[M3] Enable FP8 sparse GQA (#45744)

Signed-off-by: Thien Tran <gau.nernst@yahoo.com.sg>
Co-authored-by: Yongye Zhu <zyy1102000@gmail.com>
This commit is contained in:
Thien Tran
2026-06-16 21:38:03 -07:00
committed by GitHub
co-authored by Yongye Zhu
parent d78650cf97
commit 4c62663315
10 changed files with 179 additions and 178 deletions
+2 -1
View File
@@ -471,7 +471,8 @@ STABLE_TORCH_LIBRARY_FRAGMENT(_C, ops) {
"int num_index_heads, "
"Tensor? slot_mapping, Tensor? index_slot_mapping, "
"Tensor!? kv_cache, Tensor!? index_cache, "
"int block_size, Tensor!? q_out, Tensor!? index_q_out) -> ()");
"int block_size, Tensor!? q_out, Tensor!? index_q_out, "
"str kv_cache_dtype) -> ()");
// Apply repetition penalties to logits in-place.
ops.def(