Dilshod Tadjibaev

Your proposal suggests moving away from static backend dispatching to an "enum dispatch" approach to enable runtime backend selection. While not explicitly stated, I infer this is the central goal...

Linking prior related tickets: * https://github.com/tracel-ai/burn/issues/2276 * https://github.com/tracel-ai/burn/issues/705

It would be more awesome if we have a book section on this subject too.

I recommend having this example in a separate repo.

Yes, I agree there is a lot of room to improve for loading weights. We haven't focused on this initially. But I think this is perfect time. Since you're in...

We are close to having zero copy loading weights. It will be possible to initialize tensor data and tensors from static region. Here is the plan: 0. https://github.com/tracel-ai/burn/pull/4100 (burnpack data...

Submitted a PR to migrate from record type to burnpack https://github.com/tracel-ai/burn/pull/4122

Zero copy in burn-store and CubeCL's Bytes: https://github.com/tracel-ai/burn/issues/4123

Submitted a PR for CubeCL's Bytes: https://github.com/tracel-ai/cubecl/pull/1093

Reopening it per AdrianEddy request.