Guillaume Lagrange
Guillaume Lagrange
Do you have some code that could be helpful to reproduce the issue? Would love to pinpoint the issue. Also, is this uniquely related to the cuda backend (i.e., does...
The slice limitation currently comes from Burn's `tensor.slice(...)` which doesn't accept step != 1 for ranges. We should improve this.
Ah, I think you meant to use `Tensor::cat` instead of `Tensor::stack` here. You see, the `tensor.sum()` outputs a 1D tensor (`Tensor`), and then you're trying to stack them along a...
Just merged the linked PR, but let me know if you're still having issues!
Linked to #2326 This comes from an older `clap` version used by `github-device-flow`. ``` ├── github-device-flow v0.2.0 │ ├── chrono v0.4.38 │ │ ├── iana-time-zone v0.1.61 │ │ └── num-traits...
Ha, I was just discussing this with another user on discord last week 😄 Thanks for the PR, I'll try to look into it today!
> Hey sorry it took me some time, I made some changes and continued the discussion on some of your comments. Lemme know what you think :) No worries 🙂...
No problem, will try to make time to review this today or tomorrow!
Fyi @antimora not all changes have been addressed so further review will be pending until then. The metric should handle the different configurations for binary, multiclass and multi-label (previously suggested...
Broadcasting checks were added in #2213. Seems like your test repo contains multiple models, can you point to the one with the flagged regression? Gotta identify which node with broadcasted...