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 15, 02:15 UTC
[Misc] Relax translations tests (#20856)
Signed-off-by: NickLucche <nlucches@redhat.com>
Nicolò Lucchesi committed
Jul 14, 2025 at 22:08 UTC
149f2435a5b71c96d1bdf3d87e696daf1a793ee7
1 file changed
+3
-4
tests/entrypoints/openai/test_translation_validation.py
+3
-4
@@ -39,8 +39,8 @@ async def test_basic_audio(foscolo):
39
# TODO remove once language detection is implemented
40
extra_body=dict(language="it"),
41
temperature=0.0)
42
- out = json.loads(translation)['text'].strip()
43
- assert "Nor will I ever touch the sacred" in out
42
+ out = json.loads(translation)['text'].strip().lower()
43
+ assert "greek sea" in out
44
45
46
@pytest.mark.asyncio
@@ -168,5 +168,4 @@ async def test_long_audio_request(foscolo):
168
response_format="text",
169
temperature=0.0)
170
out = json.loads(translation)['text'].strip().lower()
171
- # TODO investigate higher model uncertainty in for longer translations.
172
- assert out.count("nor will i ever") == 2
171
+ assert out.count("greek sea") == 2