scorcher

[salernos]

Hello mlverse Team!

We have been working on a project called scorcher, an R package to create a tidy-style API for torch. Our goal to make it easier to assemble complex architectures without deep knowledge of torch or OOP. Key features include:

• Pipeline-style model assembly via initiate_scorch(), scorch_input(), scorch_layer(), scorch_output(), and compile_scorch().
• Support for complex architectures: residual blocks, concatenation, multi-headed models, ...
• Easier dataloader creation with scorch_create_dataloader(), handling multi-tensor inputs, automatic dtype conversion, and flexible batching.

We’d love to contribute scorcher to the mlverse family, collaborate on tooling, and make it discoverable alongside other torch extensions. Would you be interested in featuring scorcher in the mlverse? If so, I’d love to know the best way to contribute? Thanks for fostering such a vibrant R/torch ecosystem, and thanks for considering!