[NPU] Enable L1 prefix cache for Kimi-K3 W4A8 accuracy test (gpqa) (#38125)
Co-authored-by: EdwardXuy <EdwardXuy@users.noreply.github.com>
This commit is contained in:
@@ -96,7 +96,6 @@ KIMI_K3_W4A8_32P_OTHER_ARGS = [
|
|||||||
"triton",
|
"triton",
|
||||||
"--speculative-eagle-topk",
|
"--speculative-eagle-topk",
|
||||||
1,
|
1,
|
||||||
"--disable-radix-cache",
|
|
||||||
"--disable-custom-all-reduce",
|
"--disable-custom-all-reduce",
|
||||||
"--watchdog-timeout",
|
"--watchdog-timeout",
|
||||||
9000,
|
9000,
|
||||||
|
|||||||
Reference in New Issue
Block a user