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
Use kwargs for long lists of `EngineCoreRequest` arguments in tests and fix extra kwargs (#24987)
Signed-off-by: Andrew Sansom <andrew@protopia.ai>
Andrew Sansom committed
Sep 16, 2025 at 16:06 UTC
02d4b854543c3b2c65435a5ed9bb1c3a9856cfad
2 files changed
+16
-18
tests/detokenizer/test_min_tokens.py
+8
-10
@@ -31,16 +31,14 @@ def test_min_tokens_with_stop(min_tokens: int, stop: str, truth: str):
31
stop=stop,
32
min_tokens=min_tokens,
33
)
34
- request = EngineCoreRequest("",
35
- prompt_token_ids,
36
- None,
37
- None,
38
- None,
39
- params,
40
- None,
41
- None,
42
- 0.0,
43
- None,
34
+ request = EngineCoreRequest(request_id="",
35
+ prompt_token_ids=prompt_token_ids,
36
+ mm_features=None,
37
+ sampling_params=params,
38
+ pooling_params=None,
39
+ eos_token_id=None,
40
+ arrival_time=0.0,
41
+ lora_request=None,
42
cache_salt=None,
43
data_parallel_rank=None)
44
tests/tokenization/test_detokenize.py
+8
-8
@@ -61,14 +61,14 @@ def _run_incremental_decode(tokenizer,
61
skip_special_tokens=skip_special_tokens,
62
spaces_between_special_tokens=spaces_between_special_tokens,
63
)
64
- request = EngineCoreRequest("",
65
- prompt_token_ids,
66
- None,
67
- params,
68
- None,
69
- None,
70
- 0.0,
71
- None,
64
+ request = EngineCoreRequest(request_id="",
65
+ prompt_token_ids=prompt_token_ids,
66
+ mm_features=None,
67
+ sampling_params=params,
68
+ pooling_params=None,
69
+ eos_token_id=None,
70
+ arrival_time=0.0,
71
+ lora_request=None,
72
cache_salt=None,
73
data_parallel_rank=None)
74