def test_load_old_checkpoint(cls):
dataset = AmazonReviewSentimentCrossLingualDataset()
sha1sum_id = "4ba096cdf6bd76c06386f2c27140db055e59c91b"
checkpoint_name = "mdeberta-v3-base-checkpoint"
save_path = os.path.join(get_home_dir(), "checkpoints")
file_path = os.path.join(save_path, f"{checkpoint_name}.zip")
checkpoint_path = os.path.join(get_home_dir(), "checkpoints",
checkpoint_name)
if os.path.exists(save_path):
shutil.rmtree(save_path)
download(
url=
f"s3://automl-mm-bench/unit-tests-0.4/checkpoints/{checkpoint_name}.zip",
path=file_path,
sha1_hash=sha1sum_id,
)
protected_zip_extraction(
file_path,
sha1_hash=sha1sum_id,
folder=save_path,
)
predictor = cls.load(checkpoint_path)
verify_predictor_save_load(predictor, dataset.test_df, cls=cls)
# continuous training
predictor.fit(
dataset.train_df,
presets="multilingual",
time_limit=10,
hyperparameters={"optimization.top_k_average_method": "uniform_soup"},
)
verify_predictor_save_load(predictor, dataset.test_df, cls=cls)
def __init__(self,):
sha1sum_id = "652a17f1315ec0961336aa140cf983776400c933"
dataset = "san_francisco_airbnb"
file_name = f"{dataset}_for_unit_tests.zip"
url = get_repo_url() + file_name
save_path = os.path.join(get_data_home_dir(), file_name)
self._path = os.path.join(get_data_home_dir(), dataset)
download(
url=url,
path=save_path,
sha1_hash=sha1sum_id,
)
protected_zip_extraction(
save_path,
sha1_hash=sha1sum_id,
folder=self._path,
)
self._train_df = pd.read_csv(os.path.join(self._path, 'train.csv'), index_col=0)
self._test_df = pd.read_csv(os.path.join(self._path, 'test.csv'), index_col=0)
for img_col in self.image_columns:
self._train_df[img_col] = self._train_df[img_col].apply(
lambda ele: path_expander(ele, base_folder=os.path.join(self._path, "images")))
self._test_df[img_col] =\
self._test_df[img_col].apply(
lambda ele: path_expander(ele, base_folder=os.path.join(self._path, "images")))
print(self._train_df[img_col][0])
print(self._test_df[img_col][0])
self._train_df.reset_index(drop=True, inplace=True)
self._test_df.reset_index(drop=True, inplace=True)
print(f"train sample num: {len(self._train_df)}")
print(f"test sample num: {len(self._test_df)}")
def __init__(self,):
sha1sum_id = "2aae657b786f505004ac2922b66097d60a540a58"
dataset = "hateful_memes"
file_name = f"{dataset}_for_unit_tests.zip"
url = get_repo_url() + file_name
save_path = os.path.join(get_data_home_dir(), file_name)
self._path = os.path.join(get_data_home_dir(), dataset)
download(
url=url,
path=save_path,
sha1_hash=sha1sum_id,
)
protected_zip_extraction(
save_path,
sha1_hash=sha1sum_id,
folder=self._path,
)
self._train_df = pd.read_csv(os.path.join(self._path, 'train.csv'), index_col=0)
self._test_df = pd.read_csv(os.path.join(self._path, 'test.csv'), index_col=0)
for img_col in self.image_columns:
self._train_df[img_col] = self._train_df[img_col].apply(
lambda ele: path_expander(ele, base_folder=os.path.join(self._path, "images")))
self._test_df[img_col] =\
self._test_df[img_col].apply(
lambda ele: path_expander(ele, base_folder=os.path.join(self._path, "images")))
print(self._train_df[img_col][0])
print(self._test_df[img_col][0])
self._train_df.reset_index(drop=True, inplace=True)
self._test_df.reset_index(drop=True, inplace=True)
print(f"train sample num: {len(self._train_df)}")
print(f"test sample num: {len(self._test_df)}")
def __init__(self,):
sha1sum_id = "72cb19612318bb304d4a169804f525f88dc3f0d0"
dataset = "petfinder"
file_name = f"{dataset}_for_unit_tests.zip"
url = get_repo_url() + file_name
save_path = os.path.join(get_data_home_dir(), file_name)
self._path = os.path.join(get_data_home_dir(), dataset)
download(
url=url,
path=save_path,
sha1_hash=sha1sum_id,
)
protected_zip_extraction(
save_path,
sha1_hash=sha1sum_id,
folder=self._path,
)
self._train_df = pd.read_csv(os.path.join(self._path, 'train.csv'), index_col=0)
self._test_df = pd.read_csv(os.path.join(self._path, 'test.csv'), index_col=0)
for img_col in self.image_columns:
self._train_df[img_col] = self._train_df[img_col].apply(
lambda ele: path_expander(ele, base_folder=os.path.join(self._path, "images")))
self._test_df[img_col] =\
self._test_df[img_col].apply(
lambda ele: path_expander(ele, base_folder=os.path.join(self._path, "images")))
print(self._train_df[img_col][0])
print(self._test_df[img_col][0])
_, self._train_df = train_test_split(
self._train_df,
test_size=0.1,
random_state=np.random.RandomState(123),
stratify=self._train_df[self.label_columns[0]],
)
_, self._test_df = train_test_split(
self._test_df,
test_size=0.1,
random_state=np.random.RandomState(123),
stratify=self._test_df[self.label_columns[0]],
)
self._train_df.reset_index(drop=True, inplace=True)
self._test_df.reset_index(drop=True, inplace=True)
print(f"train sample num: {len(self._train_df)}")
print(f"test sample num: {len(self._test_df)}")
def __init__(
self,
):
sha1sum_id = "9c701aa6fc42ec3fe429bfe85a8dac4532ab9fcd"
dataset = "amazon_review_sentiment_cross_lingual"
file_name = f"{dataset}.zip"
url = get_repo_url() + file_name
save_path = os.path.join(get_data_home_dir(), file_name)
self._path = os.path.join(get_data_home_dir(), dataset)
download(
url=url,
path=save_path,
sha1_hash=sha1sum_id,
)
protected_zip_extraction(
save_path,
sha1_hash=sha1sum_id,
folder=get_data_home_dir(),
)
self._train_en_df = pd.read_csv(
os.path.join(self._path, "en_train.tsv"),
sep="\t",
header=None,
names=["label", "text"],
).sample(1000, random_state=123)
self._test_en_df = pd.read_csv(
os.path.join(self._path, "en_test.tsv"),
sep="\t",
header=None,
names=["label", "text"],
).sample(200, random_state=123)
self._train_en_df.reset_index(drop=True, inplace=True)
self._test_en_df.reset_index(drop=True, inplace=True)
print(f"train sample num: {len(self._train_en_df)}")
print(f"test sample num: {len(self._test_en_df)}")
def __init__(self,):
sha1sum_id = "8c2a25555c49ef2b30545004488022465808d03f"
dataset = "ae"
file_name = f"{dataset}_for_unit_tests.zip"
url = get_repo_url() + file_name
save_path = os.path.join(get_data_home_dir(), file_name)
self._path = os.path.join(get_data_home_dir(), dataset)
download(
url=url,
path=save_path,
sha1_hash=sha1sum_id,
)
protected_zip_extraction(
save_path,
sha1_hash=sha1sum_id,
folder=self._path,
)
self._train_df = pd.read_csv(os.path.join(self._path, 'train.csv'), index_col=0)
self._test_df = pd.read_csv(os.path.join(self._path, 'test.csv'), index_col=0)
self._train_df.reset_index(drop=True, inplace=True)
self._test_df.reset_index(drop=True, inplace=True)
print(f"train sample num: {len(self._train_df)}")
print(f"test sample num: {len(self._test_df)}")