Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 10 additions & 19 deletions pyproject.toml
Original file line number Diff line number Diff line change
Expand Up @@ -51,34 +51,25 @@ line-length = 120
# Folder to be modified
exclude = [
# Batch (1)
"tests/e2e/__init__.py",
"tests/e2e/310p/",
"tests/e2e/conftest.py",
"tests/e2e/doctests/",
"tests/e2e/model_utils.py",
"tests/e2e/models/",
"tests/e2e/multicard/2-cards/",
# "tests/e2e/__init__.py",
# "tests/e2e/310p/",
# "tests/e2e/conftest.py",
# "tests/e2e/doctests/",
# "tests/e2e/model_utils.py",
# "tests/e2e/models/",
# "tests/e2e/multicard/2-cards/",

# Batch (2)
"tests/e2e/multicard/4-cards/",
"tests/e2e/nightly/multi_node/",

# Batch (3)
"tests/e2e/nightly/single_node/models/",

# Batch (4)
"tests/e2e/nightly/single_node/ops/",

# Batch (5)
# "tests/e2e/singlecard/",

# Batch (6)
"tests/e2e/nightly/single_node/ops/singlecard_ops/triton/",
"tests/e2e/singlecard/pooling/",
"tests/e2e/singlecard/spec_decode/",
"tests/e2e/utils.py",
"tests/e2e/vllm_interface/",
"tests/e2e/weekly/",

# Batch (3)
"tests/e2e/nightly/single_node/",

"tests/ut/",
]
Expand Down
17 changes: 6 additions & 11 deletions tests/e2e/310p/multicard/test_vl_model_multicard.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,28 +15,23 @@
# limitations under the License.
# This file is a part of the vllm-ascend project.

import sys
import os
import sys

# Add 310p directory to sys.path
current_dir = os.path.dirname(os.path.abspath(__file__))
parent_dir = os.path.dirname(current_dir) # 310p directory
sys.path.insert(0, parent_dir)

# ruff: noqa: E402
from test_utils import run_vl_model_test


def test_qwen3_vl_8b_tp2_fp16():
"""Qwen3-VL-8B dual-card FP16 test"""
run_vl_model_test(
model_name="Qwen/Qwen3-VL-8B-Instruct",
tensor_parallel_size=2,
max_tokens=5
)
run_vl_model_test(model_name="Qwen/Qwen3-VL-8B-Instruct", tensor_parallel_size=2, max_tokens=5)


def test_qwen3_vl_32b_tp1_fp16():
"""Qwen3-VL-32B 4-card FP16 test"""
run_vl_model_test(
model_name="Qwen/Qwen3-VL-32B-Instruct",
tensor_parallel_size=4,
max_tokens=5
)
run_vl_model_test(model_name="Qwen/Qwen3-VL-32B-Instruct", tensor_parallel_size=4, max_tokens=5)
10 changes: 4 additions & 6 deletions tests/e2e/310p/singlecard/test_vl_model_singlecard.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,20 +15,18 @@
# limitations under the License.
# This file is a part of the vllm-ascend project.

import sys
import os
import sys

# Add 310p directory to sys.path
current_dir = os.path.dirname(os.path.abspath(__file__))
parent_dir = os.path.dirname(current_dir) # 310p directory
sys.path.insert(0, parent_dir)

# ruff: noqa: E402
from test_utils import run_vl_model_test


def test_qwen3_vl_8b_tp1_fp16():
"""Qwen3-VL-8B single-card FP16 test"""
run_vl_model_test(
model_name="Qwen/Qwen3-VL-8B-Instruct",
tensor_parallel_size=1,
max_tokens=5
)
run_vl_model_test(model_name="Qwen/Qwen3-VL-8B-Instruct", tensor_parallel_size=1, max_tokens=5)
23 changes: 10 additions & 13 deletions tests/e2e/310p/test_utils.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,10 +15,12 @@
# limitations under the License.
# This file is a part of the vllm-ascend project.

from tests.e2e.conftest import VllmRunner
from PIL import Image
import os

from PIL import Image

from tests.e2e.conftest import VllmRunner


def get_test_image():
"""Get the image object for testing"""
Expand All @@ -32,14 +34,12 @@ def get_test_prompts():
return ["<|image_pad|>Describe this image in detail."]


def run_vl_model_test(model_name: str,
tensor_parallel_size: int,
max_tokens: int,
dtype: str = "float16",
enforce_eager: bool = True):
def run_vl_model_test(
model_name: str, tensor_parallel_size: int, max_tokens: int, dtype: str = "float16", enforce_eager: bool = True
):
"""
Generic visual language model test function

Args:
model_name: Model name, e.g., "Qwen/Qwen3-VL-4B"
tensor_parallel_size: Tensor parallel size
Expand All @@ -52,9 +52,6 @@ def run_vl_model_test(model_name: str,
prompts = get_test_prompts()

with VllmRunner(
model_name,
tensor_parallel_size=tensor_parallel_size,
enforce_eager=enforce_eager,
dtype=dtype
model_name, tensor_parallel_size=tensor_parallel_size, enforce_eager=enforce_eager, dtype=dtype
) as vllm_model:
vllm_model.generate_greedy(prompts, max_tokens, images=images)
vllm_model.generate_greedy(prompts, max_tokens, images=images)
Loading
Loading