From c32a18cbe7342ac0700802b94ae98bbf928a00f0 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Eldar=20Kurti=C4=87?= <8884008+eldarkurtic@users.noreply.github.com> Date: Tue, 25 Nov 2025 20:23:36 +0100 Subject: [PATCH] Attempt to fix GPU OOM in a spec-decoding test (#29419) Signed-off-by: Eldar Kurtic <8884008+eldarkurtic@users.noreply.github.com> --- examples/offline_inference/spec_decode.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/offline_inference/spec_decode.py b/examples/offline_inference/spec_decode.py index 67a0732459..29b2e95d26 100644 --- a/examples/offline_inference/spec_decode.py +++ b/examples/offline_inference/spec_decode.py @@ -133,7 +133,7 @@ def main(args): tensor_parallel_size=args.tp, enable_chunked_prefill=args.enable_chunked_prefill, enforce_eager=args.enforce_eager, - gpu_memory_utilization=0.8, + gpu_memory_utilization=0.9, speculative_config=speculative_config, disable_log_stats=False, max_model_len=args.max_model_len,