[MyPy][3/N] Fix MyPy errors in test groups (part 3) (#55939)
Signed-off-by: Taimys <ffy2281693949@gmail.com> Signed-off-by: Luyi Xiao <81631911+luyixiao95@users.noreply.github.com> Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com> Co-authored-by: Codex <noreply@openai.com> Co-authored-by: Luyi Xiao <81631911+luyixiao95@users.noreply.github.com> Co-authored-by: Taimys <ffy2281693949@gmail.com> Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
co-authored by
Codex
Luyi Xiao
Taimys
Harry Mellor
Claude Opus 5.5
parent
c4fa6f36a7
commit
5225f1bacb
@@ -47,7 +47,7 @@ def llm_reranker_processor() -> CrossEncoderIOProcessor:
|
||||
)
|
||||
|
||||
processor = CrossEncoderIOProcessor.__new__(CrossEncoderIOProcessor)
|
||||
processor.model_config = SimpleNamespace(enable_prompt_embeds=False)
|
||||
processor.model_config = SimpleNamespace(enable_prompt_embeds=False) # type: ignore[assignment]
|
||||
processor.tokenizer = TokenizersBackend(
|
||||
tokenizer_object=backend,
|
||||
unk_token="[UNK]",
|
||||
@@ -95,12 +95,16 @@ def test_token_type_ids_stay_aligned_with_a_truncated_padded_prompt():
|
||||
truncation_side="left",
|
||||
)
|
||||
|
||||
prompt = tok_params.apply_post_tokenization(tokenizer, prompt)
|
||||
processed_prompt = tok_params.apply_post_tokenization(
|
||||
tokenizer, # type: ignore[arg-type]
|
||||
prompt,
|
||||
)
|
||||
assert "prompt_token_ids" in processed_prompt
|
||||
token_type_ids = _apply_post_tokenization_to_token_type_ids(
|
||||
tokenizer, tok_params, token_type_ids
|
||||
)
|
||||
|
||||
prompt_token_ids = prompt["prompt_token_ids"]
|
||||
prompt_token_ids = processed_prompt["prompt_token_ids"]
|
||||
assert len(token_type_ids) == len(prompt_token_ids)
|
||||
|
||||
# Keeping the last 40 tokens drops the first 10 query tokens, so 10 query
|
||||
|
||||
@@ -30,12 +30,18 @@ def test_online_forwards_truncate_prompt_tokens_to_proxy(monkeypatch):
|
||||
silently disables truncate_prompt_tokens for Jina rerank/score.
|
||||
"""
|
||||
proc = JinaRankingIOProcessor.__new__(JinaRankingIOProcessor)
|
||||
proc.valid_inputs_online = MagicMock(
|
||||
return_value=ScoringData(data_1=["query"], data_2=["doc"])
|
||||
monkeypatch.setattr(
|
||||
proc,
|
||||
"valid_inputs_online",
|
||||
MagicMock(return_value=ScoringData(data_1=["query"], data_2=["doc"])),
|
||||
)
|
||||
monkeypatch.setattr(proc, "_get_token_limits", MagicMock(return_value=(0, 0)))
|
||||
monkeypatch.setattr(
|
||||
proc, "ensure_str", MagicMock(side_effect=lambda data: list(data))
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
proc, "format_docs_prompts_func", MagicMock(return_value="formatted prompt")
|
||||
)
|
||||
proc._get_token_limits = MagicMock(return_value=(0, 0))
|
||||
proc.ensure_str = MagicMock(side_effect=lambda data: list(data))
|
||||
proc.format_docs_prompts_func = MagicMock(return_value="formatted prompt")
|
||||
|
||||
captured: dict[str, object] = {}
|
||||
|
||||
@@ -71,8 +77,10 @@ def offline_processor_and_context():
|
||||
backend.pre_tokenizer = pre_tokenizers.WhitespaceSplit()
|
||||
proc = JinaRankingIOProcessor.__new__(JinaRankingIOProcessor)
|
||||
proc.tokenizer = TokenizersBackend(tokenizer_object=backend)
|
||||
proc.model_config = SimpleNamespace(max_model_len=1024, is_encoder_decoder=False)
|
||||
proc.renderer = SimpleNamespace(
|
||||
proc.model_config = SimpleNamespace( # type: ignore[assignment]
|
||||
max_model_len=1024, is_encoder_decoder=False
|
||||
)
|
||||
proc.renderer = SimpleNamespace( # type: ignore[assignment]
|
||||
default_cmpl_tok_params=TokenizeParams(max_total_tokens=1024)
|
||||
)
|
||||
ctx = OfflineScoringInputsContext(
|
||||
|
||||
@@ -6,7 +6,7 @@ from unittest.mock import AsyncMock, Mock
|
||||
import pytest
|
||||
|
||||
from vllm.entrypoints.pooling.scoring.serving import ServingScores
|
||||
from vllm.entrypoints.pooling.typing import PoolingServeContext
|
||||
from vllm.entrypoints.pooling.typing import PoolingEngineInput, PoolingServeContext
|
||||
from vllm.inputs import tokens_input
|
||||
from vllm.pooling_params import PoolingParams
|
||||
|
||||
@@ -21,8 +21,18 @@ def _make_context() -> PoolingServeContext:
|
||||
priorities=None,
|
||||
prompt_extras=None,
|
||||
engine_inputs=[
|
||||
tokens_input(prompt_token_ids=[1]),
|
||||
tokens_input(prompt_token_ids=[2]),
|
||||
PoolingEngineInput(
|
||||
prompts=tokens_input(prompt_token_ids=[1]),
|
||||
params=PoolingParams(task="token_embed"),
|
||||
lora_requests=None,
|
||||
priorities=0,
|
||||
),
|
||||
PoolingEngineInput(
|
||||
prompts=tokens_input(prompt_token_ids=[2]),
|
||||
params=PoolingParams(task="token_embed"),
|
||||
lora_requests=None,
|
||||
priorities=0,
|
||||
),
|
||||
],
|
||||
n_queries=1,
|
||||
)
|
||||
@@ -39,10 +49,13 @@ def _query_key(context: PoolingServeContext) -> str:
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_colliding_request_ids_use_distinct_query_cache_keys():
|
||||
async def test_colliding_request_ids_use_distinct_query_cache_keys(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
):
|
||||
serving = object.__new__(ServingScores)
|
||||
serving._prepare_generators = AsyncMock()
|
||||
serving._collect_batch = AsyncMock()
|
||||
prepare_generators = AsyncMock()
|
||||
monkeypatch.setattr(serving, "_prepare_generators", prepare_generators)
|
||||
monkeypatch.setattr(serving, "_collect_batch", AsyncMock())
|
||||
|
||||
first = _make_context()
|
||||
second = _make_context()
|
||||
@@ -50,9 +63,7 @@ async def test_colliding_request_ids_use_distinct_query_cache_keys():
|
||||
await serving._flash_late_interaction_encode_queries(context)
|
||||
await serving._flash_late_interaction_encode_docs(context)
|
||||
|
||||
prepared_contexts = [
|
||||
call.args[0] for call in serving._prepare_generators.await_args_list
|
||||
]
|
||||
prepared_contexts = [call.args[0] for call in prepare_generators.await_args_list]
|
||||
first_query_key = _query_key(prepared_contexts[0])
|
||||
first_doc_key = _query_key(prepared_contexts[1])
|
||||
second_query_key = _query_key(prepared_contexts[2])
|
||||
|
||||
Reference in New Issue
Block a user