Open Models Almanac
Methodssoftware implementationkernel & quantization library

single mention · software implementation

AITER sparse MLA decode kernels

AITER kernels used for sparse multi-head latent attention decoding.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See kernel & quantization library for the methods filed beside it.

Filed under software implementation › kernel & quantization library