Skip to content

GitLab

Unverified Commit 4e256cad authored by Harry Mellor's avatar Harry Mellor Committed by GitHub
Browse files

Remove all references to `yapf` as it's no longer used (#26251) 


Signed-off-by: default avatarHarry Mellor <19981378+hmellor@users.noreply.github.com>
parent d6953beb
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 935 additions and 713 deletions
tests/models/registry.py
View file @ 4e256cad
This diff is collapsed.
tests/multimodal/test_cache.py
View file @ 4e256cad
......@@ -71,25 +71,27 @@ def _dummy_items(
)
# yapf: disable
@pytest.mark.parametrize(
("item", "expected_size"),
[
(_dummy_item("a", {"a1": 100}), 100),
(_dummy_item("a", {"a1": 100, "a2": 110}), 210),
(_dummy_items({"a": {"a1": 100, "a2": 110}, "b": {"b1": 120, "b2": 130}}), 460), # noqa: E501
(_dummy_items({"a": {"a1": 100, "a2": 110}, "b": {"b1": 120, "b2": 130}}).get_data(), 460), # noqa: E501
(
_dummy_items(
{"a": {"a1": 100, "a2": 110}, "b": {"b1": 120, "b2": 130}}
).get_data(),
460,
), # noqa: E501
],
)
# yapf: enable
def test_cache_item_size(item, expected_size):
cache = MultiModalCache.get_lru_cache(2048, type(item))
cache[""] = item
assert cache.currsize == expected_size
prompt_update = PromptInsertion("dummy", "target", "insertion") \
.resolve(0)
prompt_update = PromptInsertion("dummy", "target", "insertion").resolve(0)
cache[""] = MultiModalProcessorCacheItem(item, [prompt_update])
assert cache.currsize == expected_size
......@@ -106,9 +108,9 @@ def _create_vllm_config(
return VllmConfig(
model_config=ModelConfig(
model="llava-hf/llava-onevision-qwen2-0.5b-ov-hf",
mm_processor_cache_gb=mm_processor_cache_gb),
parallel_config=ParallelConfig(
data_parallel_size=1 if enable_ipc else 2),
mm_processor_cache_gb=mm_processor_cache_gb,
),
parallel_config=ParallelConfig(data_parallel_size=1 if enable_ipc else 2),
)
......@@ -124,11 +126,9 @@ def _compare_caches(
seed: int = 0,
):
cache_0_p0 = processor_cache_from_config(config_0, MULTIMODAL_REGISTRY)
cache_0_p1 = engine_receiver_cache_from_config(config_0,
MULTIMODAL_REGISTRY)
cache_0_p1 = engine_receiver_cache_from_config(config_0, MULTIMODAL_REGISTRY)
cache_1_p0 = processor_cache_from_config(config_1, MULTIMODAL_REGISTRY)
cache_1_p1 = engine_receiver_cache_from_config(config_1,
MULTIMODAL_REGISTRY)
cache_1_p1 = engine_receiver_cache_from_config(config_1, MULTIMODAL_REGISTRY)
cache_size_gb = max(
config_0.model_config.multimodal_config.mm_processor_cache_gb,
......@@ -142,8 +142,7 @@ def _compare_caches(
for _ in range(int(item_capacity / hit_rate))
]
all_hashes = [
MultiModalHasher.hash_kwargs(item=item.get_data())
for item in all_items
MultiModalHasher.hash_kwargs(item=item.get_data()) for item in all_items
]
# Should not be used since there is nothing to convert to text
......@@ -162,7 +161,8 @@ def _compare_caches(
for _ in range(is_cached_calls_per_iter):
cache_0_p0.is_cached(selected_hashes)
cache_0_p0_out = [
item for item, _ in cache_0_p0.get_and_update(
item
for item, _ in cache_0_p0.get_and_update(
[(item, prompt_update.content) for item in selected_items],
selected_hashes,
)
......@@ -174,7 +174,8 @@ def _compare_caches(
for _ in range(is_cached_calls_per_iter):
cache_1_p0.is_cached(selected_hashes)
cache_1_p0_out = [
item for item, _ in cache_1_p0.get_and_update(
item
for item, _ in cache_1_p0.get_and_update(
[(item, prompt_update.content) for item in selected_items],
selected_hashes,
)
......@@ -183,14 +184,12 @@ def _compare_caches(
if cache_0_p1 is None:
cache_0_p1_out = cache_0_p0_out
else:
cache_0_p1_out = cache_0_p1.get_and_update(cache_0_p0_out,
selected_hashes)
cache_0_p1_out = cache_0_p1.get_and_update(cache_0_p0_out, selected_hashes)
if cache_1_p1 is None:
cache_1_p1_out = cache_1_p0_out
else:
cache_1_p1_out = cache_1_p1.get_and_update(cache_1_p0_out,
selected_hashes)
cache_1_p1_out = cache_1_p1.get_and_update(cache_1_p0_out, selected_hashes)
assert cache_0_p1_out == cache_1_p1_out, f"Failed at {it=}"
......
tests/multimodal/test_processing.py
View file @ 4e256cad
......@@ -9,9 +9,6 @@ import pytest
from vllm.config import ModelConfig
from vllm.multimodal import MULTIMODAL_REGISTRY
# yapf conflicts with isort for this block
# yapf: disable
from vllm.multimodal.processing import (
InputProcessingContext,
PlaceholderFeaturesInfo,
......@@ -24,8 +21,6 @@ from vllm.multimodal.processing import (
iter_token_matches,
replace_token_matches,
)
# yapf: enable
from vllm.multimodal.profiling import MultiModalProfiler
from vllm.transformers_utils.tokenizer import AnyTokenizer
......@@ -34,7 +29,6 @@ from .utils import random_image
pytestmark = pytest.mark.cpu_test
# yapf: disable
@pytest.mark.parametrize(
("token_ids", "match_ids", "expected"),
[
......@@ -44,34 +38,34 @@ pytestmark = pytest.mark.cpu_test
[32000, 32000, 32000],
[32000],
[
{ "start_idx": 0, "end_idx": 1 },
{ "start_idx": 1, "end_idx": 2 },
{ "start_idx": 2, "end_idx": 3 },
{"start_idx": 0, "end_idx": 1},
{"start_idx": 1, "end_idx": 2},
{"start_idx": 2, "end_idx": 3},
],
),
(
[32000, 32000, 32000],
[32000, 32000],
[{ "start_idx": 0, "end_idx": 2 }],
[{"start_idx": 0, "end_idx": 2}],
),
(
[32000, 32000, 32000],
[32000, 32000, 32000],
[{ "start_idx": 0, "end_idx": 3 }],
[{"start_idx": 0, "end_idx": 3}],
),
(
[9833, 28747, 32000, 32000, 32000, 9833, 28747, 32000, 32000, 918],
[28747, 32000],
[
{ "start_idx": 1, "end_idx": 3 },
{ "start_idx": 6, "end_idx": 8 },
{"start_idx": 1, "end_idx": 3},
{"start_idx": 6, "end_idx": 8},
],
),
(
[9833, 28747, 32000, 32000, 32000, 9833, 28747, 32000, 32000, 918],
[28747, 32000, 32000, 32000],
[
{ "start_idx": 1, "end_idx": 5 },
{"start_idx": 1, "end_idx": 5},
],
),
(
......@@ -82,14 +76,13 @@ pytestmark = pytest.mark.cpu_test
],
)
@pytest.mark.parametrize("start_idx", [0, 4, 8])
# yapf: enable
def test_iter_token_matches(token_ids, match_ids, expected, start_idx):
result = list(iter_token_matches(token_ids, match_ids,
start_idx=start_idx))
result = list(iter_token_matches(token_ids, match_ids, start_idx=start_idx))
# Manually constructed results
assert [item._asdict() for item in result
] == [item for item in expected if item["start_idx"] >= start_idx]
assert [item._asdict() for item in result] == [
item for item in expected if item["start_idx"] >= start_idx
]
# Invariants
match_lens = [end - start for start, end in result]
......@@ -97,7 +90,6 @@ def test_iter_token_matches(token_ids, match_ids, expected, start_idx):
assert all(match_len == len(match_ids) for match_len in match_lens)
# yapf: disable
@pytest.mark.parametrize(
("token_ids", "match_ids", "new_ids", "expected"),
[
......@@ -141,7 +133,6 @@ def test_iter_token_matches(token_ids, match_ids, expected, start_idx):
),
],
)
# yapf: enable
def test_replace_token_matches(token_ids, match_ids, new_ids, expected):
result = replace_token_matches(token_ids, match_ids, new_ids)
......@@ -149,7 +140,6 @@ def test_replace_token_matches(token_ids, match_ids, new_ids, expected):
assert result == expected
# yapf: disable
@pytest.mark.parametrize(
("prompt", "target_by_key", "expected_by_key"),
[
......@@ -166,11 +156,11 @@ def test_replace_token_matches(token_ids, match_ids, new_ids, expected):
"pattern_1": [],
"pattern_2": [],
"pattern_3": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
"pattern_4": [],
"pattern_5": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
},
),
......@@ -186,26 +176,26 @@ def test_replace_token_matches(token_ids, match_ids, new_ids, expected):
},
{
"pattern_1": [
{ "start_idx": 0, "end_idx": 1 },
{ "start_idx": 1, "end_idx": 2 },
{ "start_idx": 2, "end_idx": 3 },
{ "start_idx": 3, "end_idx": 4 },
{"start_idx": 0, "end_idx": 1},
{"start_idx": 1, "end_idx": 2},
{"start_idx": 2, "end_idx": 3},
{"start_idx": 3, "end_idx": 4},
],
"pattern_2": [
{ "start_idx": 0, "end_idx": 2 },
{ "start_idx": 2, "end_idx": 4 },
{"start_idx": 0, "end_idx": 2},
{"start_idx": 2, "end_idx": 4},
],
"pattern_3": [
{ "start_idx": 0, "end_idx": 3 },
{"start_idx": 0, "end_idx": 3},
],
"pattern_4": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
"pattern_5": [
{ "start_idx": 1, "end_idx": 1 },
{"start_idx": 1, "end_idx": 1},
],
"pattern_6": [
{ "start_idx": 4, "end_idx": 4 },
{"start_idx": 4, "end_idx": 4},
],
},
),
......@@ -221,26 +211,25 @@ def test_replace_token_matches(token_ids, match_ids, new_ids, expected):
},
{
"pattern_1": [
{ "start_idx": 1, "end_idx": 3 },
{ "start_idx": 6, "end_idx": 8 },
{"start_idx": 1, "end_idx": 3},
{"start_idx": 6, "end_idx": 8},
],
"pattern_2": [
{ "start_idx": 1, "end_idx": 5 },
{"start_idx": 1, "end_idx": 5},
],
"pattern_3": [],
"pattern_4": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
"pattern_5": [],
"pattern_6": [
{ "start_idx": 10, "end_idx": 10 },
{"start_idx": 10, "end_idx": 10},
],
},
),
],
)
@pytest.mark.parametrize("update_type", [PromptInsertion, PromptReplacement])
# yapf: enable
def test_find_token_matches(
prompt,
target_by_key,
......@@ -272,7 +261,6 @@ def test_find_token_matches(
} == expected_by_key
# yapf: disable
@pytest.mark.parametrize(
("prompt", "target_by_key", "expected_by_key"),
[
......@@ -288,16 +276,16 @@ def test_find_token_matches(
"pattern_5": PromptIndexTargets.end(),
},
{
"pattern_1": [{ "start_idx": 0, "end_idx": 0 }],
"pattern_1": [{"start_idx": 0, "end_idx": 0}],
"pattern_2": [],
"pattern_3": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
"pattern_4": [],
"pattern_5": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
}
},
),
(
"<image><image><image><image>",
......@@ -311,26 +299,26 @@ def test_find_token_matches(
},
{
"pattern_1": [
{ "start_idx": 0, "end_idx": 7 },
{ "start_idx": 7, "end_idx": 14 },
{ "start_idx": 14, "end_idx": 21 },
{ "start_idx": 21, "end_idx": 28 },
{"start_idx": 0, "end_idx": 7},
{"start_idx": 7, "end_idx": 14},
{"start_idx": 14, "end_idx": 21},
{"start_idx": 21, "end_idx": 28},
],
"pattern_2": [
{ "start_idx": 0, "end_idx": 14 },
{ "start_idx": 14, "end_idx": 28 },
{"start_idx": 0, "end_idx": 14},
{"start_idx": 14, "end_idx": 28},
],
"pattern_3": [
{ "start_idx": 0, "end_idx": 21 },
{"start_idx": 0, "end_idx": 21},
],
"pattern_4": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
"pattern_5": [
{ "start_idx": 7, "end_idx": 7 },
{"start_idx": 7, "end_idx": 7},
],
"pattern_6": [
{ "start_idx": 28, "end_idx": 28 },
{"start_idx": 28, "end_idx": 28},
],
},
),
......@@ -346,21 +334,21 @@ def test_find_token_matches(
},
{
"pattern_1": [
{ "start_idx": 0, "end_idx": 13 },
{ "start_idx": 27, "end_idx": 40 },
{"start_idx": 0, "end_idx": 13},
{"start_idx": 27, "end_idx": 40},
],
"pattern_2": [
{ "start_idx": 0, "end_idx": 27 },
{"start_idx": 0, "end_idx": 27},
],
"pattern_3": [],
"pattern_4": [
{ "start_idx": 0, "end_idx": 0 },
{"start_idx": 0, "end_idx": 0},
],
"pattern_5": [
{ "start_idx": 13, "end_idx": 13 },
{"start_idx": 13, "end_idx": 13},
],
"pattern_6": [
{ "start_idx": 48, "end_idx": 48 },
{"start_idx": 48, "end_idx": 48},
],
},
),
......@@ -374,22 +362,21 @@ def test_find_token_matches(
},
{
"pattern_1": [
{ "start_idx": 0, "end_idx": 9 },
{ "start_idx": 16, "end_idx": 25 },
{"start_idx": 0, "end_idx": 9},
{"start_idx": 16, "end_idx": 25},
],
"pattern_2": [
{ "start_idx": 0, "end_idx": 16 },
{ "start_idx": 16, "end_idx": 32 },
{"start_idx": 0, "end_idx": 16},
{"start_idx": 16, "end_idx": 32},
],
"pattern_3": [
{ "start_idx": 0, "end_idx": 25 },
{"start_idx": 0, "end_idx": 25},
],
},
),
],
)
@pytest.mark.parametrize("update_type", [PromptInsertion, PromptReplacement])
# yapf: enable
def test_find_text_matches(
prompt,
target_by_key,
......@@ -421,7 +408,6 @@ def test_find_text_matches(
} == expected_by_key
# yapf: disable
@pytest.mark.parametrize(
("prompt", "target_by_key", "repl_by_key", "expected_by_update_type_mm_count"), # noqa: E501
[
......@@ -549,9 +535,8 @@ def test_find_text_matches(
},
},
),
]
],
)
# yapf: enable
def test_find_update_text(
prompt,
target_by_key,
......@@ -562,13 +547,15 @@ def test_find_update_text(
mock_tokenizer = cast(AnyTokenizer, object())
for (
update_type,
expected_by_mm_count,
update_type,
expected_by_mm_count,
) in expected_by_update_type_mm_count.items():
for mm_count, expected in expected_by_mm_count.items():
mm_prompt_updates = {
key: [[update_type(key, target, repl_by_key[key]).resolve(i)]
for i in range(mm_count)]
key: [
[update_type(key, target, repl_by_key[key]).resolve(i)]
for i in range(mm_count)
]
for key, target in target_by_key.items()
}
......@@ -589,7 +576,6 @@ def test_find_update_text(
assert new_prompt == expected
# yapf: disable
@pytest.mark.parametrize(
("prompt", "target_by_key", "repl_by_key", "expected_by_update_type_mm_count"), # noqa: E501
[
......@@ -615,8 +601,43 @@ def test_find_update_text(
{
PromptInsertion: {
0: [1, 9833, 28747, 32000, 9833, 28747, 32000, 32000, 918],
1: [1, 9833, 28747, 32000, 32000, 32000, 9833, 28747, 32000, 32000, 918, 1550, 918, 1550], # noqa: E501
2: [1, 9833, 28747, 32000, 32000, 32000, 32000, 32000, 9833, 28747, 32000, 32000, 918, 1550, 918, 1550, 1550, 918, 1550], # noqa: E501
1: [
1,
9833,
28747,
32000,
32000,
32000,
9833,
28747,
32000,
32000,
918,
1550,
918,
1550,
], # noqa: E501
2: [
1,
9833,
28747,
32000,
32000,
32000,
32000,
32000,
9833,
28747,
32000,
32000,
918,
1550,
918,
1550,
1550,
918,
1550,
], # noqa: E501
},
PromptReplacement: {
0: [1, 9833, 28747, 32000, 9833, 28747, 32000, 32000, 918],
......@@ -719,9 +740,8 @@ def test_find_update_text(
},
},
),
]
],
)
# yapf: enable
def test_find_update_tokens(
prompt,
target_by_key,
......@@ -732,13 +752,15 @@ def test_find_update_tokens(
mock_tokenizer = cast(AnyTokenizer, object())
for (
update_type,
expected_by_mm_count,
update_type,
expected_by_mm_count,
) in expected_by_update_type_mm_count.items():
for mm_count, expected in expected_by_mm_count.items():
mm_prompt_updates = {
key: [[update_type(key, target, repl_by_key[key]).resolve(i)]
for i in range(mm_count)]
key: [
[update_type(key, target, repl_by_key[key]).resolve(i)]
for i in range(mm_count)
]
for key, target in target_by_key.items()
}
......@@ -759,7 +781,6 @@ def test_find_update_tokens(
assert new_prompt == expected
# yapf: disable
@pytest.mark.parametrize(
"repl_by_key",
[
......@@ -796,8 +817,7 @@ def test_find_update_tokens(
is_embed=None,
),
],
}
},
),
(
[1, 32000, 32000, 9833, 28747, 32000, 32000, 1550, 918, 1550],
......@@ -828,7 +848,7 @@ def test_find_update_tokens(
),
],
# No match for pattern_4 as it has lower priority than pattern_1
}
},
),
(
[1, 32000, 32000, 32000, 32000, 32000, 1550, 918, 1550],
......@@ -867,12 +887,11 @@ def test_find_update_tokens(
is_embed=None,
),
],
}
},
),
]
],
)
@pytest.mark.parametrize("update_type", [PromptInsertion, PromptReplacement])
# yapf: enable
def test_find_mm_placeholders(
repl_by_key,
prompt,
......@@ -899,8 +918,15 @@ def test_find_mm_placeholders(
@pytest.mark.parametrize("model_id", ["llava-hf/llava-v1.6-mistral-7b-hf"])
@pytest.mark.parametrize(
("limit", "num_supported", "is_valid"),
[(0, 0, True), (0, 1, True), (1, 0, False), (1, 1, True), (1, 2, True),
(2, 1, False), (2, 2, True)],
[
(0, 0, True),
(0, 1, True),
(1, 0, False),
(1, 1, True),
(1, 2, True),
(2, 1, False),
(2, 2, True),
],
)
def test_limit_mm_per_prompt_dummy(model_id, limit, num_supported, is_valid):
limit_mm_per_prompt = {"image": limit}
......@@ -930,8 +956,15 @@ def test_limit_mm_per_prompt_dummy(model_id, limit, num_supported, is_valid):
@pytest.mark.parametrize("model_id", ["llava-hf/llava-v1.6-mistral-7b-hf"])
@pytest.mark.parametrize(
("num_images", "limit", "is_valid"),
[(0, 0, True), (0, 1, True), (1, 0, False), (1, 1, True), (1, 2, True),
(2, 1, False), (2, 2, True)],
[
(0, 0, True),
(0, 1, True),
(1, 0, False),
(1, 1, True),
(1, 2, True),
(2, 1, False),
(2, 2, True),
],
)
def test_limit_mm_per_prompt_apply(model_id, num_images, limit, is_valid):
limit_mm_per_prompt = {"image": limit}
......@@ -966,7 +999,6 @@ def test_limit_mm_per_prompt_apply(model_id, num_images, limit, is_valid):
class DummyProcessor:
def __init__(self, a: int = 0, b: int = 0) -> None:
super().__init__()
......@@ -982,7 +1014,6 @@ class DummyProcessor:
return dict(a=a, c=c)
# yapf: disable
@pytest.mark.parametrize("model_id", ["Qwen/Qwen2-VL-2B-Instruct"]) # Dummy
@pytest.mark.parametrize(
("config_kwargs", "inference_kwargs", "expected_kwargs"),
......@@ -996,7 +1027,6 @@ class DummyProcessor:
({"b": 1, "c": 1}, {}, {"a": 0, "b": 1}),
],
)
# yapf: enable
def test_hf_processor_init_kwargs(
model_id,
config_kwargs,
......@@ -1020,7 +1050,6 @@ def test_hf_processor_init_kwargs(
assert getattr(processor, k) == v
# yapf: disable
@pytest.mark.parametrize("model_id", ["Qwen/Qwen2-VL-2B-Instruct"]) # Dummy
@pytest.mark.parametrize(
("config_kwargs", "inference_kwargs", "expected_kwargs"),
......@@ -1034,7 +1063,6 @@ def test_hf_processor_init_kwargs(
({"b": 1, "c": 1}, {}, {"a": 0, "c": 1}),
],
)
# yapf: enable
def test_hf_processor_call_kwargs(
model_id,
config_kwargs,
......
tests/multimodal/test_utils.py
View file @ 4e256cad
......@@ -233,7 +233,6 @@ async def test_fetch_video_http_with_dynamic_loader(
assert metadata_sync["video_backend"] == "opencv_dynamic"
# yapf: disable
@pytest.mark.parametrize(
"case",
[
......@@ -264,7 +263,6 @@ async def test_fetch_video_http_with_dynamic_loader(
("image", 0),
],
),
# Two modalities
## Internally sorted
dict(
......@@ -276,7 +274,7 @@ async def test_fetch_video_http_with_dynamic_loader(
"audio": [
PlaceholderRange(offset=0, length=2),
PlaceholderRange(offset=2, length=3),
]
],
},
expected_modality_idxs=[
("audio", 0),
......@@ -295,7 +293,7 @@ async def test_fetch_video_http_with_dynamic_loader(
"audio": [
PlaceholderRange(offset=5, length=2),
PlaceholderRange(offset=11, length=4),
]
],
},
expected_modality_idxs=[
("image", 0),
......@@ -314,7 +312,7 @@ async def test_fetch_video_http_with_dynamic_loader(
"audio": [
PlaceholderRange(offset=11, length=4),
PlaceholderRange(offset=5, length=2),
]
],
},
expected_modality_idxs=[
("image", 1),
......@@ -323,7 +321,6 @@ async def test_fetch_video_http_with_dynamic_loader(
("audio", 0),
],
),
# Three modalities
## Internally sorted
dict(
......@@ -339,7 +336,7 @@ async def test_fetch_video_http_with_dynamic_loader(
PlaceholderRange(offset=3, length=4),
PlaceholderRange(offset=7, length=5),
PlaceholderRange(offset=12, length=6),
]
],
},
expected_modality_idxs=[
("audio", 0),
......@@ -363,7 +360,7 @@ async def test_fetch_video_http_with_dynamic_loader(
],
"video": [
PlaceholderRange(offset=8, length=5),
]
],
},
expected_modality_idxs=[
("image", 0),
......@@ -386,7 +383,7 @@ async def test_fetch_video_http_with_dynamic_loader(
],
"video": [
PlaceholderRange(offset=8, length=5),
]
],
},
expected_modality_idxs=[
("image", 0),
......@@ -398,7 +395,6 @@ async def test_fetch_video_http_with_dynamic_loader(
),
],
)
# yapf: enable
def test_argsort_mm_positions(case):
mm_positions = case["mm_positions"]
expected_modality_idxs = case["expected_modality_idxs"]
......@@ -413,13 +409,16 @@ def test_argsort_mm_positions(case):
@pytest.mark.parametrize("num_frames", [-1, 32, 1800])
async def test_allowed_media_domains(video_url: str, num_frames: int):
connector = MediaConnector(
media_io_kwargs={"video": {
"num_frames": num_frames,
}},
media_io_kwargs={
"video": {
"num_frames": num_frames,
}
},
allowed_media_domains=[
"www.bogotobogo.com",
"github.com",
])
],
)
video_sync, metadata_sync = connector.fetch_video(video_url)
video_async, metadata_async = await connector.fetch_video_async(video_url)
......
tests/test_inputs.py
View file @ 4e256cad
......@@ -59,48 +59,52 @@ def test_parse_raw_single_batch_string_slice(inputs_slice: slice):
)
# yapf: disable
@pytest.mark.parametrize('mm_processor_kwargs,expected_mm_kwargs', [
(None, [{}, {}]),
({}, [{}, {}]),
({"foo": 100}, [{"foo": 100}, {"foo": 100}]),
([{"foo": 100}, {"bar": 200}], [{"foo": 100}, {"bar": 200}]),
])
# yapf: enable
@pytest.mark.parametrize(
"mm_processor_kwargs,expected_mm_kwargs",
[
(None, [{}, {}]),
({}, [{}, {}]),
({"foo": 100}, [{"foo": 100}, {"foo": 100}]),
([{"foo": 100}, {"bar": 200}], [{"foo": 100}, {"bar": 200}]),
],
)
def test_zip_enc_dec_prompts(mm_processor_kwargs, expected_mm_kwargs):
"""Test mm_processor_kwargs init for zipping enc/dec prompts."""
encoder_prompts = ['An encoder prompt', 'Another encoder prompt']
decoder_prompts = ['A decoder prompt', 'Another decoder prompt']
zipped_prompts = zip_enc_dec_prompts(encoder_prompts, decoder_prompts,
mm_processor_kwargs)
encoder_prompts = ["An encoder prompt", "Another encoder prompt"]
decoder_prompts = ["A decoder prompt", "Another decoder prompt"]
zipped_prompts = zip_enc_dec_prompts(
encoder_prompts, decoder_prompts, mm_processor_kwargs
)
assert len(zipped_prompts) == len(encoder_prompts) == len(decoder_prompts)
for enc, dec, exp_kwargs, zipped in zip(encoder_prompts, decoder_prompts,
expected_mm_kwargs,
zipped_prompts):
for enc, dec, exp_kwargs, zipped in zip(
encoder_prompts, decoder_prompts, expected_mm_kwargs, zipped_prompts
):
assert isinstance(zipped, dict)
assert len(zipped.keys()) == 3
assert zipped['encoder_prompt'] == enc
assert zipped['decoder_prompt'] == dec
assert zipped['mm_processor_kwargs'] == exp_kwargs
@pytest.mark.parametrize("model_id", [
"facebook/opt-125m",
])
@pytest.mark.parametrize("prompt", [
{
"prompt": "",
"multi_modal_data": {
"dummy": []
assert zipped["encoder_prompt"] == enc
assert zipped["decoder_prompt"] == dec
assert zipped["mm_processor_kwargs"] == exp_kwargs
@pytest.mark.parametrize(
"model_id",
[
"facebook/opt-125m",
],
)
@pytest.mark.parametrize(
"prompt",
[
{
"prompt": "",
"multi_modal_data": {"dummy": []},
},
},
{
"prompt_token_ids": [],
"multi_modal_data": {
"dummy": []
{
"prompt_token_ids": [],
"multi_modal_data": {"dummy": []},
},
},
])
],
)
def test_preprocessor_text_no_mm_inputs(model_id, prompt):
model_config = ModelConfig(model=model_id)
tokenizer = init_tokenizer_from_configs(model_config)
......@@ -110,15 +114,19 @@ def test_preprocessor_text_no_mm_inputs(model_id, prompt):
input_preprocessor.preprocess(prompt)
@pytest.mark.parametrize("model_id", [
"facebook/chameleon-7b",
])
@pytest.mark.parametrize("prompt", [
"",
{
"prompt_token_ids": []
},
])
@pytest.mark.parametrize(
"model_id",
[
"facebook/chameleon-7b",
],
)
@pytest.mark.parametrize(
"prompt",
[
"",
{"prompt_token_ids": []},
],
)
def test_preprocessor_always_mm_code_path(model_id, prompt):
model_config = ModelConfig(model=model_id)
tokenizer = init_tokenizer_from_configs(model_config)
......
tests/tpu/test_moe_pallas.py
View file @ 4e256cad
......@@ -9,14 +9,10 @@ import pytest
import torch
import torch_xla
# yapf conflicts with isort for this block
# yapf: disable
from vllm.model_executor.layers.fused_moe.moe_pallas import fused_moe as pallas_moe
from vllm.model_executor.layers.fused_moe.moe_torch_iterative import (
fused_moe as torch_moe,
)
# yapf: enable
from vllm.platforms import current_platform
if not current_platform.is_tpu():
......
tests/utils_/test_utils.py
View file @ 4e256cad
......@@ -388,7 +388,6 @@ def test_duplicate_dict_args(caplog_vllm, parser):
assert "-O.level" in caplog_vllm.text
# yapf: enable
@pytest.mark.parametrize(
"callable,kw_name,requires_kw_only,allow_var_kwargs,is_supported",
[
......@@ -408,7 +407,6 @@ def test_duplicate_dict_args(caplog_vllm, parser):
(lambda foo, **kwargs: None, "foo", True, True, False),
],
)
# yapf: disable
def test_supports_kw(
callable, kw_name, requires_kw_only, allow_var_kwargs, is_supported
):
......@@ -681,7 +679,6 @@ def test_lru_cache():
assert 6 in cache
# yapf: disable
@pytest.mark.parametrize(
("src_dtype", "tgt_dtype", "expected_result"),
[
......@@ -715,12 +712,10 @@ def test_lru_cache():
(torch.complex64, torch.complex32, False),
],
)
# yapf: enable
def test_is_lossless_cast(src_dtype, tgt_dtype, expected_result):
assert is_lossless_cast(src_dtype, tgt_dtype) == expected_result
# yapf: disable
@pytest.mark.parametrize(
("dtypes", "expected_result"),
[
......@@ -730,7 +725,6 @@ def test_is_lossless_cast(src_dtype, tgt_dtype, expected_result):
([torch.bool, torch.int8, torch.float16, torch.complex32], torch.complex32), # noqa: E501
],
)
# yapf: enable
def test_common_broadcastable_dtype(dtypes, expected_result):
assert common_broadcastable_dtype(dtypes) == expected_result
......@@ -775,7 +769,6 @@ def test_placeholder_module_error_handling():
_ = placeholder_attr.module
# yapf: disable
@pytest.mark.parametrize(
"obj,key1,key2",
[
......@@ -785,8 +778,8 @@ def test_placeholder_module_error_handling():
({1: "a", 2: "b"}, 1, 3),
# Tests for both keys do not exist
({1: "a", 2: "b"}, 3, 4),
])
# yapf: enable
],
)
def test_swap_dict_values(obj, key1, key2):
original_obj = obj.copy()
swap_dict_values(obj, key1, key2)
......@@ -800,26 +793,30 @@ def test_swap_dict_values(obj, key1, key2):
assert key1 not in obj
def test_model_specification(parser_with_config, cli_config_file,
cli_config_file_with_model):
def test_model_specification(
parser_with_config, cli_config_file, cli_config_file_with_model
):
# Test model in CLI takes precedence over config
args = parser_with_config.parse_args(
['serve', 'cli-model', '--config', cli_config_file_with_model])
assert args.model_tag == 'cli-model'
assert args.served_model_name == 'mymodel'
["serve", "cli-model", "--config", cli_config_file_with_model]
)
assert args.model_tag == "cli-model"
assert args.served_model_name == "mymodel"
# Test model from config file works
args = parser_with_config.parse_args([
'serve',
'--config',
cli_config_file_with_model,
])
assert args.model == 'config-model'
assert args.served_model_name == 'mymodel'
args = parser_with_config.parse_args(
[
"serve",
"--config",
cli_config_file_with_model,
]
)
assert args.model == "config-model"
assert args.served_model_name == "mymodel"
# Test no model specified anywhere raises error
with pytest.raises(ValueError, match="No model specified!"):
parser_with_config.parse_args(['serve', '--config', cli_config_file])
parser_with_config.parse_args(["serve", "--config", cli_config_file])
# Test using --model option raises error
# with pytest.raises(
......@@ -833,47 +830,52 @@ def test_model_specification(parser_with_config, cli_config_file,
# Test using --model option back-compatibility
# (when back-compatibility ends, the above test should be uncommented
# and the below test should be removed)
args = parser_with_config.parse_args([
'serve',
'--tensor-parallel-size',
'2',
'--model',
'my-model',
'--trust-remote-code',
'--port',
'8001',
])
args = parser_with_config.parse_args(
[
"serve",
"--tensor-parallel-size",
"2",
"--model",
"my-model",
"--trust-remote-code",
"--port",
"8001",
]
)
assert args.model is None
assert args.tensor_parallel_size == 2
assert args.trust_remote_code is True
assert args.port == 8001
args = parser_with_config.parse_args([
'serve',
'--tensor-parallel-size=2',
'--model=my-model',
'--trust-remote-code',
'--port=8001',
])
args = parser_with_config.parse_args(
[
"serve",
"--tensor-parallel-size=2",
"--model=my-model",
"--trust-remote-code",
"--port=8001",
]
)
assert args.model is None
assert args.tensor_parallel_size == 2
assert args.trust_remote_code is True
assert args.port == 8001
# Test other config values are preserved
args = parser_with_config.parse_args([
'serve',
'cli-model',
'--config',
cli_config_file_with_model,
])
args = parser_with_config.parse_args(
[
"serve",
"cli-model",
"--config",
cli_config_file_with_model,
]
)
assert args.tensor_parallel_size == 2
assert args.trust_remote_code is True
assert args.port == 12312
@pytest.mark.parametrize("input", [(), ("abc", ), (None, ),
(None, bool, [1, 2, 3])])
@pytest.mark.parametrize("input", [(), ("abc",), (None,), (None, bool, [1, 2, 3])])
def test_sha256(input: tuple):
digest = sha256(input)
assert digest is not None
......@@ -887,7 +889,7 @@ def test_sha256(input: tuple):
assert digest == sha256(input)
# hashing different input, returns different value
assert digest != sha256(input + (1, ))
assert digest != sha256(input + (1,))
@pytest.mark.parametrize(
......@@ -897,7 +899,8 @@ def test_sha256(input: tuple):
("tcp://127.0.0.1:5555", ("tcp", "127.0.0.1", "5555")),
("tcp://[::1]:5555", ("tcp", "::1", "5555")), # IPv6 address
("inproc://some_identifier", ("inproc", "some_identifier", "")),
])
],
)
def test_split_zmq_path(path, expected):
assert split_zmq_path(path) == expected
......@@ -909,7 +912,8 @@ def test_split_zmq_path(path, expected):
"tcp://127.0.0.1", # Missing port
"tcp://[::1]", # Missing port for IPv6
"tcp://:5555", # Missing host
])
],
)
def test_split_zmq_path_invalid(invalid_path):
with pytest.raises(ValueError):
split_zmq_path(invalid_path)
......@@ -931,8 +935,9 @@ def test_make_zmq_socket_ipv6():
zsock: zmq.Socket = make_zmq_socket(ctx, ipv6_path, socket_type)
# Verify that the IPV6 option is set
assert zsock.getsockopt(
zmq.IPV6) == 1, "IPV6 option should be enabled for IPv6 addresses"
assert zsock.getsockopt(zmq.IPV6) == 1, (
"IPV6 option should be enabled for IPv6 addresses"
)
# Clean up
zsock.close()
......@@ -1019,15 +1024,14 @@ def test_convert_ids_list_to_tokens():
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-1.5B-Instruct")
token_ids = tokenizer.encode("Hello, world!")
# token_ids = [9707, 11, 1879, 0]
assert tokenizer.convert_ids_to_tokens(token_ids) == [
'Hello', ',', 'Ġworld', '!'
]
assert tokenizer.convert_ids_to_tokens(token_ids) == ["Hello", ",", "Ġworld", "!"]
tokens = convert_ids_list_to_tokens(tokenizer, token_ids)
assert tokens == ['Hello', ',', ' world', '!']
assert tokens == ["Hello", ",", " world", "!"]
def test_current_stream_multithread():
import threading
if not torch.cuda.is_available():
pytest.skip("CUDA not available")
......@@ -1046,13 +1050,18 @@ def test_current_stream_multithread():
child_thread.start()
try:
assert thread_stream_ready.wait(
timeout=5), "Child thread failed to enter stream context in time"
assert thread_stream_ready.wait(timeout=5), (
"Child thread failed to enter stream context in time"
)
main_current_stream = current_stream()
assert main_current_stream != child_stream, "Main thread's current_stream was contaminated by child thread"
assert main_current_stream == main_default_stream, "Main thread's current_stream is not the default stream"
assert main_current_stream != child_stream, (
"Main thread's current_stream was contaminated by child thread"
)
assert main_current_stream == main_default_stream, (
"Main thread's current_stream is not the default stream"
)
# Notify child thread it can exit
thread_can_exit.set()
......@@ -1070,7 +1079,7 @@ def test_load_config_file(tmp_path):
"enable-logging": True,
"list-arg": ["item1", "item2"],
"port": 12323,
"tensor-parallel-size": 4
"tensor-parallel-size": 4,
}
# Write the configuration data to a temporary YAML file
......
tests/v1/core/test_kv_cache_utils.py
View file @ 4e256cad
......@@ -16,9 +16,6 @@ from vllm.multimodal.inputs import (
from vllm.sampling_params import SamplingParams
from vllm.utils import GiB_bytes, sha256, sha256_cbor
from vllm.v1.core.kv_cache_manager import KVCacheManager
# disable yapf here as it formats differently than isort such that both fail
# yapf: disable
from vllm.v1.core.kv_cache_utils import (
BlockHash,
FreeKVCacheBlockQueue,
......@@ -48,8 +45,6 @@ from vllm.v1.kv_cache_interface import (
from vllm.v1.metrics.stats import PrefixCacheStats
from vllm.v1.request import Request
# yapf: enable
pytestmark = pytest.mark.cpu_test
......
tests/v1/logits_processors/test_correctness.py
View file @ 4e256cad
......@@ -22,8 +22,6 @@ from vllm.config import VllmConfig
from vllm.platforms import current_platform
from vllm.sampling_params import SamplingParams
from vllm.utils import is_pin_memory_available
# yapf: disable
from vllm.v1.sample.logits_processor import (
BatchUpdate,
BatchUpdateBuilder,
......@@ -34,8 +32,6 @@ from vllm.v1.sample.logits_processor import (
MoveDirectionality,
build_logitsprocs,
)
# yapf: enable
from vllm.v1.sample.metadata import SamplingMetadata
PIN_MEMORY_AVAILABLE = is_pin_memory_available()
......
tests/v1/logits_processors/test_custom_offline.py
View file @ 4e256cad
......@@ -7,8 +7,6 @@ from typing import Union
import pytest
from tests.utils import create_new_process_for_each_test
# yapf: disable
from tests.v1.logits_processors.utils import (
DUMMY_LOGITPROC_ARG,
DUMMY_LOGITPROC_FQCN,
......@@ -24,8 +22,6 @@ from tests.v1.logits_processors.utils import (
prompts,
)
from tests.v1.logits_processors.utils import entry_points as fake_entry_points
# yapf: enable
from vllm import LLM, SamplingParams
from vllm.v1.sample.logits_processor import (
STR_POOLING_REJECTS_LOGITSPROCS,
......
tests/v1/logits_processors/test_custom_online.py
View file @ 4e256cad
......@@ -11,8 +11,6 @@ import pytest
import pytest_asyncio
from tests.utils import RemoteOpenAIServerCustom, create_new_process_for_each_test
# yapf: disable
from tests.v1.logits_processors.utils import (
DUMMY_LOGITPROC_ARG,
DUMMY_LOGITPROC_FQCN,
......@@ -25,8 +23,6 @@ from tests.v1.logits_processors.utils import (
)
from tests.v1.logits_processors.utils import entry_points as fake_entry_points
# yapf: enable
def _server_with_logitproc_entrypoint(
env_dict: Optional[dict[str, str]],
......
vllm/distributed/kv_transfer/kv_connector/factory.py
View file @ 4e256cad
......@@ -4,7 +4,6 @@
import importlib
from typing import TYPE_CHECKING, Callable
# yapf: disable
import vllm.envs as envs
from vllm.distributed.kv_transfer.kv_connector.base import (
KVConnectorBase,
......@@ -13,8 +12,6 @@ from vllm.distributed.kv_transfer.kv_connector.base import (
from vllm.distributed.kv_transfer.kv_connector.v1 import KVConnectorRole
from vllm.logger import init_logger
# yapf: enable
if TYPE_CHECKING:
from vllm.config import VllmConfig
from vllm.config.kv_transfer import KVTransferConfig
......
vllm/engine/arg_utils.py
View file @ 4e256cad
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
# yapf: disable
import argparse
import copy
import dataclasses
......@@ -88,8 +87,6 @@ from vllm.transformers_utils.utils import check_gguf_file
from vllm.utils import FlexibleArgumentParser, GiB_bytes, get_ip, is_in_ray_actor
from vllm.v1.sample.logits_processor import LogitsProcessor
# yapf: enable
if TYPE_CHECKING:
from vllm.executor.executor_base import ExecutorBase
from vllm.model_executor.layers.quantization import QuantizationMethods
......
vllm/entrypoints/chat_utils.py
View file @ 4e256cad
......@@ -17,9 +17,6 @@ import jinja2.nodes
import jinja2.parser
import jinja2.sandbox
import transformers.utils.chat_template_utils as hf_chat_utils
# yapf conflicts with isort for this block
# yapf: disable
from openai.types.chat import (
ChatCompletionAssistantMessageParam,
ChatCompletionContentPartImageParam,
......@@ -40,8 +37,6 @@ from openai.types.responses import ResponseInputImageParam
from openai_harmony import Message as OpenAIHarmonyMessage
from PIL import Image
from pydantic import BaseModel, ConfigDict, TypeAdapter
# yapf: enable
from transformers import PreTrainedTokenizer, PreTrainedTokenizerFast, ProcessorMixin
# pydantic needs the TypedDict from typing_extensions
......@@ -52,11 +47,7 @@ from vllm.logger import init_logger
from vllm.model_executor.models import SupportsMultiModal
from vllm.multimodal import MULTIMODAL_REGISTRY, MultiModalDataDict, MultiModalUUIDDict
from vllm.multimodal.utils import MediaConnector
# yapf: disable
from vllm.transformers_utils.chat_templates import get_chat_template_fallback_path
# yapf: enable
from vllm.transformers_utils.processor import cached_get_processor
from vllm.transformers_utils.tokenizer import AnyTokenizer, MistralTokenizer
from vllm.utils import random_uuid, supports_kw
......@@ -317,11 +308,7 @@ def _is_var_or_elems_access(
):
return _is_var_or_elems_access(node.node, varname, key)
# yapf: disable
return (
_is_attr_access(node, varname, key) if key
else _is_var_access(node, varname)
) # yapf: enable
return _is_attr_access(node, varname, key) if key else _is_var_access(node, varname)
def _iter_nodes_assign_var_or_elems(root: jinja2.nodes.Node, varname: str):
......
vllm/entrypoints/llm.py
View file @ 4e256cad
......@@ -39,9 +39,6 @@ from vllm.entrypoints.chat_utils import (
parse_chat_messages,
resolve_chat_template_content_format,
)
# yapf conflicts with isort for this block
# yapf: disable
from vllm.entrypoints.score_utils import (
ScoreContentPartParam,
ScoreMultiModalParam,
......@@ -50,8 +47,6 @@ from vllm.entrypoints.score_utils import (
compress_token_type_ids,
get_score_prompt,
)
# yapf: enable
from vllm.entrypoints.utils import _validate_truncation_size, log_non_default_args
from vllm.inputs import (
DataPrompt,
......
vllm/entrypoints/openai/api_server.py
View file @ 4e256cad
......@@ -49,9 +49,6 @@ from vllm.entrypoints.chat_utils import (
from vllm.entrypoints.launcher import serve_http
from vllm.entrypoints.logger import RequestLogger
from vllm.entrypoints.openai.cli_args import make_arg_parser, validate_parsed_serve_args
# yapf conflicts with isort for this block
# yapf: disable
from vllm.entrypoints.openai.protocol import (
ChatCompletionRequest,
ChatCompletionResponse,
......@@ -84,8 +81,6 @@ from vllm.entrypoints.openai.protocol import (
TranslationResponse,
UnloadLoRAAdapterRequest,
)
# yapf: enable
from vllm.entrypoints.openai.serving_chat import OpenAIServingChat
from vllm.entrypoints.openai.serving_classification import ServingClassification
from vllm.entrypoints.openai.serving_completion import OpenAIServingCompletion
......
vllm/entrypoints/openai/protocol.py
View file @ 4e256cad
......@@ -11,8 +11,6 @@ from typing import Annotated, Any, ClassVar, Generic, Literal, Optional, TypeVar
import regex as re
import torch
from fastapi import HTTPException, UploadFile
# yapf: disable
from openai.types.chat.chat_completion_audio import (
ChatCompletionAudio as OpenAIChatCompletionAudio,
)
......@@ -46,8 +44,6 @@ from openai.types.responses import ResponseCreatedEvent as OpenAIResponseCreated
from openai.types.responses import (
ResponseInProgressEvent as OpenAIResponseInProgressEvent,
)
# yapf: enable
from openai.types.responses.response_reasoning_item import (
Content as ResponseReasoningTextContent,
)
......
vllm/entrypoints/openai/run_batch.py
View file @ 4e256cad
......@@ -18,8 +18,6 @@ from vllm.config import VllmConfig
from vllm.engine.arg_utils import AsyncEngineArgs, optional_type
from vllm.engine.protocol import EngineClient
from vllm.entrypoints.logger import RequestLogger
# yapf: disable
from vllm.entrypoints.openai.protocol import (
BatchRequestInput,
BatchRequestOutput,
......@@ -30,8 +28,6 @@ from vllm.entrypoints.openai.protocol import (
RerankResponse,
ScoreResponse,
)
# yapf: enable
from vllm.entrypoints.openai.serving_chat import OpenAIServingChat
from vllm.entrypoints.openai.serving_embedding import OpenAIServingEmbedding
from vllm.entrypoints.openai.serving_models import BaseModelPath, OpenAIServingModels
......
vllm/entrypoints/openai/serving_chat.py
View file @ 4e256cad
......@@ -1733,13 +1733,15 @@ class OpenAIServingChat(OpenAIServing):
is a tool call with arguments.
"""
# yapf: disable
return bool(
# if there is a delta message that includes tool calls which
# include a function that has arguments
output.finish_reason is not None
and self.enable_auto_tools and self.tool_parser and delta_message
and delta_message.tool_calls and delta_message.tool_calls[0]
and self.enable_auto_tools
and self.tool_parser
and delta_message
and delta_message.tool_calls
and delta_message.tool_calls[0]
and delta_message.tool_calls[0].function
and delta_message.tool_calls[0].function.arguments is not None
)
......
vllm/entrypoints/openai/serving_classification.py
View file @ 4e256cad
......@@ -18,8 +18,6 @@ from vllm.entrypoints.openai.protocol import (
ErrorResponse,
UsageInfo,
)
# yapf: enable
from vllm.entrypoints.openai.serving_engine import (
ClassificationServeContext,
OpenAIServing,
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment