A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
(Misc): add missing test for zero truncation size. (#23457)
Signed-off-by: teekenl <teekenlau@gmail.com>
TeeKen Lau committed
Aug 24, 2025 at 20:12 UTC
5e021b4981c14724b57744df182d876e6a07e4b9
1 file changed
+22
tests/entrypoints/openai/test_truncation.py
+22
@@ -64,6 +64,28 @@ async def test_smaller_truncation_size(client: openai.AsyncOpenAI):
64
assert response["usage"]["prompt_tokens"] == truncation_size
65
66
67
+@pytest.mark.asyncio
68
+async def test_zero_truncation_size(client: openai.AsyncOpenAI):
69
+ truncation_size = 0
70
+ kwargs: dict[str, Any] = {
71
+ "model": MODEL_NAME,
72
+ "input": input,
73
+ "truncate_prompt_tokens": truncation_size
74
+ }
75
+
76
+ with pytest.raises(openai.BadRequestError) as err:
77
+ await client.post(path="embeddings", cast_to=object, body={**kwargs})
78
+
79
+ assert err.value.status_code == 400
80
+ error_details = err.value.response.json()["error"]
81
+
82
+ assert error_details["type"] == "BadRequestError"
83
+ assert "This model's maximum context length is" in error_details["message"]
84
+ assert "tokens in the input for embedding generation" in error_details[
85
+ "message"]
86
+ assert "Please reduce the length of the input" in error_details["message"]
87
+
88
+
89
@pytest.mark.asyncio
90
async def test_bigger_truncation_size(client: openai.AsyncOpenAI):
91
truncation_size = max_model_len + 1