EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[Revision] Replace enable_flashinfer_mla argument with attention_backend (#5052)

Browse Source
This commit is contained in:
Baizhou Zhang
2025-04-05 01:23:02 -07:00
committed by GitHub
parent ca8d02abd5
commit efbae697b3
9 changed files with 92 additions and 82 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
python/sglang/srt/managers/schedule_batch.py
View File

@@ -76,7 +76,6 @@ global_server_args_dict = {
"device": ServerArgs.device,
"speculative_accept_threshold_single": ServerArgs.speculative_accept_threshold_single,
"speculative_accept_threshold_acc": ServerArgs.speculative_accept_threshold_acc,
"enable_flashinfer_mla": ServerArgs.enable_flashinfer_mla,
"enable_flashmla": ServerArgs.enable_flashmla,
"disable_radix_cache": ServerArgs.disable_radix_cache,
"flashinfer_mla_disable_ragged": ServerArgs.flashinfer_mla_disable_ragged,
@@ -1437,7 +1436,10 @@ class ScheduleBatch(ScheduleBatchDisaggregationDecodeMixin):
# Create seq_lens_cpu when needed
if (
global_server_args_dict["enable_flashinfer_mla"]
(
global_server_args_dict["use_mla_backend"]
and global_server_args_dict["attention_backend"] == "flashinfer"
)
or global_server_args_dict["enable_flashmla"]
or global_server_args_dict["attention_backend"] == "fa3"
):