Guillaume Lagrange

> This suggests that the error occurs when trying to deserialize the weight of Conv1d (Tensor) into a Linear weight (Tensor) while attempting to match each variant in the enum....

As [mentioned on discord](https://discord.com/channels/1038839012602941528/1336347155409010788/1336428174703333480), this is actually a side effect of the issue #2776. Let me know if you still see the issue, otherwise will close when the repeat dim...

Linking #1970 Guess there might still be an issue with fusion. If you try without the feature (enabled by default), do you still hit the limit?

The issue here is that the book is rendered for the last released version (i.e., 0.13.2) while the examples linked are on the main branch. We could probably link to...

> Thank you for your comment! But even if I check your link (in order to see the examples in version/0.13) the `training.rs` file has a function `run` which does...

The examples are not identical. The guide only uses the wgpu backend while the mnist example is configurable to demonstrate how to use different backends. I agree though that they...

I think you bring a valid point 🙂 we should probably review the examples name to make them a bit more clear

Glad to see you found a workaround, but that's weird 😅 it should work with the default strategy, which is: https://github.com/tracel-ai/burn/blob/27f67188a87e270f231b575908b06741d3441554/crates/burn-train/src/learner/builder.rs#L119-L129 The error you posted ``` thread 'main' panicked at...

Huh, weird. I was hoping for something that could somehow give a bit more info as to why the receiver channel was closed. It looks like the thread exited, but...

Oh thanks for sharing the logs! It does look like something else panicked, and the top level error message is just a product of the drop chain. So unsure how...