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

chore: cleanup garbage code (#29770)

Browse Source
This commit is contained in:
Xiaoyu Zhang
2026-07-02 16:14:01 +08:00
committed by GitHub
parent 0ae76117ef
commit b276a9acee
55 changed files with 86 additions and 262 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/manual/chunked_prefill/test_scripted_regression.py
+1 -1
View File
@@ -112,7 +112,7 @@ class TestRegressionBasic(ScriptedTestCase):
"dbdcdde245 mamba_pool_idx cleanup-skip is mamba-architecture-specific. "
"The shared test fixture does not configure a mamba model; running this "
"regression against the default transformer model would not exercise the "
"mamba NO_TOKEN cleanup path, so the body would be a pure smoke test "
"mamba NO_TOKEN cleanup path, so the body would be a check "
"with no real protection. Re-enable when a mamba fixture is wired in."
)
def test_mamba_chunked_resume_no_token(self):
test/manual/dsv4/_common.py
+3 -3
View File
@@ -8,7 +8,7 @@ and contains one ``CustomTestCase`` subclass per recipe
Each subclass launches the server with the cookbook's exact flags and
runs two sgl-eval evaluations (https://github.com/sgl-project/sgl-eval):
- ``test_smoke_gsm8k`` — short, cheap GSM8K pass to verify the server
can produce coherent math answers at all (smoke gate).
can produce coherent math answers at all (sanity gate).
- ``test_aime25`` — full AIME25 accuracy run (heavy; 16 repeats default).
Cookbook reference:
@@ -19,7 +19,7 @@ These are MANUAL tests (not CI). ``sgl-eval`` must be on PATH.
Per-variant defaults (set on the Flash/Pro intermediate base classes):
Flash recipes -> AIME25 score threshold 0.93
Pro recipes -> AIME25 score threshold 0.95
GSM8K smoke threshold (0.93) is shared across Flash and Pro.
GSM8K sanity threshold (0.93) is shared across Flash and Pro.
AIME25 knobs (env vars):
DSV4_AIME25_NUM_REPEATS (default 16 -> --n-repeats)
@@ -30,7 +30,7 @@ AIME25 knobs (env vars):
DSV4_AIME25_SCORE_METRIC (default "score"; sgl-eval JSON key under "aggregate")
DSV4_AIME25_SCORE_THRESHOLD (default 0; >0 overrides per-variant default)
GSM8K smoke knobs (env vars):
GSM8K sanity knobs (env vars):
DSV4_GSM8K_NUM_EXAMPLES (default 50 -> --num-examples)
DSV4_GSM8K_N_REPEATS (default 1 -> --n-repeats)
DSV4_GSM8K_TEMPERATURE (default 0.6 -> --temperature)