[Docs] Use trtllm_mha for Qwen3.6 B300 (#29964)
This commit is contained in:
@@ -222,12 +222,9 @@ export const Qwen36Deployment = () => {
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if (hardware === 'b200') {
|
if (hardware === 'b200' || hardware === 'b300') {
|
||||||
cmd += ` \\\n --attention-backend trtllm_mha`;
|
cmd += ` \\\n --attention-backend trtllm_mha`;
|
||||||
}
|
}
|
||||||
if (hardware === 'b300') {
|
|
||||||
cmd += ` \\\n --attention-backend flashinfer`;
|
|
||||||
}
|
|
||||||
if (hwConfig.mem !== undefined) {
|
if (hwConfig.mem !== undefined) {
|
||||||
cmd += ` \\\n --mem-fraction-static ${hwConfig.mem}`;
|
cmd += ` \\\n --mem-fraction-static ${hwConfig.mem}`;
|
||||||
}
|
}
|
||||||
|
|||||||
Reference in New Issue
Block a user