qqtang/sskj
1
2
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update VLLM + GLM + H20

Browse Source
This commit is contained in:
iiGray 2026-07-20 15:13:09 +00:00
parent 866af70760
commit 3f50608aa7
2 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
experiments/h20/glm_h20_vllm_tp_dp_matrix/adaptive_config.env
View File

@ -21,7 +21,7 @@ PLATEAU_PATIENCE="${PLATEAU_PATIENCE:-2}"
# TTFT SLO early-stop settings.
# When ttft_p95_ms exceeds TTFT_SLO_MS, stop searching the current (ISL, OSL)
# shape and move on to the next scenario.
TTFT_SLO_MS="${TTFT_SLO_MS:-4000}"
TTFT_SLO_MS="${TTFT_SLO_MS:-6000}"
ENABLE_TTFT_SLO_STOP="${ENABLE_TTFT_SLO_STOP:-1}"
# Keep the same random workload semantics as the fixed matrix baseline.

2
experiments/h20/glm_h20_vllm_tp_dp_matrix/compare.py
View File

@ -18,7 +18,7 @@ def load_result(result_root: Path) -> dict:
def slo_status(ttft_p95_ms: float, tpot_mean_ms: float,
ttft_limit_ms: float = 3000.0, tpot_limit_ms: float = 50.0) -> str:
ttft_limit_ms: float = 5000.0, tpot_limit_ms: float = 50.0) -> str:
ttft_ok = ttft_p95_ms < ttft_limit_ms
tpot_ok = tpot_mean_ms < tpot_limit_ms
if ttft_ok and tpot_ok: