Implement a single MLA layer (down-proj to latent, up-proj, decoupled RoPE)

Transformer Architecture Internals6 / 7

Chapter 016 / 7

Transformer Architecture Internals

Implement a single MLA layer (down-proj to latent, up-proj, decoupled RoPE)

Hard
Target interface
class MLA

Implement the function/class skeleton in the editor. Any correct approach is accepted.

Hints0 / 2
Reference solution
Your own code stays in the editor.