xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 09:25:44 +08:00
Code Issues Packages Projects Releases Wiki Activity

[CI] Nits for bad initialization of SeqGroup in testing (#4748)

Browse Source
This commit is contained in:
Robert Shaw 2024-05-10 16:01:01 -06:00 committed by GitHub
parent 2e7796f2cf
commit fcc2994be6
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 16 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
tests/core/test_block_manager.py
View File

@ -142,8 +142,10 @@ def test_append_slot_cow():
child = prompt.fork(new_seq_id=2)
# Allocate space for the sequence group.
seq_group = SequenceGroup("1", [prompt, child], SamplingParams(),
time.time(), time.perf_counter)
seq_group = SequenceGroup(request_id="1",
seqs=[prompt, child],
arrival_time=time.time(),
sampling_params=SamplingParams())
block_manager.allocate(seq_group)
# Fork and append a new token id. We expect a COW to be scheduled.
@ -303,8 +305,11 @@ def test_sliding_window_multi_seq():
assert block_manager.get_num_free_gpu_blocks() == num_gpu_blocks
parent = Sequence(1, "one two three", [0, 1, 2], block_size)
seq_group = SequenceGroup("1", [parent], SamplingParams(), time.time(),
None)
seq_group = SequenceGroup(request_id="1",
seqs=[parent],
arrival_time=time.time(),
sampling_params=SamplingParams(),
lora_request=None)
block_manager.allocate(seq_group)
# assert the number of blocks allocated is correct

11
tests/core/utils.py
View File

@ -22,10 +22,13 @@ def create_dummy_prompt(
prompt_tokens = list(range(prompt_length))
prompt_str = " ".join([str(t) for t in prompt_tokens])
prompt = Sequence(int(request_id), prompt_str, prompt_tokens, block_size)
seq_group = SequenceGroup(
request_id, [prompt],
SamplingParams(use_beam_search=use_beam_search, best_of=best_of),
time.time(), lora_request)
seq_group = SequenceGroup(request_id=request_id,
seqs=[prompt],
arrival_time=time.time(),
sampling_params=SamplingParams(
use_beam_search=use_beam_search,
best_of=best_of),
lora_request=lora_request)
return prompt, seq_group