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 11, 23:45 UTC
[CI] Re-use whisper_client for all tests (#30148)
Signed-off-by: NickLucche <nlucches@redhat.com>
Nicolò Lucchesi committed
Dec 5, 2025 at 20:47 UTC
e23ca3a0e8bbf8d1e386bd619e0056ec7255810c
1 file changed
+13
-17
tests/entrypoints/openai/test_transcription_validation_whisper.py
+13
-17
@@ -32,24 +32,20 @@ async def whisper_client(server):
32
33
34
@pytest.mark.asyncio
35
-async def test_basic_audio(mary_had_lamb):
36
- server_args = ["--enforce-eager"]
37
-
35
+async def test_basic_audio(whisper_client, mary_had_lamb):
36
# Based on https://github.com/openai/openai-cookbook/blob/main/examples/Whisper_prompting_guide.ipynb.
39
- with RemoteOpenAIServer(MODEL_NAME, server_args) as remote_server:
40
- client = remote_server.get_async_client()
41
- transcription = await client.audio.transcriptions.create(
42
- model=MODEL_NAME,
43
- file=mary_had_lamb,
44
- language="en",
45
- response_format="text",
46
- temperature=0.0,
47
- )
48
- out = json.loads(transcription)
49
- out_text = out["text"]
50
- out_usage = out["usage"]
51
- assert "Mary had a little lamb," in out_text
52
- assert out_usage["seconds"] == 16, out_usage["seconds"]
37
+ transcription = await whisper_client.audio.transcriptions.create(
38
+ model=MODEL_NAME,
39
+ file=mary_had_lamb,
40
+ language="en",
41
+ response_format="text",
42
+ temperature=0.0,
43
+ )
44
+ out = json.loads(transcription)
45
+ out_text = out["text"]
46
+ out_usage = out["usage"]
47
+ assert "Mary had a little lamb," in out_text
48
+ assert out_usage["seconds"] == 16, out_usage["seconds"]
49
50
51
@pytest.mark.asyncio