From 2b1e2111b025d5f832d62167afab8756deaec22d Mon Sep 17 00:00:00 2001 From: Lu Fang <30275821+houseroad@users.noreply.github.com> Date: Wed, 11 Jun 2025 12:54:59 +0800 Subject: [PATCH] Fix test_max_model_len in tests/entrypoints/llm/test_generate.py (#19451) Signed-off-by: Lu Fang --- tests/entrypoints/llm/test_generate.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/tests/entrypoints/llm/test_generate.py b/tests/entrypoints/llm/test_generate.py index 3c3281c34d569..707891f6bdd8d 100644 --- a/tests/entrypoints/llm/test_generate.py +++ b/tests/entrypoints/llm/test_generate.py @@ -125,4 +125,7 @@ def test_max_model_len(): for output in outputs: num_total_tokens = len(output.prompt_token_ids) + len( output.outputs[0].token_ids) - assert num_total_tokens == max_model_len + # Total tokens must not exceed max_model_len. + # It can be less if generation finishes due to other reasons (e.g., EOS) + # before reaching the absolute model length limit. + assert num_total_tokens <= max_model_len