diff --git a/tests/entrypoints/llm/test_generate.py b/tests/entrypoints/llm/test_generate.py index 34465b7d27..82f38adfb7 100644 --- a/tests/entrypoints/llm/test_generate.py +++ b/tests/entrypoints/llm/test_generate.py @@ -91,6 +91,12 @@ def test_multiple_priority(llm: LLM): outputs = llm.generate(PROMPTS, sampling_params=None, priority=[]) +def test_single_prompt_priority(llm: LLM): + # Single string prompts should be normalized to one request. + outputs = llm.generate(PROMPTS[0], sampling_params=None, priority=[0]) + assert len(outputs) == 1 + + def test_max_model_len(): max_model_len = 20 llm = LLM( diff --git a/vllm/entrypoints/llm.py b/vllm/entrypoints/llm.py index 3a43dc591a..d135a9ec97 100644 --- a/vllm/entrypoints/llm.py +++ b/vllm/entrypoints/llm.py @@ -1575,7 +1575,7 @@ class LLM: seq_prompts = prompt_to_seq(prompts) seq_params = self._params_to_seq(params, len(seq_prompts)) seq_lora_requests = self._lora_request_to_seq(lora_request, len(seq_prompts)) - seq_priority = self._priority_to_seq(priority, len(prompts)) + seq_priority = self._priority_to_seq(priority, len(seq_prompts)) return self._render_and_add_requests( prompts=(