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

Refactor global_server_args_dict (#6866)

Browse Source
This commit is contained in:
fzyzcjy
2025-06-07 18:10:35 +08:00
committed by GitHub
parent c499591ac8
commit f5599ef124
2 changed files with 38 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
python/sglang/srt/model_executor/model_runner.py
View File

@@ -65,7 +65,10 @@ from sglang.srt.managers.expert_location import (
get_global_expert_location_metadata,
set_global_expert_location_metadata,
)
from sglang.srt.managers.schedule_batch import global_server_args_dict
from sglang.srt.managers.schedule_batch import (
GLOBAL_SERVER_ARGS_KEYS,
global_server_args_dict,
)
from sglang.srt.mem_cache.memory_pool import (
DoubleSparseTokenToKVPool,
MHATokenToKVPool,
@@ -187,33 +190,10 @@ class ModelRunner:
# Global vars
global_server_args_dict.update(
{
"attention_backend": server_args.attention_backend,
"debug_tensor_dump_inject": server_args.debug_tensor_dump_inject,
"debug_tensor_dump_output_folder": server_args.debug_tensor_dump_output_folder,
"deepep_mode": server_args.deepep_mode,
"device": server_args.device,
"disable_chunked_prefix_cache": server_args.disable_chunked_prefix_cache,
"disable_radix_cache": server_args.disable_radix_cache,
"enable_nan_detection": server_args.enable_nan_detection,
"enable_dp_attention": server_args.enable_dp_attention,
"enable_two_batch_overlap": server_args.enable_two_batch_overlap,
"enable_dp_lm_head": server_args.enable_dp_lm_head,
"enable_ep_moe": server_args.enable_ep_moe,
"enable_deepep_moe": server_args.enable_deepep_moe,
"deepep_config": server_args.deepep_config,
"flashinfer_mla_disable_ragged": server_args.flashinfer_mla_disable_ragged,
"moe_dense_tp_size": server_args.moe_dense_tp_size,
"ep_dispatch_algorithm": server_args.ep_dispatch_algorithm,
"disable_shared_experts_fusion": server_args.disable_shared_experts_fusion,
"triton_attention_reduce_in_fp32": server_args.triton_attention_reduce_in_fp32,
"torchao_config": server_args.torchao_config,
"sampling_backend": server_args.sampling_backend,
"speculative_accept_threshold_single": server_args.speculative_accept_threshold_single,
"speculative_accept_threshold_acc": server_args.speculative_accept_threshold_acc,
{k: getattr(server_args, k) for k in GLOBAL_SERVER_ARGS_KEYS}
| {
# TODO it is indeed not a "server args"
"use_mla_backend": self.use_mla_backend,
"mm_attention_backend": server_args.mm_attention_backend,
"ep_num_redundant_experts": server_args.ep_num_redundant_experts,
}
)