Hello, I am interested in how the bitmask representation is processed in GPU.
By scanning the nm-vllm code, SparseBitmaskStorageFormat seems to be from a pip package called nm-magic-wand.
I was wondering if the source code that pip package is not open sourced.
Anything you want to discuss about vllm.
Hello, I am interested in how the bitmask representation is processed in GPU. By scanning the nm-vllm code, SparseBitmaskStorageFormat seems to be from a pip package called nm-magic-wand.
I was wondering if the source code that pip package is not open sourced.