Deprecated 1st generation Qwen and QwenVL models (#45131)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2026-06-10 14:55:33 +00:00
committed by GitHub
parent ccc05de038
commit 3cc9fecd58
27 changed files with 6 additions and 1349 deletions
@@ -192,7 +192,6 @@ MULTIMODAL_MODELS = {
"AIDC-AI/Ovis2.5-2B": PPTestSettings.fast(),
"microsoft/Phi-3.5-vision-instruct": PPTestSettings.fast(),
"mistralai/Pixtral-12B-2409": PPTestSettings.fast(load_format="dummy"),
"Qwen/Qwen-VL-Chat": PPTestSettings.fast(),
"Qwen/Qwen2-Audio-7B-Instruct": PPTestSettings.fast(),
"Qwen/Qwen2-VL-2B-Instruct": PPTestSettings.fast(),
"fixie-ai/ultravox-v0_5-llama-3_2-1b": PPTestSettings.fast(),
-10
View File
@@ -5,21 +5,11 @@
import os
import warnings
import pytest
import torch
from tests.utils import prewarm_hf_cache
from vllm.platforms import current_platform
@pytest.fixture(scope="session", autouse=True)
def _prewarm_hf_cache():
# tokenization_qwen.py downloads SimSun.ttf from
# qianwen-res.oss-cn-beijing.aliyuncs.com; both Qwen/Qwen-VL and
# Qwen/Qwen-VL-Chat look it up from the Chat repo.
prewarm_hf_cache([("Qwen/Qwen-VL-Chat", "SimSun.ttf")])
def pytest_configure(config):
"""Early ROCm configuration that must happen before test collection."""
if not current_platform.is_rocm():
@@ -976,16 +976,6 @@ VLM_TEST_SETTINGS = {
auto_cls=AutoModelForImageTextToText,
hf_model_kwargs=model_utils.qianfan_ocr_hf_model_kwargs("baidu/Qianfan-OCR"),
),
"qwen_vl": VLMTestInfo(
models=["Qwen/Qwen-VL"],
test_type=(VLMTestType.IMAGE, VLMTestType.MULTI_IMAGE),
prompt_formatter=identity,
img_idx_to_prompt=lambda idx: f"Picture {idx}: <img></img>\n",
max_model_len=1024,
max_num_seqs=2,
vllm_output_post_proc=model_utils.qwen_vllm_to_hf_output,
prompt_path_encoder=model_utils.qwen_prompt_path_encoder,
),
"qwen2_vl": VLMTestInfo(
models=["Qwen/Qwen2-VL-2B-Instruct"],
test_type=(VLMTestType.IMAGE, VLMTestType.MULTI_IMAGE, VLMTestType.VIDEO),
-18
View File
@@ -505,14 +505,6 @@ _TEXT_GENERATION_EXAMPLE_MODELS = {
"pfnet/plamo-3-nict-2b-base",
trust_remote_code=True,
),
"QWenLMHeadModel": _HfExamplesInfo(
"Qwen/Qwen-7B-Chat",
max_transformers_version="4.53",
transformers_version_reason={
"hf": "HF model uses remote code that is not compatible with latest Transformers" # noqa: E501
},
trust_remote_code=True,
),
"Qwen2ForCausalLM": _HfExamplesInfo(
"Qwen/Qwen2-0.5B-Instruct",
extras={
@@ -1297,16 +1289,6 @@ _MULTIMODAL_EXAMPLE_MODELS = {
"baidu/Qianfan-OCR",
min_transformers_version="5.6.0",
),
"QwenVLForConditionalGeneration": _HfExamplesInfo(
"Qwen/Qwen-VL",
extras={"chat": "Qwen/Qwen-VL-Chat"},
trust_remote_code=True,
max_transformers_version="4.53.3",
transformers_version_reason={
"hf": "HF model uses deprecated imports which have been removed."
}, # noqa: E501
hf_overrides={"architectures": ["QwenVLForConditionalGeneration"]},
),
"Qwen2AudioForConditionalGeneration": _HfExamplesInfo(
"Qwen/Qwen2-Audio-7B-Instruct"
),
-14
View File
@@ -1,14 +0,0 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import pytest
from tests.utils import prewarm_hf_cache
@pytest.fixture(scope="session", autouse=True)
def _prewarm_hf_cache():
# tokenization_qwen.py downloads SimSun.ttf from
# qianwen-res.oss-cn-beijing.aliyuncs.com; both Qwen/Qwen-VL and
# Qwen/Qwen-VL-Chat look it up from the Chat repo.
prewarm_hf_cache([("Qwen/Qwen-VL-Chat", "SimSun.ttf")])
-2
View File
@@ -428,8 +428,6 @@ def test_resolve_content_format_hf_defined(model, expected_format):
("deepseek-ai/deepseek-vl2-tiny", "string"),
("adept/fuyu-8b", "string"),
("google/paligemma-3b-mix-224", "string"),
("Qwen/Qwen-VL", "string"),
("Qwen/Qwen-VL-Chat", "string"),
],
)
def test_resolve_content_format_fallbacks(model, expected_format):
-14
View File
@@ -1,14 +0,0 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import pytest
from tests.utils import prewarm_hf_cache
@pytest.fixture(scope="session", autouse=True)
def _prewarm_hf_cache():
# tokenization_qwen.py downloads SimSun.ttf from
# qianwen-res.oss-cn-beijing.aliyuncs.com; both Qwen/Qwen-VL and
# Qwen/Qwen-VL-Chat look it up from the Chat repo.
prewarm_hf_cache([("Qwen/Qwen-VL-Chat", "SimSun.ttf")])
-8
View File
@@ -47,14 +47,6 @@ def test_tokenizer_like_protocol():
assert "DSV32" in tokenizer.__class__.__name__
_assert_tokenizer_like(tokenizer)
tokenizer = get_tokenizer(
"Qwen/Qwen-VL",
tokenizer_mode="qwen_vl",
trust_remote_code=True,
)
assert isinstance(tokenizer, HfTokenizer)
assert "WithoutImagePad" in tokenizer.__class__.__name__
@pytest.mark.parametrize("tokenizer_name", ["facebook/opt-125m", "gpt2"])
def test_tokenizer_revision(tokenizer_name: str):