From 8f4f77a7275ecac594f84bdb41b67c95cf3eb26d Mon Sep 17 00:00:00 2001 From: Lucas Wilkinson Date: Wed, 19 Nov 2025 16:43:54 -0500 Subject: [PATCH] [BugFix] Fix false assertion with spec-decode=[2,4,..] and TP>2 (#29036) Signed-off-by: Lucas Wilkinson --- vllm/config/compilation.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vllm/config/compilation.py b/vllm/config/compilation.py index ca01cb3fb55d5..1c3ef502f0f45 100644 --- a/vllm/config/compilation.py +++ b/vllm/config/compilation.py @@ -921,7 +921,7 @@ class CompilationConfig: self, uniform_decode_query_len: int, tensor_parallel_size: int ): multiple_of = uniform_decode_query_len - if tensor_parallel_size > 1: + if tensor_parallel_size > 1 and self.pass_config.enable_sequence_parallelism: multiple_of = max(uniform_decode_query_len, tensor_parallel_size) if ( multiple_of % uniform_decode_query_len != 0