[CI/Build] Test both text and token IDs in batched OpenAI Completions API (#5568)

tests/entrypoints/test_openai_server.py

@@ -655,10 +655,12 @@ async def test_completion_stream_options(client: openai.AsyncOpenAI,
     [MODEL_NAME, "zephyr-lora"],
 )
 async def test_batch_completions(client: openai.AsyncOpenAI, model_name: str):
+    # test both text and token IDs
+    for prompts in (["Hello, my name is"] * 2, [[0, 0, 0, 0, 0]] * 2):
         # test simple list
         batch = await client.completions.create(
             model=model_name,
-        prompt=["Hello, my name is", "Hello, my name is"],
+            prompt=prompts,
             max_tokens=5,
             temperature=0.0,
         )
@@ -668,7 +670,7 @@ async def test_batch_completions(client: openai.AsyncOpenAI, model_name: str):
         # test n = 2
         batch = await client.completions.create(
             model=model_name,
-        prompt=["Hello, my name is", "Hello, my name is"],
+            prompt=prompts,
             n=2,
             max_tokens=5,
             temperature=0.0,
@@ -688,7 +690,7 @@ async def test_batch_completions(client: openai.AsyncOpenAI, model_name: str):
         # test streaming
         batch = await client.completions.create(
             model=model_name,
-        prompt=["Hello, my name is", "Hello, my name is"],
+            prompt=prompts,
             max_tokens=5,
             temperature=0.0,
             stream=True,