Jean-Baptiste Cordonnier

Hi @NamburiSrinath, I am not surprised that this repo is not up to date with new packages versions. See `setup.py` for versions that were working at least at one point...

Did anyone manage to reproduce this bug? Happy to provide more context if needed.

Can you provide the hyperparameters of the training of the ResNet50? Thank you.

Hello, You can check the explanation here #5 maybe? Here is the difference: ```python import torch def downsample_concatenate(X, kernel): b, h, w, c = X.shape Y = X.contiguous().view(b, h, w...

Another beautiful solution with einops: ```python einops.rearrange(x, "batch (h p_h) (w p_w) c -> batch h w (p_h p_w c)", p_h=kernel, p_w=kernel) ```

Hi, Yes maybe, the epsilon pickle file will be ~9GB so you need approximately this much RAM. Can you paste the error message here just to check ?

Unfortunately I cannot share the whole pyproject.toml. Thanks for the pointer at environments, most of the lockfile resolution-markers entries were for darwin/linux, so resolving only for linux should avoid that...

> Does `resolution-markers` contain duplicates? Yes, it was containing many lines like ``` resolution-markers = [ "sys_platform == 'linux'", "sys_platform == 'linux'", "sys_platform == 'linux'", "sys_platform != 'darwin' and sys_platform...

> Separately, do you have any conflicts defined in your pyproject.toml? Yes, we have a conflict defined between two extra, `torch-cpu` and `torch-gpu` that follow the PyTorch guide. We also...

> My guess is that resolution-markers somehow grows exponentially over time... Are you able to share, like, a subset of the pyproject.toml? Unfortunately I cannot, but I can run what...