Dushyant Mehta

I too think it would be useful to have the forward pass and backward pass timings of the models, with different batch sizes and different image resolutions, compared to different...

In my opinion this is a serious enough bug (if I am correct in my assumptions) to warrant putting a note on the leaderboard saying that the speed values should...

and to add one more point to this: The intention of the 'speed' column is to indicate the intended throughput, but even when measured by factoring out the disk read...

Hi Robert Thanks for the detailed response. I can work on a patch to also report the loading + inference breakdown, and perhaps code to try different batch sizes and...

I am using Pytorch 1.9. I can reproduce the reported numbers for avit_tiny if I use avit_small batch size hyperparameters, but the numbers are still different. I get 72.1 for...

I don't recall off the top of my head, but if I recall correctly, the DOFs per joint are specified in one of the asset files that accompany the XNect...

Bumping this @zdaxie I'd like to build upon this work, but I am also running into this error in trying to reproduce the results

@Eashwar93 If the last part of what you said were true, it would be a major Pytorch bug. The derived module carries the parameters of batch norm, but the forward...

@feinanshan This bug raises suspicions about the correctness of the rest of the implementation too vs. the model actually used to generate the results for the paper. It would be...

There is a different bug here then. The bias terms in the convs being fused with BN should be true.