minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Remove # fmt: off from environ.py Envs class (#30153)

Browse Source
This commit is contained in:
Lianmin Zheng
2026-07-05 12:45:24 -07:00
committed by GitHub
parent 92a1f6e06c
commit 8673e85e6c
1 changed files with 11 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/environ.py
+8 -6
View File
@@ -216,7 +216,6 @@ class ToolStrictLevel(IntEnum):
class Envs: class Envs:
# fmt: off
# Model & File Download # Model & File Download
SGLANG_USE_MODELSCOPE = EnvBool(False) SGLANG_USE_MODELSCOPE = EnvBool(False)
@@ -362,7 +361,8 @@ class Envs:
SGLANG_DISAGGREGATION_FORCE_QUERY_PREFILL_DP_RANK = EnvBool(False) SGLANG_DISAGGREGATION_FORCE_QUERY_PREFILL_DP_RANK = EnvBool(False)
# Scheduler: others: # Scheduler: others:
SGLANG_EMPTY_CACHE_INTERVAL = EnvFloat(-1) # in seconds. Set if you observe high memory accumulation over a long serving period. # in seconds. Set if you observe high memory accumulation over a long serving period.
SGLANG_EMPTY_CACHE_INTERVAL = EnvFloat(-1)
SGLANG_DISABLE_CONSECUTIVE_PREFILL_OVERLAP = EnvBool(False) SGLANG_DISABLE_CONSECUTIVE_PREFILL_OVERLAP = EnvBool(False)
# Force-enable the WAR (write-after-read) barrier for the overlap scheduler # Force-enable the WAR (write-after-read) barrier for the overlap scheduler
# even when is_cuda() is False (e.g. AMD/ROCm). On CUDA the barrier is # even when is_cuda() is False (e.g. AMD/ROCm). On CUDA the barrier is
@@ -651,7 +651,9 @@ class Envs:
SGLANG_NIXL_EP_NUM_MAX_DISPATCH_TOKENS_PER_RANK = EnvInt(128) SGLANG_NIXL_EP_NUM_MAX_DISPATCH_TOKENS_PER_RANK = EnvInt(128)
# DSA Backend (canonical names; fall back to SGLANG_NSA_* with deprecation warning) # DSA Backend (canonical names; fall back to SGLANG_NSA_* with deprecation warning)
SGLANG_DSA_FUSE_TOPK = EnvBoolWithAlias(True, deprecated_name="SGLANG_NSA_FUSE_TOPK") SGLANG_DSA_FUSE_TOPK = EnvBoolWithAlias(
True, deprecated_name="SGLANG_NSA_FUSE_TOPK"
)
SGLANG_DSA_TOPK_FLASHINFER_DETERMINISTIC = EnvBool(False) SGLANG_DSA_TOPK_FLASHINFER_DETERMINISTIC = EnvBool(False)
SGLANG_DSA_TOPK_FLASHINFER_TIE_BREAK = EnvStr(None) SGLANG_DSA_TOPK_FLASHINFER_TIE_BREAK = EnvStr(None)
SGLANG_DSA_ENABLE_MTP_PRECOMPUTE_METADATA = EnvBoolWithAlias( SGLANG_DSA_ENABLE_MTP_PRECOMPUTE_METADATA = EnvBoolWithAlias(
@@ -745,7 +747,6 @@ class Envs:
# preserve the user's original tokens to avoid retokenization drift. # preserve the user's original tokens to avoid retokenization drift.
SGLANG_MM_AVOID_RETOKENIZE = EnvBool(True) SGLANG_MM_AVOID_RETOKENIZE = EnvBool(True)
# VLM Item CUDA IPC Transport # VLM Item CUDA IPC Transport
SGLANG_USE_CUDA_IPC_TRANSPORT = EnvBool(False) SGLANG_USE_CUDA_IPC_TRANSPORT = EnvBool(False)
SGLANG_USE_IPC_POOL_HANDLE_CACHE = EnvBool(False) SGLANG_USE_IPC_POOL_HANDLE_CACHE = EnvBool(False)
@@ -784,7 +785,9 @@ class Envs:
SGLANG_NGRAM_FORCE_GREEDY_VERIFY = EnvBool(False) SGLANG_NGRAM_FORCE_GREEDY_VERIFY = EnvBool(False)
# Warmup # Warmup
SGLANG_WARMUP_TIMEOUT = EnvFloat(-1) # in seconds. If a warmup forward batch takes longer than this, the server will crash to prevent hanging. Recommend to increase warmup timeout to 1800 to accommodate some kernel JIT precache e.g. deep gemm # in seconds. If a warmup forward batch takes longer than this, the server will crash to prevent hanging.
# Recommend to increase warmup timeout to 1800 to accommodate some kernel JIT precache e.g. deep gemm
SGLANG_WARMUP_TIMEOUT = EnvFloat(-1)
# HTTP Server # HTTP Server
SGLANG_TIMEOUT_KEEP_ALIVE = EnvInt(5) SGLANG_TIMEOUT_KEEP_ALIVE = EnvInt(5)
@@ -958,7 +961,6 @@ class Envs:
# Aiter # Aiter
SGLANG_USE_AITER_FP8_PER_TOKEN = EnvBool(False) SGLANG_USE_AITER_FP8_PER_TOKEN = EnvBool(False)
# fmt: on
# EPD # EPD
SGLANG_ENCODER_RECV_TIMEOUT = EnvFloat(180.0) SGLANG_ENCODER_RECV_TIMEOUT = EnvFloat(180.0)