Make bare hugging_face imports forbidden (#49726)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com> Co-authored-by: Andreas Karatzas <akaratza@amd.com>
This commit is contained in:
co-authored by
Andreas Karatzas
parent
0b1a8bb1f6
commit
ca0defa343
@@ -126,10 +126,11 @@ def _load_hf_model(model_name: str, hf_spec: dict, device: torch.device):
|
||||
|
||||
def _load_projection_weight(model_name: str, hf_spec: dict, device: torch.device):
|
||||
"""Download and return the ColBERT linear projection weight."""
|
||||
from huggingface_hub import hf_hub_download
|
||||
from safetensors.torch import load_file
|
||||
|
||||
path = hf_hub_download(model_name, filename=hf_spec["weights_file"])
|
||||
from vllm.transformers_utils.repo_utils import hf_api
|
||||
|
||||
path = hf_api().hf_hub_download(model_name, filename=hf_spec["weights_file"])
|
||||
weights = load_file(path)
|
||||
return weights[hf_spec["weights_key"]].to(device)
|
||||
|
||||
|
||||
@@ -6,7 +6,6 @@ from collections.abc import Sequence
|
||||
|
||||
import pytest
|
||||
import regex as re
|
||||
from huggingface_hub import snapshot_download
|
||||
from transformers import AutoTokenizer
|
||||
|
||||
from vllm.assets.image import ImageAsset
|
||||
@@ -14,6 +13,7 @@ from vllm.logprobs import SampleLogprobs
|
||||
from vllm.lora.request import LoRARequest
|
||||
from vllm.multimodal.image import convert_image_mode, rescale_image_size
|
||||
from vllm.multimodal.media.audio import load_audio
|
||||
from vllm.transformers_utils.repo_utils import hf_api
|
||||
|
||||
from ....conftest import (
|
||||
IMAGE_ASSETS,
|
||||
@@ -35,7 +35,7 @@ HF_MULTIIMAGE_IMAGE_PROMPT = (
|
||||
"<|user|>\n<|image_1|>\n<|image_2|>\nDescribe these images.<|end|>\n<|assistant|>\n" # noqa: E501
|
||||
)
|
||||
|
||||
model_path = snapshot_download("microsoft/Phi-4-multimodal-instruct")
|
||||
model_path = hf_api().snapshot_download("microsoft/Phi-4-multimodal-instruct")
|
||||
# Since the vision-lora and speech-lora co-exist with the base model,
|
||||
# we have to manually specify the path of the lora weights.
|
||||
vision_lora_path = os.path.join(model_path, "vision-lora")
|
||||
|
||||
@@ -3,11 +3,11 @@
|
||||
import pytest
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
from huggingface_hub import snapshot_download
|
||||
from transformers import AutoConfig, AutoModel, CLIPImageProcessor
|
||||
|
||||
from vllm.distributed import cleanup_dist_env_and_memory
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.transformers_utils.repo_utils import hf_api
|
||||
from vllm.utils.torch_utils import STR_DTYPE_TO_TORCH_DTYPE
|
||||
|
||||
from ....conftest import ImageTestAssets
|
||||
@@ -31,7 +31,7 @@ def run_intern_vit_test(
|
||||
*,
|
||||
dtype: str,
|
||||
):
|
||||
model = snapshot_download(model_id, allow_patterns=DOWNLOAD_PATTERN)
|
||||
model = hf_api().snapshot_download(model_id, allow_patterns=DOWNLOAD_PATTERN)
|
||||
torch_dtype = STR_DTYPE_TO_TORCH_DTYPE[dtype]
|
||||
|
||||
img_processor = CLIPImageProcessor.from_pretrained(model)
|
||||
|
||||
@@ -3,13 +3,13 @@
|
||||
import pytest
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
from huggingface_hub import snapshot_download
|
||||
from transformers import AutoConfig, AutoModel, CLIPImageProcessor
|
||||
|
||||
from vllm.distributed import cleanup_dist_env_and_memory
|
||||
from vllm.model_executor.models.radio import RadioModel
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.transformers_utils.configs.radio import RadioConfig
|
||||
from vllm.transformers_utils.repo_utils import hf_api
|
||||
from vllm.utils.torch_utils import STR_DTYPE_TO_TORCH_DTYPE
|
||||
|
||||
from ....conftest import ImageTestAssets
|
||||
@@ -28,7 +28,7 @@ def run_radio_test(
|
||||
*,
|
||||
dtype: str,
|
||||
):
|
||||
model = snapshot_download(model_id, allow_patterns=DOWNLOAD_PATTERN)
|
||||
model = hf_api().snapshot_download(model_id, allow_patterns=DOWNLOAD_PATTERN)
|
||||
torch_dtype = STR_DTYPE_TO_TORCH_DTYPE[dtype]
|
||||
|
||||
img_processor = CLIPImageProcessor.from_pretrained(model)
|
||||
|
||||
@@ -22,6 +22,7 @@ import pytest
|
||||
from packaging import version
|
||||
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.transformers_utils.repo_utils import hf_api
|
||||
|
||||
if current_platform.is_rocm():
|
||||
from vllm.platforms.rocm import on_gfx950
|
||||
@@ -47,7 +48,7 @@ QUARK_MXFP4_AVAILABLE = importlib.util.find_spec("quark") is not None and versio
|
||||
|
||||
def has_huggingface_access(repo):
|
||||
try:
|
||||
huggingface_hub.list_repo_refs(repo)
|
||||
hf_api().list_repo_refs(repo)
|
||||
return True
|
||||
except huggingface_hub.errors.RepositoryNotFoundError:
|
||||
return False
|
||||
|
||||
Reference in New Issue
Block a user