Dilshod Tadjibaev

@syl20bnr can we try updating it again and see if this is fixed? We should try it before the upcoming release.

Labeled this PR as needs help completing it. It would be a great feature to have in Burn.

Before merging, please file uncompleted refactor or fixes.

If we do #1773 alone, then we can deprecate serialization. So you don't need to do other accumulating item types.

Since Burn uses different backends, we should we take advantage of this. Doing purely for CPU will have data transfer overhead for GPU backends. I believe @Gadersd has implemented something...

Having it for the forward pass will cover the majority of use cases. As you mentioned many simply need to extract features and it is better to do using the...

The only thing I could suggest is to file issues for the remaining work so others can pick up or refer to the deficiencies.

I am personally not in favor we support training in the browser. This is such niche use case that requires the code base to accommodate.

If you could provide a minimal code, we will look into this. PyTorch screen shot appears to store weights as an array, which is supported by PyTorchFileRecorder.

Here is what I would recommend you to unblock you quickly. It will be a while since we handle this use case. Create a corresponding PyTorch module for lstm type...