Nikolai Ilinykh

Hello everyone, same question here: how do I do it during the decoding phase? During decoding, we have multiple attention weights, since there are several heads and also several layers,...

Ah, I see, I think I found where exactly it is in the code, so all good, thanks. I have another question, more of just learning about image captioning transformer:...

you mean to say, that each individual cell in the grid attends to all cells in the grid? this is why we have 196 x 196?

Is there any update on this issue? Will it be merged? A very nice functionality, I would say. Just a quick question related to how initial states for multi-layer LSTM...

@MenSanYan what would be the intuition behind it? I mean, I have not seen anywhere that people are using other numbers then 1 (as it is in the paper). Could...

This might not be the question directly related to the discussion here, but, perhaps, it's a good place (and time) to ask the question: what would be the theoretical and...

In addition to this problem, I have encountered a few others with other packages. It is important to install numpy==1.21.0 or (probably) higher as, for example, numpy 1.20.0 would cause...

I am not sure if this is the right way, but it seems like simply commenting out the problematic line worked: def load_scene(self, scene) -> None: #embed(); raise #self.config.defrost() #self.config.simulator.scene...

Ok, the issue is still not solved. Commenting these lines out leads to the same scene being used every time. I am not sure how I can use these defrost()...

One hack is to set OmegaConf to True in default.py (habitat/config) and comment out defrost() and freeze() lines. This would reconfigure the simulator with the required scene.