ONNX backend does not work when cache_folder is specified and local_files_only is True

[lcorcodilos]

Reproducible example

Fresh environment:

mkdir test
cd test
uv init
uv add "sentence-transformers[onnx]"
mkdir model_cache

Populate the cache:

from sentence_transformers import SentenceTransformer

t = SentenceTransformer(
    "sentence-transformers/all-MiniLM-L6-v2",
    cache_folder="./model_cache/",
    backend="onnx",
    device="cpu",
    local_files_only=False,
)

Now use the cache (local_files_only=True):

from sentence_transformers import SentenceTransformer

t = SentenceTransformer(
    "sentence-transformers/all-MiniLM-L6-v2",
    cache_folder="./model_cache/",
    backend="onnx",
    device="cpu",
    local_files_only=True,
)

Results in the traceback pasted below.

Note that things work without the cache_folder and without using onnx.

Traceback (most recent call last):
  File "/Users/lcorcodilos/Projects/temp/test.py", line 3, in <module>
    t = SentenceTransformer(
        ^^^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/sentence_transformers/SentenceTransformer.py", line 327, in __init__
    modules, self.module_kwargs = self._load_sbert_model(
                                  ^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/sentence_transformers/SentenceTransformer.py", line 2305, in _load_sbert_model
    module = module_class.load(
             ^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/sentence_transformers/models/Transformer.py", line 365, in load
    return cls(model_name_or_path=model_name_or_path, **init_kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/sentence_transformers/models/Transformer.py", line 88, in __init__
    self._load_model(model_name_or_path, config, cache_dir, backend, is_peft_model, **model_args)
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/sentence_transformers/models/Transformer.py", line 200, in _load_model
    self.auto_model = load_onnx_model(
                      ^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/sentence_transformers/backend/load.py", line 73, in load_onnx_model
    model = model_cls.from_pretrained(
            ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/optimum/onnxruntime/modeling.py", line 575, in from_pretrained
    return super().from_pretrained(
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/optimum/modeling_base.py", line 356, in from_pretrained
    library_name = TasksManager.infer_library_from_model(
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/optimum/exporters/tasks.py", line 994, in infer_library_from_model
    library_name = cls._infer_library_from_model_name_or_path(
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/lcorcodilos/Projects/temp/.venv/lib/python3.12/site-packages/optimum/exporters/tasks.py", line 956, in _infer_library_from_model_name_or_path
    raise ValueError(
ValueError: The library name could not be automatically inferred. If using the command-line, please provide the argument --library {transformers,diffusers,timm,sentence_transformers}. Example: `--library diffusers`.