xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 06:35:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Avoid use of deprecated AutoModelForVision2Seq (#25065)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung 2025-09-17 20:19:15 +08:00 committed by GitHub
parent 2b85697031
commit e120533d7a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 7 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
tests/models/multimodal/generation/test_common.py
View File

@ -10,7 +10,7 @@ from pathlib import PosixPath
import pytest import pytest
from transformers import (AutoModel, AutoModelForImageTextToText, from transformers import (AutoModel, AutoModelForImageTextToText,
AutoModelForTextToWaveform, AutoModelForVision2Seq) AutoModelForTextToWaveform)
from vllm.platforms import current_platform from vllm.platforms import current_platform
from vllm.utils import identity from vllm.utils import identity
@ -137,7 +137,7 @@ VLM_TEST_SETTINGS = {
video_idx_to_prompt=lambda idx: "<|vision_start|><|video_pad|><|vision_end|>", # noqa: E501 video_idx_to_prompt=lambda idx: "<|vision_start|><|video_pad|><|vision_end|>", # noqa: E501
max_model_len=4096, max_model_len=4096,
max_num_seqs=2, max_num_seqs=2,
auto_cls=AutoModelForVision2Seq, auto_cls=AutoModelForImageTextToText,
vllm_output_post_proc=model_utils.qwen2_vllm_to_hf_output, vllm_output_post_proc=model_utils.qwen2_vllm_to_hf_output,
image_size_factors=[(), (0.25,), (0.25, 0.25, 0.25), (0.25, 0.2, 0.15)], image_size_factors=[(), (0.25,), (0.25, 0.25, 0.25), (0.25, 0.2, 0.15)],
marks=[pytest.mark.core_model, pytest.mark.cpu_model], marks=[pytest.mark.core_model, pytest.mark.cpu_model],
@ -502,7 +502,7 @@ VLM_TEST_SETTINGS = {
num_video_frames=16, num_video_frames=16,
max_model_len=16384, max_model_len=16384,
hf_model_kwargs=model_utils.llava_onevision_hf_model_kwargs("llava-hf/llava-onevision-qwen2-0.5b-ov-hf"), # noqa: E501 hf_model_kwargs=model_utils.llava_onevision_hf_model_kwargs("llava-hf/llava-onevision-qwen2-0.5b-ov-hf"), # noqa: E501
auto_cls=AutoModelForVision2Seq, auto_cls=AutoModelForImageTextToText,
vllm_output_post_proc=model_utils.llava_onevision_vllm_to_hf_output, vllm_output_post_proc=model_utils.llava_onevision_vllm_to_hf_output,
custom_test_opts=[CustomTestOptions( custom_test_opts=[CustomTestOptions(
inputs=custom_inputs.multi_video_multi_aspect_ratio_inputs( inputs=custom_inputs.multi_video_multi_aspect_ratio_inputs(
@ -518,7 +518,7 @@ VLM_TEST_SETTINGS = {
num_video_frames=16, num_video_frames=16,
max_model_len=4096, max_model_len=4096,
max_num_seqs=2, max_num_seqs=2,
auto_cls=AutoModelForVision2Seq, auto_cls=AutoModelForImageTextToText,
vllm_output_post_proc=model_utils.llava_video_vllm_to_hf_output, vllm_output_post_proc=model_utils.llava_video_vllm_to_hf_output,
), ),
"mantis": VLMTestInfo( "mantis": VLMTestInfo(
@ -680,7 +680,7 @@ VLM_TEST_SETTINGS = {
multi_image_prompt="Picture 1: <vlm_image>\nPicture 2: <vlm_image>\nDescribe these two images with one paragraph respectively.", # noqa: E501 multi_image_prompt="Picture 1: <vlm_image>\nPicture 2: <vlm_image>\nDescribe these two images with one paragraph respectively.", # noqa: E501
max_model_len=4096, max_model_len=4096,
max_num_seqs=2, max_num_seqs=2,
auto_cls=AutoModelForVision2Seq, auto_cls=AutoModelForImageTextToText,
vllm_output_post_proc=model_utils.qwen2_vllm_to_hf_output, vllm_output_post_proc=model_utils.qwen2_vllm_to_hf_output,
image_size_factors=[(), (0.25,), (0.25, 0.25, 0.25), (0.25, 0.2, 0.15)], image_size_factors=[(), (0.25,), (0.25, 0.25, 0.25), (0.25, 0.2, 0.15)],
marks=[pytest.mark.cpu_model], marks=[pytest.mark.cpu_model],
@ -784,7 +784,7 @@ VLM_TEST_SETTINGS = {
test_type=VLMTestType.CUSTOM_INPUTS, test_type=VLMTestType.CUSTOM_INPUTS,
max_model_len=16384, max_model_len=16384,
max_num_seqs=2, max_num_seqs=2,
auto_cls=AutoModelForVision2Seq, auto_cls=AutoModelForImageTextToText,
hf_model_kwargs=model_utils.llava_onevision_hf_model_kwargs("llava-hf/llava-onevision-qwen2-0.5b-ov-hf"), # noqa: E501 hf_model_kwargs=model_utils.llava_onevision_hf_model_kwargs("llava-hf/llava-onevision-qwen2-0.5b-ov-hf"), # noqa: E501
vllm_output_post_proc=model_utils.llava_onevision_vllm_to_hf_output, vllm_output_post_proc=model_utils.llava_onevision_vllm_to_hf_output,
custom_test_opts=[CustomTestOptions( custom_test_opts=[CustomTestOptions(
@ -800,7 +800,7 @@ VLM_TEST_SETTINGS = {
test_type=VLMTestType.CUSTOM_INPUTS, test_type=VLMTestType.CUSTOM_INPUTS,
max_model_len=4096, max_model_len=4096,
max_num_seqs=2, max_num_seqs=2,
auto_cls=AutoModelForVision2Seq, auto_cls=AutoModelForImageTextToText,
vllm_output_post_proc=model_utils.qwen2_vllm_to_hf_output, vllm_output_post_proc=model_utils.qwen2_vllm_to_hf_output,
custom_test_opts=[CustomTestOptions( custom_test_opts=[CustomTestOptions(
inputs=custom_inputs.windows_attention_image_qwen2_5_vl(), inputs=custom_inputs.windows_attention_image_qwen2_5_vl(),