-
Vadim Gimpelson authored
[PERF] Qwen3-next MTP speedup (change bool mask indexing to index_select / index_copy to reduce d2h) (#26437) Signed-off-by:Vadim Gimpelson <vadim.gimpelson@gmail.com>
785d8b64
[PERF] Qwen3-next MTP speedup (change bool mask indexing to index_select / index_copy to reduce d2h) (#26437)
Signed-off-by:
Vadim Gimpelson <vadim.gimpelson@gmail.com>