From 9e455f55d9bb480fa774067206b5d1b8083d5c54 Mon Sep 17 00:00:00 2001 From: Jiejing <113606+kzjeef@users.noreply.github.com> Date: Sun, 25 Oct 2026 21:36:15 +0000 Subject: [PATCH] serve/pd_server: drop the redundant expert slots when EPLB is off, as openai_server.py does Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01YNgkNrJL5oR1pZqVYzZMHW --- serve/pd_server.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/serve/pd_server.py b/serve/pd_server.py index 3c7813f..342c223 100644 --- a/serve/pd_server.py +++ b/serve/pd_server.py @@ -107,6 +107,9 @@ def worker_main(args: argparse.Namespace, dp_rank: int, in_q: mp.Queue, out_q: m os.environ["DEEPINFER_PREFILL_STORE_EXTRA_STATE"] = "1" os.environ.setdefault("DEEPINFER_PREFILL_STORE_EXTRA_SIZE", "512") # identical state layout on PE and CE os.environ.setdefault("DEEPINFER_PREFILL_ANY_ATTEMPT_INTERVAL", str(args.prefill_any_interval)) + if args.eplb_interval == 0: + # Same as openai_server.py: without EPLB the redundant expert slots only add expert weight per rank. + os.environ.setdefault("DEEPINFER_OVERRIDE_NUM_EXTRA_EXPERTS", "0") setup_common_env(model_name=args.model, sampler_force_topk=1024, set_override_experts=True, enable_eplb=False) if args.eplb_interval > 0: os.environ["DEEPINFER_EXPERT_REARRANGEMENT_STEP_INTERVAL"] = str(args.eplb_interval)