearch/AVT

Doubts regarding the Causality

Anirudh257 opened this issue · 2 comments

@rohitgirdhar Thanks for this excellent work and releasing the code.

I went through the code but I am unable to see where the masking of inputs is occurring. In

outputs = self.gpt_model(inputs_embeds=feats,
you directly pass the entire input features to the GPT itself. Does the masking and output per timesteps get taken care by GPT itself?

Hi @Anirudh257 thanks for your interest. Yes, the masking operations happen within the GPT implementation itself.

Thanks for the clarification @rohitgirdhar. I was confused as to how the features were changing at different time-steps.