FIX [Trainer / tags]: Fix trainer + tags when users do not pass "tags" to trainer.push_to_hub() (#29009)

* fix trainer tags

* add test
This commit is contained in:
Younes Belkada
2024-02-14 23:56:35 +01:00
committed by GitHub
parent 5f06053dd8
commit 7a0fccc6eb
2 changed files with 37 additions and 3 deletions

View File

@@ -30,7 +30,7 @@ from typing import Dict, List
from unittest.mock import Mock, patch
import numpy as np
from huggingface_hub import HfFolder, delete_repo, list_repo_commits, list_repo_files
from huggingface_hub import HfFolder, ModelCard, delete_repo, list_repo_commits, list_repo_files
from parameterized import parameterized
from requests.exceptions import HTTPError
@@ -2564,7 +2564,13 @@ class TrainerIntegrationWithHubTester(unittest.TestCase):
@classmethod
def tearDownClass(cls):
for model in ["test-trainer", "test-trainer-epoch", "test-trainer-step", "test-trainer-tensorboard"]:
for model in [
"test-trainer",
"test-trainer-epoch",
"test-trainer-step",
"test-trainer-tensorboard",
"test-trainer-tags",
]:
try:
delete_repo(token=cls._token, repo_id=model)
except HTTPError:
@@ -2695,6 +2701,31 @@ class TrainerIntegrationWithHubTester(unittest.TestCase):
assert found_log is True, "No tensorboard log found in repo"
def test_push_to_hub_tags(self):
# Checks if `trainer.push_to_hub()` works correctly by adding the desired
# tag without having to pass `tags` in `push_to_hub`
# see:
with tempfile.TemporaryDirectory() as tmp_dir:
trainer = get_regression_trainer(
output_dir=os.path.join(tmp_dir, "test-trainer-tags"),
push_to_hub=True,
hub_token=self._token,
)
trainer.model.add_model_tags(["test-trainer-tags"])
url = trainer.push_to_hub()
# Extract repo_name from the url
re_search = re.search(ENDPOINT_STAGING + r"/([^/]+/[^/]+)/", url)
self.assertTrue(re_search is not None)
repo_name = re_search.groups()[0]
self.assertEqual(repo_name, f"{USER}/test-trainer-tags")
model_card = ModelCard.load(repo_name)
self.assertTrue("test-trainer-tags" in model_card.data.tags)
@require_torch
@require_optuna