def test_init_from_existing_local_clone(self):
subprocess.run(
["git", "clone", self._repo_url, WORKING_REPO_DIR],
check=True,
)
repo = Repository(WORKING_REPO_DIR)
repo.lfs_track(["*.pdf"])
repo.lfs_enable_largefiles()
repo.git_pull()
def test_init_clone_in_nonempty_folder(self):
# Create dummy files
# one is lfs-tracked, the other is not.
os.makedirs(WORKING_REPO_DIR, exist_ok=True)
with open(os.path.join(WORKING_REPO_DIR, "dummy.txt"), "w") as f:
f.write("hello")
with open(os.path.join(WORKING_REPO_DIR, "model.bin"), "w") as f:
f.write("hello")
repo = Repository(WORKING_REPO_DIR, clone_from=self._repo_url)
repo.lfs_track(["*.pdf"])
repo.lfs_enable_largefiles()
repo.git_pull()
class RepocardUpdateTest(unittest.TestCase):
_api = HfApi(endpoint=ENDPOINT_STAGING)
@classmethod
def setUpClass(cls):
"""
Share this valid token in all tests below.
"""
cls._token = TOKEN
cls._api.set_access_token(TOKEN)
@retry_endpoint
def setUp(self) -> None:
self.repo_path = Path(tempfile.mkdtemp())
self.REPO_NAME = repo_name()
self.repo = Repository(
self.repo_path / self.REPO_NAME,
clone_from=f"{USER}/{self.REPO_NAME}",
use_auth_token=self._token,
git_user="******",
git_email="*****@*****.**",
)
with self.repo.commit("Add README to main branch"):
with open("README.md", "w+") as f:
f.write(DUMMY_MODELCARD_EVAL_RESULT)
self.existing_metadata = yaml.safe_load(
DUMMY_MODELCARD_EVAL_RESULT.strip().strip("---"))
def tearDown(self) -> None:
self._api.delete_repo(repo_id=f"{self.REPO_NAME}", token=self._token)
shutil.rmtree(self.repo_path)
def test_update_dataset_name(self):
new_datasets_data = {"datasets": "['test/test_dataset']"}
metadata_update(f"{USER}/{self.REPO_NAME}",
new_datasets_data,
token=self._token)
self.repo.git_pull()
updated_metadata = metadata_load(self.repo_path / self.REPO_NAME /
"README.md")
expected_metadata = copy.deepcopy(self.existing_metadata)
expected_metadata.update(new_datasets_data)
self.assertDictEqual(updated_metadata, expected_metadata)
def test_update_existing_result_with_overwrite(self):
new_metadata = copy.deepcopy(self.existing_metadata)
new_metadata["model-index"][0]["results"][0]["metrics"][0][
"value"] = 0.2862102282047272
metadata_update(f"{USER}/{self.REPO_NAME}",
new_metadata,
token=self._token,
overwrite=True)
self.repo.git_pull()
updated_metadata = metadata_load(self.repo_path / self.REPO_NAME /
"README.md")
self.assertDictEqual(updated_metadata, new_metadata)
def test_update_existing_result_without_overwrite(self):
new_metadata = copy.deepcopy(self.existing_metadata)
new_metadata["model-index"][0]["results"][0]["metrics"][0][
"value"] = 0.2862102282047272
with pytest.raises(
ValueError,
match=
("You passed a new value for the existing metric 'name: Accuracy, type:"
" accuracy'. Set `overwrite=True` to overwrite existing metrics."
),
):
metadata_update(
f"{USER}/{self.REPO_NAME}",
new_metadata,
token=self._token,
overwrite=False,
)
def test_update_existing_field_without_overwrite(self):
new_datasets_data = {"datasets": "['test/test_dataset']"}
metadata_update(f"{USER}/{self.REPO_NAME}",
new_datasets_data,
token=self._token)
with pytest.raises(
ValueError,
match=
("You passed a new value for the existing meta data field 'datasets'."
" Set `overwrite=True` to overwrite existing metadata."),
):
new_datasets_data = {"datasets": "['test/test_dataset_2']"}
metadata_update(
f"{USER}/{self.REPO_NAME}",
new_datasets_data,
token=self._token,
overwrite=False,
)
def test_update_new_result_existing_dataset(self):
new_result = metadata_eval_result(
model_pretty_name="RoBERTa fine-tuned on ReactionGIF",
task_pretty_name="Text Classification",
task_id="text-classification",
metrics_pretty_name="Recall",
metrics_id="recall",
metrics_value=0.7762102282047272,
dataset_pretty_name="ReactionGIF",
dataset_id="julien-c/reactiongif",
)
metadata_update(f"{USER}/{self.REPO_NAME}",
new_result,
token=self._token,
overwrite=False)
expected_metadata = copy.deepcopy(self.existing_metadata)
expected_metadata["model-index"][0]["results"][0]["metrics"].append(
new_result["model-index"][0]["results"][0]["metrics"][0])
self.repo.git_pull()
updated_metadata = metadata_load(self.repo_path / self.REPO_NAME /
"README.md")
self.assertDictEqual(updated_metadata, expected_metadata)
def test_update_new_result_new_dataset(self):
new_result = metadata_eval_result(
model_pretty_name="RoBERTa fine-tuned on ReactionGIF",
task_pretty_name="Text Classification",
task_id="text-classification",
metrics_pretty_name="Accuracy",
metrics_id="accuracy",
metrics_value=0.2662102282047272,
dataset_pretty_name="ReactionJPEG",
dataset_id="julien-c/reactionjpeg",
)
metadata_update(f"{USER}/{self.REPO_NAME}",
new_result,
token=self._token,
overwrite=False)
expected_metadata = copy.deepcopy(self.existing_metadata)
expected_metadata["model-index"][0]["results"].append(
new_result["model-index"][0]["results"][0])
self.repo.git_pull()
updated_metadata = metadata_load(self.repo_path / self.REPO_NAME /
"README.md")
self.assertDictEqual(updated_metadata, expected_metadata)
def test_init_clone_in_empty_folder(self):
repo = Repository(WORKING_REPO_DIR, clone_from=self._repo_url)
repo.lfs_track(["*.pdf"])
repo.lfs_enable_largefiles()
repo.git_pull()
def push_to_hub_fastai(
learner,
repo_id: str,
commit_message: Optional[str] = "Add model",
private: Optional[bool] = None,
token: Optional[str] = None,
config: Optional[dict] = None,
**kwargs,
):
"""
Upload learner checkpoint files to the Hub while synchronizing a local clone of the repo in
:obj:`repo_id`.
Args:
learner (`Learner`):
The `fastai.Learner' you'd like to push to the Hub.
repo_id (`str`):
The repository id for your model in Hub in the format of "namespace/repo_name". The namespace can be your individual account or an organization to which you have write access (for example, 'stanfordnlp/stanza-de').
commit_message (`str`, *optional*):
Message to commit while pushing. Will default to :obj:`"add model"`.
private (`bool`, *optional*):
Whether or not the repository created should be private.
token (`str`, *optional*):
The Hugging Face account token to use as HTTP bearer authorization for remote files. If :obj:`None`, the token will be asked by a prompt.
config (`dict`, *optional*):
Configuration object to be saved alongside the model weights.
Keyword Args:
api_endpoint (`str`, *optional*):
The API endpoint to use when pushing the model to the hub.
git_user (`str`, *optional*):
Will override the ``git config user.name`` for committing and pushing files to the hub.
git_email (`str`, *optional*):
Will override the ``git config user.email`` for committing and pushing files to the hub.
Returns:
The url of the commit of your model in the given repository.
<Tip>
Raises the following error:
- [`ValueError`](https://docs.python.org/3/library/exceptions.html#ValueError)
if the user is not log on to the Hugging Face Hub.
</Tip>
"""
_check_fastai_fastcore_versions()
api_endpoint: str = kwargs.get("api_endpoint", None)
git_user: str = kwargs.get("git_user", None)
git_email: str = kwargs.get("git_email", None)
if token is None:
token = HfFolder.get_token()
if token is None:
raise ValueError(
"You must login to the Hugging Face Hub. There are two options: "
"(1) Type `huggingface-cli login` in your terminal and enter your token. "
"(2) Enter your token in the `token` argument. "
"Your token is available in the Settings of your Hugging Face account. "
)
# Create repo using `HfApi()`.
repo_url = HfApi(endpoint=api_endpoint).create_repo(
repo_id,
token=token,
private=private,
repo_type=None,
exist_ok=True,
)
# If repository exists in the Hugging Face Hub then clone it locally in `repo_id`.
repo = Repository(
repo_id,
clone_from=repo_url,
use_auth_token=token,
git_user=git_user,
git_email=git_email,
)
repo.git_pull(rebase=True)
_save_pretrained_fastai(learner, repo_id, config=config)
return repo.push_to_hub(commit_message=commit_message)