xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-07 17:42:21 +08:00
Code Issues Packages Projects Releases Wiki Activity

[CI/Build][VLM] Cleanup multiple images inputs model test (#7897)

Browse Source
This commit is contained in:
Isotr0py 2024-08-27 23:28:30 +08:00 committed by GitHub
parent 6fc4e6e07a
commit 9db642138b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 74 additions and 203 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

136
tests/models/test_minicpmv.py
View File

@ -1,14 +1,15 @@
from typing import List, Optional, Tuple, Type from typing import List, Optional, Tuple, Type, Union
import pytest import pytest
import torch import torch
import torch.types import torch.types
from PIL import Image
from transformers import BatchEncoding from transformers import BatchEncoding
from vllm.multimodal.utils import rescale_image_size from vllm.multimodal.utils import rescale_image_size
from vllm.sequence import SampleLogprobs from vllm.sequence import SampleLogprobs
from ..conftest import IMAGE_ASSETS, HfRunner, VllmRunner, _ImageAssets from ..conftest import IMAGE_ASSETS, HfRunner, VllmRunner
from .utils import check_logprobs_close from .utils import check_logprobs_close
pytestmark = pytest.mark.vlm pytestmark = pytest.mark.vlm
@ -24,6 +25,11 @@ HF_IMAGE_PROMPTS = IMAGE_ASSETS.prompts({
"(<image>./</image>)\nWhat is the season?<|eot_id|>" \ "(<image>./</image>)\nWhat is the season?<|eot_id|>" \
"<|start_header_id|>assistant<|end_header_id|>\n\n", "<|start_header_id|>assistant<|end_header_id|>\n\n",
}) })
HF_MULTIIMAGE_IMAGE_PROMPT = \
"<|begin_of_text|><|start_header_id|>user<|end_header_id|>\n\n" \
"(<image>./</image>)\n(<image>./</image>)\n" \
"Describe these images.<|eot_id|>" \
"<|start_header_id|>assistant<|end_header_id|>\n\n"
models = ["openbmb/MiniCPM-Llama3-V-2_5"] models = ["openbmb/MiniCPM-Llama3-V-2_5"]
@ -46,13 +52,14 @@ target_dtype = "half"
def run_test( def run_test(
hf_runner: Type[HfRunner], hf_runner: Type[HfRunner],
vllm_runner: Type[VllmRunner], vllm_runner: Type[VllmRunner],
image_assets: _ImageAssets, inputs: List[Tuple[List[str], Union[List[Image.Image],
List[List[Image.Image]]]]],
model: str, model: str,
*, *,
size_factors: List[float],
dtype: str, dtype: str,
max_tokens: int, max_tokens: int,
num_logprobs: int, num_logprobs: int,
mm_limit: int,
tensor_parallel_size: int, tensor_parallel_size: int,
distributed_executor_backend: Optional[str] = None, distributed_executor_backend: Optional[str] = None,
): ):
@ -65,12 +72,6 @@ def run_test(
Note, the text input is also adjusted to abide by vllm contract. Note, the text input is also adjusted to abide by vllm contract.
The text output is sanitized to be able to compare with hf. The text output is sanitized to be able to compare with hf.
""" """
images = [asset.pil_image for asset in image_assets]
inputs_per_image = [(
[prompt for _ in size_factors],
[rescale_image_size(image, factor) for factor in size_factors],
) for image, prompt in zip(images, HF_IMAGE_PROMPTS)]
# NOTE: take care of the order. run vLLM first, and then run HF. # NOTE: take care of the order. run vLLM first, and then run HF.
# vLLM needs a fresh new process without cuda initialization. # vLLM needs a fresh new process without cuda initialization.
@ -82,6 +83,7 @@ def run_test(
max_model_len=4096, max_model_len=4096,
max_num_seqs=1, max_num_seqs=1,
dtype=dtype, dtype=dtype,
limit_mm_per_prompt={"image": mm_limit},
tensor_parallel_size=tensor_parallel_size, tensor_parallel_size=tensor_parallel_size,
distributed_executor_backend=distributed_executor_backend, distributed_executor_backend=distributed_executor_backend,
enforce_eager=True) as vllm_model: enforce_eager=True) as vllm_model:
@ -93,7 +95,7 @@ def run_test(
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
images=images, images=images,
stop_token_ids=stop_token_ids) stop_token_ids=stop_token_ids)
for prompts, images in inputs_per_image for prompts, images in inputs
] ]
hf_model = hf_runner(model, dtype=dtype, postprocess_inputs=_wrap_inputs) hf_model = hf_runner(model, dtype=dtype, postprocess_inputs=_wrap_inputs)
@ -104,7 +106,7 @@ def run_test(
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
images=images, images=images,
tokenizer=tokenizer) tokenizer=tokenizer)
for prompts, images in inputs_per_image for prompts, images in inputs
] ]
for hf_outputs, vllm_outputs in zip(hf_outputs_per_image, for hf_outputs, vllm_outputs in zip(hf_outputs_per_image,
@ -138,104 +140,26 @@ def run_test(
@pytest.mark.parametrize("num_logprobs", [5]) @pytest.mark.parametrize("num_logprobs", [5])
def test_models(hf_runner, vllm_runner, image_assets, model, size_factors, def test_models(hf_runner, vllm_runner, image_assets, model, size_factors,
dtype: str, max_tokens: int, num_logprobs: int) -> None: dtype: str, max_tokens: int, num_logprobs: int) -> None:
images = [asset.pil_image for asset in image_assets]
inputs_per_image = [(
[prompt for _ in size_factors],
[rescale_image_size(image, factor) for factor in size_factors],
) for image, prompt in zip(images, HF_IMAGE_PROMPTS)]
run_test( run_test(
hf_runner, hf_runner,
vllm_runner, vllm_runner,
image_assets, inputs_per_image,
model, model,
size_factors=size_factors,
dtype=dtype, dtype=dtype,
max_tokens=max_tokens, max_tokens=max_tokens,
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
mm_limit=1,
tensor_parallel_size=1, tensor_parallel_size=1,
) )
HF_MULTIIMAGE_IMAGE_PROMPT = \
"<|begin_of_text|><|start_header_id|>user<|end_header_id|>\n\n" \
"(<image>./</image>)\n(<image>./</image>)\n" \
"Describe these images.<|eot_id|>" \
"<|start_header_id|>assistant<|end_header_id|>\n\n"
def run_multi_image_test(
hf_runner: Type[HfRunner],
vllm_runner: Type[VllmRunner],
image_assets: _ImageAssets,
model: str,
*,
size_factors: List[float],
dtype: str,
max_tokens: int,
num_logprobs: int,
tensor_parallel_size: int,
distributed_executor_backend: Optional[str] = None,
):
"""Inference result should be the same between hf and vllm.
All the image fixtures for the test is under tests/images.
For huggingface runner, we provide the PIL images as input.
For vllm runner, we provide MultiModalDataDict objects
and corresponding MultiModalConfig as input.
Note, the text input is also adjusted to abide by vllm contract.
The text output is sanitized to be able to compare with hf.
"""
images = [asset.pil_image for asset in image_assets]
inputs_per_case = [
([HF_MULTIIMAGE_IMAGE_PROMPT for _ in size_factors],
[[rescale_image_size(image, factor) for image in images]
for factor in size_factors])
]
# NOTE: take care of the order. run vLLM first, and then run HF.
# vLLM needs a fresh new process without cuda initialization.
# if we run HF first, the cuda initialization will be done and it
# will hurt multiprocessing backend with fork method (the default method).
# max_model_len should be greater than image_feature_size
with vllm_runner(model,
max_model_len=4096,
max_num_seqs=1,
limit_mm_per_prompt={"image": len(images)},
dtype=dtype,
tensor_parallel_size=tensor_parallel_size,
distributed_executor_backend=distributed_executor_backend,
enforce_eager=True) as vllm_model:
tokenizer = vllm_model.model.get_tokenizer()
stop_token_ids = [tokenizer.eos_id, tokenizer.eot_id]
vllm_outputs_per_case = [
vllm_model.generate_greedy_logprobs(prompts,
max_tokens,
num_logprobs=num_logprobs,
images=images,
stop_token_ids=stop_token_ids)
for prompts, images in inputs_per_case
]
hf_model = hf_runner(model, dtype=dtype, postprocess_inputs=_wrap_inputs)
with hf_model, torch.no_grad():
hf_outputs_per_case = [
hf_model.generate_greedy_logprobs_limit(prompts,
max_tokens,
num_logprobs=num_logprobs,
images=images,
tokenizer=tokenizer)
for prompts, images in inputs_per_case
]
for hf_outputs, vllm_outputs in zip(hf_outputs_per_case,
vllm_outputs_per_case):
check_logprobs_close(
outputs_0_lst=[
trunc_hf_output(hf_output) for hf_output in hf_outputs
],
outputs_1_lst=vllm_outputs,
name_0="hf",
name_1="vllm",
)
@pytest.mark.parametrize("model", models) @pytest.mark.parametrize("model", models)
@pytest.mark.parametrize( @pytest.mark.parametrize(
"size_factors", "size_factors",
@ -256,14 +180,22 @@ def run_multi_image_test(
def test_multi_images_models(hf_runner, vllm_runner, image_assets, model, def test_multi_images_models(hf_runner, vllm_runner, image_assets, model,
size_factors, dtype: str, max_tokens: int, size_factors, dtype: str, max_tokens: int,
num_logprobs: int) -> None: num_logprobs: int) -> None:
run_multi_image_test( images = [asset.pil_image for asset in image_assets]
inputs_per_case = [
([HF_MULTIIMAGE_IMAGE_PROMPT for _ in size_factors],
[[rescale_image_size(image, factor) for image in images]
for factor in size_factors])
]
run_test(
hf_runner, hf_runner,
vllm_runner, vllm_runner,
image_assets, inputs_per_case,
model, model,
size_factors=size_factors,
dtype=dtype, dtype=dtype,
max_tokens=max_tokens, max_tokens=max_tokens,
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
mm_limit=2,
tensor_parallel_size=1, tensor_parallel_size=1,
) )

141
tests/models/test_phi3v.py
View File

@ -1,6 +1,6 @@
import os import os
import re import re
from typing import List, Optional, Tuple, Type from typing import List, Optional, Tuple, Type, Union
import pytest import pytest
from PIL import Image from PIL import Image
@ -60,13 +60,14 @@ if is_hip():
def run_test( def run_test(
hf_runner: Type[HfRunner], hf_runner: Type[HfRunner],
vllm_runner: Type[VllmRunner], vllm_runner: Type[VllmRunner],
images: List[Image.Image], inputs: List[Tuple[List[str], Union[List[Image.Image],
List[List[Image.Image]]]]],
model: str, model: str,
*, *,
size_factors: List[float],
dtype: str, dtype: str,
max_tokens: int, max_tokens: int,
num_logprobs: int, num_logprobs: int,
mm_limit: int,
tensor_parallel_size: int, tensor_parallel_size: int,
distributed_executor_backend: Optional[str] = None, distributed_executor_backend: Optional[str] = None,
): ):
@ -79,13 +80,6 @@ def run_test(
Note, the text input is also adjusted to abide by vllm contract. Note, the text input is also adjusted to abide by vllm contract.
The text output is sanitized to be able to compare with hf. The text output is sanitized to be able to compare with hf.
""" """
inputs_per_image = [(
[prompt for _ in size_factors],
[
rescale_image_size(image, factor, transpose=idx)
for idx, factor in enumerate(size_factors)
],
) for image, prompt in zip(images, HF_IMAGE_PROMPTS)]
# NOTE: take care of the order. run vLLM first, and then run HF. # NOTE: take care of the order. run vLLM first, and then run HF.
# vLLM needs a fresh new process without cuda initialization. # vLLM needs a fresh new process without cuda initialization.
@ -97,15 +91,16 @@ def run_test(
max_model_len=4096, max_model_len=4096,
max_num_seqs=1, max_num_seqs=1,
dtype=dtype, dtype=dtype,
limit_mm_per_prompt={"image": mm_limit},
tensor_parallel_size=tensor_parallel_size, tensor_parallel_size=tensor_parallel_size,
distributed_executor_backend=distributed_executor_backend, distributed_executor_backend=distributed_executor_backend,
enforce_eager=True) as vllm_model: enforce_eager=True) as vllm_model:
vllm_outputs_per_image = [ vllm_outputs_per_case = [
vllm_model.generate_greedy_logprobs(prompts, vllm_model.generate_greedy_logprobs(prompts,
max_tokens, max_tokens,
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
images=images) images=images)
for prompts, images in inputs_per_image for prompts, images in inputs
] ]
# use eager mode for hf runner, since phi3_v didn't work with flash_attn # use eager mode for hf runner, since phi3_v didn't work with flash_attn
@ -113,17 +108,17 @@ def run_test(
with hf_runner(model, dtype=dtype, with hf_runner(model, dtype=dtype,
model_kwargs=hf_model_kwargs) as hf_model: model_kwargs=hf_model_kwargs) as hf_model:
eos_token_id = hf_model.processor.tokenizer.eos_token_id eos_token_id = hf_model.processor.tokenizer.eos_token_id
hf_outputs_per_image = [ hf_outputs_per_case = [
hf_model.generate_greedy_logprobs_limit(prompts, hf_model.generate_greedy_logprobs_limit(prompts,
max_tokens, max_tokens,
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
images=images, images=images,
eos_token_id=eos_token_id) eos_token_id=eos_token_id)
for prompts, images in inputs_per_image for prompts, images in inputs
] ]
for hf_outputs, vllm_outputs in zip(hf_outputs_per_image, for hf_outputs, vllm_outputs in zip(hf_outputs_per_case,
vllm_outputs_per_image): vllm_outputs_per_case):
check_logprobs_close( check_logprobs_close(
outputs_0_lst=hf_outputs, outputs_0_lst=hf_outputs,
outputs_1_lst=[ outputs_1_lst=[
@ -156,15 +151,22 @@ def run_test(
@pytest.mark.parametrize("num_logprobs", [10]) @pytest.mark.parametrize("num_logprobs", [10])
def test_models(hf_runner, vllm_runner, image_assets, model, size_factors, def test_models(hf_runner, vllm_runner, image_assets, model, size_factors,
dtype: str, max_tokens: int, num_logprobs: int) -> None: dtype: str, max_tokens: int, num_logprobs: int) -> None:
images = [asset.pil_image for asset in image_assets]
inputs_per_image = [(
[prompt for _ in size_factors],
[rescale_image_size(image, factor) for factor in size_factors],
) for image, prompt in zip(images, HF_IMAGE_PROMPTS)]
run_test( run_test(
hf_runner, hf_runner,
vllm_runner, vllm_runner,
[asset.pil_image for asset in image_assets], inputs_per_image,
model, model,
size_factors=size_factors,
dtype=dtype, dtype=dtype,
max_tokens=max_tokens, max_tokens=max_tokens,
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
mm_limit=1,
tensor_parallel_size=1, tensor_parallel_size=1,
) )
@ -173,97 +175,26 @@ def test_models(hf_runner, vllm_runner, image_assets, model, size_factors,
@pytest.mark.parametrize("dtype", [target_dtype]) @pytest.mark.parametrize("dtype", [target_dtype])
def test_regression_7840(hf_runner, vllm_runner, image_assets, model, def test_regression_7840(hf_runner, vllm_runner, image_assets, model,
dtype) -> None: dtype) -> None:
images = [asset.pil_image for asset in image_assets]
inputs_regresion_7840 = [
([prompt], [image]) for image, prompt in zip(images, HF_IMAGE_PROMPTS)
]
# Regression test for #7840. # Regression test for #7840.
run_test( run_test(
hf_runner, hf_runner,
vllm_runner, vllm_runner,
[image_assets[0].pil_image.resize((465, 226))], inputs_regresion_7840,
model, model,
size_factors=[1.0],
dtype=dtype, dtype=dtype,
max_tokens=128, max_tokens=128,
num_logprobs=10, num_logprobs=10,
mm_limit=1,
tensor_parallel_size=1, tensor_parallel_size=1,
) )
def run_multi_image_test(
hf_runner: Type[HfRunner],
vllm_runner: Type[VllmRunner],
images: List[Image.Image],
model: str,
*,
size_factors: List[float],
dtype: str,
max_tokens: int,
num_logprobs: int,
tensor_parallel_size: int,
distributed_executor_backend: Optional[str] = None,
):
"""Inference result should be the same between hf and vllm.
All the image fixtures for the test is under tests/images.
For huggingface runner, we provide the PIL images as input.
For vllm runner, we provide MultiModalDataDict objects
and corresponding MultiModalConfig as input.
Note, the text input is also adjusted to abide by vllm contract.
The text output is sanitized to be able to compare with hf.
"""
inputs_per_case = [
([HF_MULTIIMAGE_IMAGE_PROMPT for _ in size_factors],
[[rescale_image_size(image, factor) for image in images]
for factor in size_factors])
]
# NOTE: take care of the order. run vLLM first, and then run HF.
# vLLM needs a fresh new process without cuda initialization.
# if we run HF first, the cuda initialization will be done and it
# will hurt multiprocessing backend with fork method (the default method).
# max_model_len should be greater than image_feature_size
with vllm_runner(model,
max_model_len=4096,
max_num_seqs=1,
limit_mm_per_prompt={"image": len(images)},
dtype=dtype,
tensor_parallel_size=tensor_parallel_size,
distributed_executor_backend=distributed_executor_backend,
enforce_eager=True) as vllm_model:
vllm_outputs_per_case = [
vllm_model.generate_greedy_logprobs(prompts,
max_tokens,
num_logprobs=num_logprobs,
images=images)
for prompts, images in inputs_per_case
]
hf_model_kwargs = {"_attn_implementation": "eager"}
with hf_runner(model, dtype=dtype,
model_kwargs=hf_model_kwargs) as hf_model:
eos_token_id = hf_model.processor.tokenizer.eos_token_id
hf_outputs_per_case = [
hf_model.generate_greedy_logprobs_limit(prompts,
max_tokens,
num_logprobs=num_logprobs,
images=images,
eos_token_id=eos_token_id)
for prompts, images in inputs_per_case
]
for hf_outputs, vllm_outputs in zip(hf_outputs_per_case,
vllm_outputs_per_case):
check_logprobs_close(
outputs_0_lst=hf_outputs,
outputs_1_lst=[
vllm_to_hf_output(vllm_output, model)
for vllm_output in vllm_outputs
],
name_0="hf",
name_1="vllm",
)
@pytest.mark.parametrize("model", models) @pytest.mark.parametrize("model", models)
@pytest.mark.parametrize( @pytest.mark.parametrize(
"size_factors", "size_factors",
@ -280,18 +211,26 @@ def run_multi_image_test(
) )
@pytest.mark.parametrize("dtype", [target_dtype]) @pytest.mark.parametrize("dtype", [target_dtype])
@pytest.mark.parametrize("max_tokens", [128]) @pytest.mark.parametrize("max_tokens", [128])
@pytest.mark.parametrize("num_logprobs", [5]) @pytest.mark.parametrize("num_logprobs", [10])
def test_multi_images_models(hf_runner, vllm_runner, image_assets, model, def test_multi_images_models(hf_runner, vllm_runner, image_assets, model,
size_factors, dtype: str, max_tokens: int, size_factors, dtype: str, max_tokens: int,
num_logprobs: int) -> None: num_logprobs: int) -> None:
run_multi_image_test( images = [asset.pil_image for asset in image_assets]
inputs_per_case = [
([HF_MULTIIMAGE_IMAGE_PROMPT for _ in size_factors],
[[rescale_image_size(image, factor) for image in images]
for factor in size_factors])
]
run_test(
hf_runner, hf_runner,
vllm_runner, vllm_runner,
[asset.pil_image for asset in image_assets], inputs_per_case,
model, model,
size_factors=size_factors,
dtype=dtype, dtype=dtype,
max_tokens=max_tokens, max_tokens=max_tokens,
num_logprobs=num_logprobs, num_logprobs=num_logprobs,
mm_limit=2,
tensor_parallel_size=1, tensor_parallel_size=1,
) )