feat(ml): export clip models to ONNX and host models on Hugging Face (#4700)

* export clip models

* export to hf

refactored export code

* export mclip, general refactoring

cleanup

* updated conda deps

* do transforms with pillow and numpy, add tokenization config to export, general refactoring

* moved conda dockerfile, re-added poetry

* minor fixes

* updated link

* updated tests

* removed `requirements.txt` from workflow

* fixed mimalloc path

* removed torchvision

* cleaner np typing

* review suggestions

* update default model name

* update test
This commit is contained in:
Mert 2023-10-31 06:02:04 -04:00 committed by GitHub
parent 3212a47720
commit 87a0ba3db3
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
29 changed files with 6192 additions and 2043 deletions

View file

@ -25,7 +25,7 @@ class InferenceModel(ABC):
) -> None:
self.model_name = model_name
self.loaded = False
self._cache_dir = Path(cache_dir) if cache_dir is not None else get_cache_dir(model_name, self.model_type)
self._cache_dir = Path(cache_dir) if cache_dir is not None else None
self.providers = model_kwargs.pop("providers", ["CPUExecutionProvider"])
# don't pre-allocate more memory than needed
self.provider_options = model_kwargs.pop(
@ -92,7 +92,7 @@ class InferenceModel(ABC):
@property
def cache_dir(self) -> Path:
return self._cache_dir
return self._cache_dir if self._cache_dir is not None else get_cache_dir(self.model_name, self.model_type)
@cache_dir.setter
def cache_dir(self, cache_dir: Path) -> None: