7fe0469e27
### What this PR does / why we need it?
1. For all parts of the current test module involving the millisecond
download model, add the `local_file_only` parameter to specify offline
mode; this ensures that CI will not fail due to network instability.
2. Install modelscope from a fixed commit until it next release
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
check if the env or arg `local_files_only` works
1) set the env:
```shell
export HF_HUB_OFFLINE=1
```
2) run the script
```python
from transformers import PretrainedConfig
import huggingface_hub
from modelscope.utils.hf_util import patch_hub
patch_hub()
model="Qwen/Qwen3-0.6B"
kwargs = {}
config_dict, _ = PretrainedConfig.get_config_dict(
model,
trust_remote_code=True,
local_files_only=huggingface_hub.constants.HF_HUB_OFFLINE,
**kwargs,
)
print(config_dict)
```
it works well:
```shell
2026-03-06 06:40:12,546 - modelscope - WARNING - We can not confirm the cached file is for revision: master
The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.
{'architectures': ['Qwen3ForCausalLM'], 'attention_bias': False, 'attention_dropout': 0.0, 'bos_token_id': 151643, 'eos_token_id': 151645, 'head_dim': 128, 'hidden_act': 'silu', 'hidden_size': 1024, 'initializer_range': 0.02, 'intermediate_size': 3072, 'max_position_embeddings': 40960, 'max_window_layers': 28, 'model_type': 'qwen3', 'num_attention_heads': 16, 'num_hidden_layers': 28, 'num_key_value_heads': 8, 'rms_norm_eps': 1e-06, 'rope_scaling': None, 'rope_theta': 1000000, 'sliding_window': None, 'tie_word_embeddings': True, 'torch_dtype': 'bfloat16', 'transformers_version': '4.51.0', 'use_cache': True, 'use_sliding_window': False, 'vocab_size': 151936, '_commit_hash': None}
```
3) test the model repo does not cached locally when the env
`HF_HUB_OFFLINE`==True
```python
from transformers import PretrainedConfig
import huggingface_hub
from modelscope.utils.hf_util import patch_hub
patch_hub()
model="FireRedTeam/FireRed-OCR"
kwargs = {}
config_dict, _ = PretrainedConfig.get_config_dict(
model,
trust_remote_code=True,
local_files_only=huggingface_hub.constants.HF_HUB_OFFLINE,
**kwargs,
)
print(config_dict)
```
and the result is as expected:
```shell
File "/workspace/demo.py", line 12, in <module>
config_dict, _ = PretrainedConfig.get_config_dict(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/python3.11.14/lib/python3.11/site-packages/modelscope/utils/hf_util/patcher.py", line 189, in patch_get_config_dict
model_dir = get_model_dir(pretrained_model_name_or_path,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/python3.11.14/lib/python3.11/site-packages/modelscope/utils/hf_util/patcher.py", line 164, in get_model_dir
model_dir = snapshot_download(
^^^^^^^^^^^^^^^^^^
File "/usr/local/python3.11.14/lib/python3.11/site-packages/modelscope/hub/snapshot_download.py", line 137, in snapshot_download
return _snapshot_download(
^^^^^^^^^^^^^^^^^^^
File "/usr/local/python3.11.14/lib/python3.11/site-packages/modelscope/hub/snapshot_download.py", line 283, in _snapshot_download
raise ValueError(
ValueError: Cannot find the requested files in the cached path and outgoing traffic has been disabled. To enable look-ups and downloads online, set 'local_files_only' to False
```
- vLLM version: v0.16.0
- vLLM main:
15d76f74e2
---------
Signed-off-by: wangli <wangli858794774@gmail.com>
188 lines
5.8 KiB
Python
188 lines
5.8 KiB
Python
# SPDX-License-Identifier: Apache-2.0
|
|
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
|
import pytest
|
|
import torch
|
|
import torch.nn.functional as F
|
|
import huggingface_hub
|
|
from modelscope import snapshot_download # type: ignore[import-untyped]
|
|
|
|
from tests.e2e.conftest import HfRunner, VllmRunner
|
|
|
|
CROSS_ENCODER_MODELS = [
|
|
"dengcao/ms-marco-MiniLM-L6-v2", # Bert
|
|
"BAAI/bge-reranker-v2-m3", # Roberta
|
|
]
|
|
|
|
EMBEDDING_MODELS = [
|
|
"sentence-transformers/all-MiniLM-L12-v2",
|
|
]
|
|
|
|
TEXTS_1 = [
|
|
"What is the capital of France?",
|
|
"What is the capital of Germany?",
|
|
]
|
|
|
|
TEXTS_2 = [
|
|
"The capital of France is Paris.",
|
|
"The capital of Germany is Berlin.",
|
|
]
|
|
|
|
DTYPE = "half"
|
|
|
|
|
|
@pytest.fixture(scope="module", params=CROSS_ENCODER_MODELS)
|
|
def model_name(request):
|
|
yield snapshot_download(request.param, local_files_only=huggingface_hub.constants.HF_HUB_OFFLINE,)
|
|
|
|
def test_cross_encoder_score_1_to_1(model_name):
|
|
text_pair = [TEXTS_1[0], TEXTS_2[0]]
|
|
|
|
with HfRunner(model_name, dtype=DTYPE, is_cross_encoder=True) as hf_model:
|
|
hf_outputs = hf_model.predict([text_pair]).tolist()
|
|
|
|
with VllmRunner(model_name,
|
|
runner="pooling",
|
|
dtype=DTYPE,
|
|
cudagraph_capture_sizes=[4],
|
|
max_model_len=None) as vllm_model:
|
|
vllm_outputs = vllm_model.score(text_pair[0], text_pair[1])
|
|
|
|
assert len(vllm_outputs) == 1
|
|
assert len(hf_outputs) == 1
|
|
|
|
assert hf_outputs[0] == pytest.approx(vllm_outputs[0], rel=0.01)
|
|
|
|
|
|
def test_cross_encoder_score_1_to_N(model_name):
|
|
text_pairs = [
|
|
[TEXTS_1[0], TEXTS_2[0]],
|
|
[TEXTS_1[0], TEXTS_2[1]],
|
|
]
|
|
|
|
with HfRunner(model_name, dtype=DTYPE, is_cross_encoder=True) as hf_model:
|
|
hf_outputs = hf_model.predict(text_pairs).tolist()
|
|
|
|
with VllmRunner(model_name,
|
|
runner="pooling",
|
|
dtype=DTYPE,
|
|
cudagraph_capture_sizes=[4],
|
|
max_model_len=None) as vllm_model:
|
|
vllm_outputs = vllm_model.score(TEXTS_1[0], TEXTS_2)
|
|
|
|
assert len(vllm_outputs) == 2
|
|
assert len(hf_outputs) == 2
|
|
|
|
assert hf_outputs[0] == pytest.approx(vllm_outputs[0], rel=0.01)
|
|
assert hf_outputs[1] == pytest.approx(vllm_outputs[1], rel=0.01)
|
|
|
|
|
|
def test_cross_encoder_score_N_to_N(model_name):
|
|
text_pairs = [
|
|
[TEXTS_1[0], TEXTS_2[0]],
|
|
[TEXTS_1[1], TEXTS_2[1]],
|
|
]
|
|
|
|
with HfRunner(model_name, dtype=DTYPE, is_cross_encoder=True) as hf_model:
|
|
hf_outputs = hf_model.predict(text_pairs).tolist()
|
|
|
|
with VllmRunner(model_name,
|
|
runner="pooling",
|
|
dtype=DTYPE,
|
|
cudagraph_capture_sizes=[4],
|
|
max_model_len=None) as vllm_model:
|
|
vllm_outputs = vllm_model.score(TEXTS_1, TEXTS_2)
|
|
|
|
assert len(vllm_outputs) == 2
|
|
assert len(hf_outputs) == 2
|
|
|
|
assert hf_outputs[0] == pytest.approx(vllm_outputs[0], rel=0.01)
|
|
assert hf_outputs[1] == pytest.approx(vllm_outputs[1], rel=0.01)
|
|
|
|
|
|
@pytest.fixture(scope="module", params=EMBEDDING_MODELS)
|
|
def emb_model_name(request):
|
|
yield snapshot_download(request.param, local_files_only=huggingface_hub.constants.HF_HUB_OFFLINE,)
|
|
|
|
|
|
def test_embedding_score_1_to_1(emb_model_name):
|
|
text_pair = [TEXTS_1[0], TEXTS_2[0]]
|
|
|
|
with HfRunner(emb_model_name, dtype=DTYPE,
|
|
is_sentence_transformer=True) as hf_model:
|
|
hf_embeddings = hf_model.encode(text_pair)
|
|
hf_outputs = [
|
|
F.cosine_similarity(*map(torch.tensor, hf_embeddings), dim=0)
|
|
]
|
|
|
|
with VllmRunner(emb_model_name,
|
|
runner="pooling",
|
|
dtype=DTYPE,
|
|
cudagraph_capture_sizes=[4],
|
|
max_model_len=None) as vllm_model:
|
|
vllm_outputs = vllm_model.score(text_pair[0], text_pair[1])
|
|
|
|
assert len(vllm_outputs) == 1
|
|
assert len(hf_outputs) == 1
|
|
|
|
assert hf_outputs[0] == pytest.approx(vllm_outputs[0], rel=0.01)
|
|
|
|
|
|
def test_embedding_score_1_to_N(emb_model_name):
|
|
text_pairs = [
|
|
[TEXTS_1[0], TEXTS_2[0]],
|
|
[TEXTS_1[0], TEXTS_2[1]],
|
|
]
|
|
|
|
with HfRunner(emb_model_name, dtype=DTYPE,
|
|
is_sentence_transformer=True) as hf_model:
|
|
hf_embeddings = [
|
|
hf_model.encode(text_pair) for text_pair in text_pairs
|
|
]
|
|
hf_outputs = [
|
|
F.cosine_similarity(*map(torch.tensor, pair), dim=0)
|
|
for pair in hf_embeddings
|
|
]
|
|
|
|
with VllmRunner(emb_model_name,
|
|
runner="pooling",
|
|
dtype=DTYPE,
|
|
cudagraph_capture_sizes=[4],
|
|
max_model_len=None) as vllm_model:
|
|
vllm_outputs = vllm_model.score(TEXTS_1[0], TEXTS_2)
|
|
|
|
assert len(vllm_outputs) == 2
|
|
assert len(hf_outputs) == 2
|
|
|
|
assert hf_outputs[0] == pytest.approx(vllm_outputs[0], rel=0.01)
|
|
assert hf_outputs[1] == pytest.approx(vllm_outputs[1], rel=0.01)
|
|
|
|
|
|
def test_embedding_score_N_to_N(emb_model_name):
|
|
text_pairs = [
|
|
[TEXTS_1[0], TEXTS_2[0]],
|
|
[TEXTS_1[1], TEXTS_2[1]],
|
|
]
|
|
|
|
with HfRunner(emb_model_name, dtype=DTYPE,
|
|
is_sentence_transformer=True) as hf_model:
|
|
hf_embeddings = [
|
|
hf_model.encode(text_pair) for text_pair in text_pairs
|
|
]
|
|
hf_outputs = [
|
|
F.cosine_similarity(*map(torch.tensor, pair), dim=0)
|
|
for pair in hf_embeddings
|
|
]
|
|
|
|
with VllmRunner(emb_model_name,
|
|
runner="pooling",
|
|
dtype=DTYPE,
|
|
cudagraph_capture_sizes=[4],
|
|
max_model_len=None) as vllm_model:
|
|
vllm_outputs = vllm_model.score(TEXTS_1, TEXTS_2)
|
|
|
|
assert len(vllm_outputs) == 2
|
|
assert len(hf_outputs) == 2
|
|
|
|
assert hf_outputs[0] == pytest.approx(vllm_outputs[0], rel=0.01)
|
|
assert hf_outputs[1] == pytest.approx(vllm_outputs[1], rel=0.01)
|