biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Core] Consolidate prompt arguments to LLM engines (#4328)

Browse Source 
Co-authored-by: Roger Wang <ywang@roblox.com>
This commit is contained in:
Cyrus Leung
2024-05-29 04:29:31 +08:00
committed by GitHub
parent 290f4ada2b
commit 5ae5ed1e60
43 changed files with 1407 additions and 442 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
tests/samplers/test_logits_processor.py
View File

@@ -35,28 +35,25 @@ def test_logits_processor_force_generate(
# test logits_processors when prompt_logprobs is not None
vllm_model.model._add_request(
prompt=example_prompts[0],
example_prompts[0],
params=params_with_logprobs,
prompt_token_ids=None,
)
# test prompt_logprobs is not None
vllm_model.model._add_request(
prompt=example_prompts[1],
example_prompts[1],
params=SamplingParams(
prompt_logprobs=3,
max_tokens=max_tokens,
),
prompt_token_ids=None,
)
# test grouped requests
vllm_model.model._add_request(
prompt=example_prompts[2],
example_prompts[2],
params=SamplingParams(max_tokens=max_tokens),
prompt_token_ids=None,
)
outputs = vllm_model.model._run_engine(False)
outputs = vllm_model.model._run_engine(use_tqdm=False)
assert outputs[0].outputs[0].text == enforced_answers * repeat_times