Guillaume Lagrange

Sorry for the delayed response! > You can get the onnx file here: https://github.com/martinjrobins/diffsol/raw/refs/heads/workspace/examples/neural-ode-weather-prediction/rhs.onnx The link is invalid, but there seems to be a script to generate it [here](https://github.com/martinjrobins/diffsol/blob/main/examples/neural-ode-weather-prediction/src/model/model.py)? Could...

Ok so my initial hypothesis was correct. > If this is valid ONNX, then this is a bug. During code generation we parse the names and format them for variables...

> CCing [@nathanielsimard](https://github.com/nathanielsimard) and [@laggui](https://github.com/laggui), I am also getting this error when I am evaluating Yolo11x but passing with torch backend. There shouldn't be any shader error when using torch...

Thanks for opening this! I'll link [this comment](https://github.com/tracel-ai/burn/pull/2423#discussion_r1818106985) from our previous discussion to add the initial context.

We absolutely need to improve this on docs.rs 😅

We don't currently support the `Split` operator, but we have `tensor.chunk` on the Burn side which could be equivalent to an even split (part of the `Split` spec). So a...

Closing in favor of #2440

So this unit test is failing on your machine? https://github.com/tracel-ai/burn/blob/3ad5343f85b37e49e519ae9250bfc47dd092e876/crates/burn-tensor/src/tests/ops/remainder.rs#L90-L106 For what backend? If it's a GPU backend, what GPU do you have?

We have [`tensor.round()`](https://burn.dev/burn-book/building-blocks/tensor.html#float-operations) but the number of decimal places to round to is not configurable. /edit: if you want to achieve that in the meantime, you can easily do it...

This could be added to the `tensor.round()` as a default implementation, but as you can see it's not so difficult to achieve without 🙂