From 0bbaf42635528b7ed54bc448d5df803bc3ab7590 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Wed, 20 Jul 2022 17:21:26 +0200
Subject: [PATCH 01/16] ENH add examples and dtypes to CardData and config.json

---
 examples/plot_hf_hub.py              |  13 ++-
 skops/card/_model_card.py            |  37 +++++--
 skops/hub_utils/_hf_hub.py           | 148 ++++++++++++++++++++++++++-
 skops/hub_utils/tests/test_hf_hub.py |   4 +
 4 files changed, 187 insertions(+), 15 deletions(-)

diff --git a/examples/plot_hf_hub.py b/examples/plot_hf_hub.py
index d1fb4f4c..c79da92d 100644
--- a/examples/plot_hf_hub.py
+++ b/examples/plot_hf_hub.py
@@ -15,12 +15,12 @@
 import json
 import os
 import pickle
+from pathlib import Path
 from tempfile import mkdtemp, mkstemp
 from uuid import uuid4
 
 import sklearn
 from huggingface_hub import HfApi
-from modelcards import CardData
 from sklearn.datasets import load_breast_cancer
 from sklearn.ensemble import HistGradientBoostingClassifier
 from sklearn.experimental import enable_halving_search_cv  # noqa
@@ -76,7 +76,11 @@
 
 local_repo = mkdtemp(prefix="skops-")
 hub_utils.init(
-    model=pkl_name, requirements=[f"scikit-learn={sklearn.__version__}"], dst=local_repo
+    model=pkl_name,
+    requirements=[f"scikit-learn={sklearn.__version__}"],
+    dst=local_repo,
+    task="tabular-classification",
+    data=X_test,
 )
 
 # %%
@@ -86,9 +90,8 @@
 # %%
 # Model Card
 # ==========
-card_data = CardData(tags=["tabular-classification"])
-model_card = card.create_model_card(model, card_data)
-model_card.save(os.path.join(f"{local_repo}", "README.md"))
+model_card = card.create_model_card(local_repo)
+model_card.save(Path(local_repo) / "README.md")
 
 # %%
 # Push to Hub
diff --git a/skops/card/_model_card.py b/skops/card/_model_card.py
index 49b7f1ac..b5e27863 100644
--- a/skops/card/_model_card.py
+++ b/skops/card/_model_card.py
@@ -1,11 +1,15 @@
 import os
+import pickle
 import re
+from pathlib import Path
 
-from modelcards import ModelCard
+from modelcards import CardData, ModelCard
 from sklearn.utils import estimator_html_repr
 
 import skops
 
+from ..hub_utils import get_config
+
 
 def _extract_estimator_config(model):
     """Extracts estimator configuration and renders them into a vertical table.
@@ -27,23 +31,44 @@ def _extract_estimator_config(model):
 
 
 def create_model_card(
-    model,
-    card_data,
+    path,
+    card_data=None,
     **card_kwargs,
 ):
     """Creates a model card for the model and saves it to the target directory.
 
     Parameters:
     ----------
-    model: estimator
-        scikit-learn compatible estimator.
-    card_data: CardData
+    path: str
+        The path to the local directory containing the model and corresponding
+        configuration file.
+    card_data: CardData, optional
         CardData object.
     card_kwargs:
         Card kwargs are information you can pass to fill in the sections of the
         card template, e.g. model_description, citation_bibtex, get_started_code.
     """
     ROOT = skops.__path__
+
+    # Load the model from the existing directory.
+    config = get_config(path)
+    model_path = Path(path) / config["sklearn"]["model"]["file"]
+    with open(model_path, "rb") as f:
+        model = pickle.load(f)
+
+    card_data = card_data or CardData()
+    card_data.tags = card_data.tags or list()
+
+    # Read relevant info from the config file and add them to the CardData
+    # object.
+    if "sklearn" not in card_data.tags:
+        card_data.tags += ["sklearn"]
+
+    if config["sklearn"]["task"] not in card_data.tags:
+        card_data.tags += [config["sklearn"]["task"]]
+
+    setattr(card_data, "widget", config["sklearn"]["example_input"])
+
     model_plot = re.sub(r"\n\s+", "", str(estimator_html_repr(model)))
     hyperparameter_table = _extract_estimator_config(model)
     card_data.library_name = "sklearn"
diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index 1e214f36..de245362 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -9,9 +9,17 @@
 from pathlib import Path
 from typing import List, Union
 
+import numpy as np
 from huggingface_hub import HfApi, snapshot_download
 from requests import HTTPError
 
+SUPPORTED_TASKS = [
+    "tabular-classification",
+    "tabular-regression",
+    "text-classification",
+    "text-regression",
+]
+
 
 def _validate_folder(path: Union[str, Path]):
     """Validate the contents of a folder.
@@ -56,7 +64,76 @@ def _validate_folder(path: Union[str, Path]):
         raise TypeError(f"Model file {model_path} does not exist.")
 
 
-def _create_config(*, model_path: str, requirements: List[str], dst: str):
+def _get_example_input(data):
+    """Returns the example input of a model.
+
+    The input is converted into a dictionary which is then stored in the config
+    file.
+
+    Parameters
+    ----------
+    data: array-like
+        The input needs to be either a ``pandas.DataFrame`` or a
+        ``numpy.ndarray``. The first 3 rows are used as example input.
+
+    Returns
+    -------
+    example_input: dict of lists
+        The example input of the model as accepted by HuggingFace's backend.
+    """
+    try:
+        import pandas as pd
+
+        if isinstance(data, pd.DataFrame):
+            return {x: data[x][:3].to_list() for x in data.columns}
+    except ImportError:
+        # pandas is not installed, the data cannot be a pandas DataFrame
+        pass
+
+    # here we convert the first three rows of the numpy array to a dict of lists
+    # to be stored in the config file
+    if isinstance(data, np.ndarray):
+        return {x: data[:3, x].tolist() for x in range(data.shape[1])}
+
+    raise ValueError("The data is not a pandas.DataFrame or a numpy.ndarray.")
+
+
+def _get_column_dtypes(data):
+    """Returns the dtype of the columns of the data.
+
+    If data is a numpy.ndarray, column names are assumed to be ``X0`` to
+    ``Xn-1``, where ``n`` is the number of columns.
+
+    Parameters
+    ----------
+    data: pandas.DataFrame or numpy.ndarray
+        The data whose columns along with their dtypes are to be returned.
+
+    Returns
+    -------
+    columns: list of tuples
+        A list of tuples of the form (column name, dtype).
+    """
+    try:
+        import pandas as pd
+
+        if isinstance(data, pd.DataFrame):
+            return list(zip(data.dtypes.index, data.dtypes.astype("str")))
+    except ImportError:
+        # pandas is not installed, the data cannot be a pandas DataFrame
+        pass
+
+    # TODO: this is going to fail for Structured Arrays. We can add support for
+    # them later if we see need for it.
+    if isinstance(data, np.ndarray):
+        return [(f"X{x}", str(data.dtype)) for x in range(data.shape[1])]
+
+    raise ValueError("The data is not a pandas.DataFrame or a numpy.ndarray.")
+
+
+def _create_config(
+    *, model_path: str, requirements: List[str], dst: str, task: str, data
+):
     """Write the configuration into a `config.json` file.
 
     Parameters
@@ -71,6 +148,21 @@ def _create_config(*, model_path: str, requirements: List[str], dst: str):
     dst : str, or Path
         The path to an existing folder where the config file should be created.
 
+    task: str
+        The task of the model, which determines the input and output type of
+        the model. It can be one of: ``tabular-classification``,
+        ``tabular-regression``, ``text-classification``, ``text-regression``.
+
+    data: array-like
+        The input to the model. This is used for two purposes:
+
+            1. Save an example input to the model, which is used by HuggingFace's
+                backend and shown in the widget of the model's page.
+            2. Store the dtype of the input, which is used by HugfingFace's backend
+                to pass the data with the right dtype to the model.
+
+        The first 3 input values are used as example inputs.
+
     Returns
     -------
     None
@@ -84,12 +176,29 @@ def recursively_default_dict():
     config = recursively_default_dict()
     config["sklearn"]["model"]["file"] = model_path
     config["sklearn"]["environment"] = requirements
+    config["sklearn"]["task"] = task
+
+    if "tabular" in task:
+        config["sklearn"]["example_input"] = _get_example_input(data)
+        config["sklearn"]["columns"] = _get_column_dtypes(data)
+    elif "text" in task:
+        if isinstance(data, list):
+            config["sklearn"]["example_input"] = {"data": data[:3]}
+        else:
+            raise ValueError("The data needs to be a list of strings.")
 
     with open(Path(dst) / "config.json", mode="w") as f:
         json.dump(config, f, sort_keys=True, indent=4)
 
 
-def init(*, model: Union[str, Path], requirements: List[str], dst: Union[str, Path]):
+def init(
+    *,
+    model: Union[str, Path],
+    requirements: List[str],
+    dst: Union[str, Path],
+    task: str,
+    data,
+):
     """Initialize a scikit-learn based HuggingFace repo.
 
     Given a model pickle and a set of required packages, this function
@@ -107,6 +216,26 @@ def init(*, model: Union[str, Path], requirements: List[str], dst: Union[str, Pa
     dst: str, or Path
         The path to a non-existing or empty folder which is to be initialized.
 
+    task: str
+        The task of the model, which determines the input and output type of
+        the model. It can be one of: ``tabular-classification``,
+        ``tabular-regression``, ``text-classification``, ``text-regression``.
+
+    data: array-like
+        The input to the model. This is used for two purposes:
+
+            1. Save an example input to the model, which is used by HuggingFace's
+                backend and shown in the widget of the model's page.
+            2. Store the dtype of the input, which is used by HugfingFace's backend
+                to pass the data with the right dtype to the model.
+
+        The first 3 input values are used as example inputs.
+
+        If ``task`` is ``tabular-classification`` or ``tabular-regression``,
+        the data needs to be a ``pandas.DataFrame`` or a ``numpy.ndarray``. If
+        ``task`` is ``text-classification`` or ``text-regression``, the data
+        needs to be a ``list`` of strings.
+
     Returns
     -------
     None
@@ -114,12 +243,23 @@ def init(*, model: Union[str, Path], requirements: List[str], dst: Union[str, Pa
     dst = Path(dst)
     if dst.exists() and next(dst.iterdir(), None):
         raise OSError("None-empty dst path already exists!")
+
+    if task not in SUPPORTED_TASKS:
+        raise ValueError(
+            f"Task {task} not supported. Supported tasks are: {SUPPORTED_TASKS}"
+        )
     dst.mkdir(parents=True, exist_ok=True)
 
     shutil.copy2(src=model, dst=dst)
 
     model_name = Path(model).name
-    _create_config(model_path=model_name, requirements=requirements, dst=dst)
+    _create_config(
+        model_path=model_name,
+        requirements=requirements,
+        dst=dst,
+        task=task,
+        data=data,
+    )
 
 
 def update_env(*, path: Union[str, Path], requirements: List[str] = None):
@@ -215,7 +355,7 @@ def get_config(path: Union[str, Path]):
     Parameters
     ----------
     path: str
-        The path to the director holding the project and its ``config.json``
+        The path to the directory holding the project and its ``config.json``
         configuration file.
 
     Returns
diff --git a/skops/hub_utils/tests/test_hf_hub.py b/skops/hub_utils/tests/test_hf_hub.py
index 678bbade..a34f82d6 100644
--- a/skops/hub_utils/tests/test_hf_hub.py
+++ b/skops/hub_utils/tests/test_hf_hub.py
@@ -57,6 +57,7 @@ def test_create_config():
         model_path="model.pkl",
         requirements=['scikit-learn="1.1.1"', "numpy"],
         dst=dir_path,
+        data=None,
     )
 
     config_content = {
@@ -81,6 +82,7 @@ def test_init():
         model=_get_cwd() / "sample_repo/model.pkl",
         requirements=[f'scikit-learn="{version}"'],
         dst=dir_path,
+        data=None,
     )
     _validate_folder(path=dir_path)
 
@@ -90,6 +92,7 @@ def test_init():
             model=_get_cwd() / "sample_repo/model.pkl",
             requirements=[f'scikit-learn="{version}"'],
             dst=dir_path,
+            data=None,
         )
 
 
@@ -102,6 +105,7 @@ def test_push_download(explicit_create):
             model=_get_cwd() / "sample_repo/model.pkl",
             requirements=[f'scikit-learn="{version}"'],
             dst=dir_path,
+            data=None,
         )
 
         user = client.whoami(token=HF_HUB_TOKEN)["name"]

From 65b191c5df102811966a649b4b0710badf1a0288 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Fri, 22 Jul 2022 14:07:57 +0200
Subject: [PATCH 02/16] address some feedback

---
 skops/hub_utils/_hf_hub.py | 29 +++++++++++++++++++++--------
 1 file changed, 21 insertions(+), 8 deletions(-)

diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index de245362..efc4a101 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -7,7 +7,7 @@
 import json
 import shutil
 from pathlib import Path
-from typing import List, Union
+from typing import List, Literal, Union
 
 import numpy as np
 from huggingface_hub import HfApi, snapshot_download
@@ -132,7 +132,17 @@ def _get_column_dtypes(data):
 
 
 def _create_config(
-    *, model_path: str, requirements: List[str], dst: str, task: str, data
+    *,
+    model_path: str,
+    requirements: List[str],
+    dst: str,
+    task: Literal[
+        "tabular-classification",
+        "tabular-regression",
+        "text-classification",
+        "text-regression",
+    ],
+    data,
 ):
     """Write the configuration into a `config.json` file.
 
@@ -148,7 +158,9 @@ def _create_config(
     dst : str, or Path
         The path to an existing folder where the config file should be created.
 
-    task: str
+    task: "tabular-classification", "tabular-regression",
+    "text-classification", /
+            or "text-regression"
         The task of the model, which determines the input and output type of
         the model. It can be one of: ``tabular-classification``,
         ``tabular-regression``, ``text-classification``, ``text-regression``.
@@ -156,10 +168,11 @@ def _create_config(
     data: array-like
         The input to the model. This is used for two purposes:
 
-            1. Save an example input to the model, which is used by HuggingFace's
-                backend and shown in the widget of the model's page.
-            2. Store the dtype of the input, which is used by HugfingFace's backend
-                to pass the data with the right dtype to the model.
+            1. Save an example input to the model, which is used by
+               HuggingFace's backend and shown in the widget of the model's
+               page.
+            2. Store the dtype of the input, which is used by HuggingFace's
+               backend to pass the data with the right dtype to the model.
 
         The first 3 input values are used as example inputs.
 
@@ -182,7 +195,7 @@ def recursively_default_dict():
         config["sklearn"]["example_input"] = _get_example_input(data)
         config["sklearn"]["columns"] = _get_column_dtypes(data)
     elif "text" in task:
-        if isinstance(data, list):
+        if isinstance(data, list) and all(isinstance(x, str) for x in data):
             config["sklearn"]["example_input"] = {"data": data[:3]}
         else:
             raise ValueError("The data needs to be a list of strings.")

From 475bc43c47a2c386764003bdf8565802e9bad876 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 12:39:09 +0200
Subject: [PATCH 03/16] address more comments

---
 skops/card/_model_card.py  | 2 +-
 skops/hub_utils/_hf_hub.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/skops/card/_model_card.py b/skops/card/_model_card.py
index b5e27863..29a10d28 100644
--- a/skops/card/_model_card.py
+++ b/skops/card/_model_card.py
@@ -67,7 +67,7 @@ def create_model_card(
     if config["sklearn"]["task"] not in card_data.tags:
         card_data.tags += [config["sklearn"]["task"]]
 
-    setattr(card_data, "widget", config["sklearn"]["example_input"])
+    card_data.widget = config["sklearn"]["example_input"]
 
     model_plot = re.sub(r"\n\s+", "", str(estimator_html_repr(model)))
     hyperparameter_table = _extract_estimator_config(model)
diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index efc4a101..d9cdb73e 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -126,7 +126,7 @@ def _get_column_dtypes(data):
     # TODO: this is going to fail for Structured Arrays. We can add support for
     # them later if we see need for it.
     if isinstance(data, np.ndarray):
-        return [(f"X{x}", str(data.dtype)) for x in range(data.shape[1])]
+        return [(f"x{x}", str(data.dtype)) for x in range(data.shape[1])]
 
     raise ValueError("The data is not a pandas.DataFrame or a numpy.ndarray.")
 

From d85a7f49ddff4a358d624b6d7b0d0d6384afa40d Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 13:28:24 +0200
Subject: [PATCH 04/16] improve tests

---
 skops/hub_utils/tests/test_hf_hub.py | 67 +++++++++++++++++++++-------
 1 file changed, 51 insertions(+), 16 deletions(-)

diff --git a/skops/hub_utils/tests/test_hf_hub.py b/skops/hub_utils/tests/test_hf_hub.py
index e83355b5..a283123b 100644
--- a/skops/hub_utils/tests/test_hf_hub.py
+++ b/skops/hub_utils/tests/test_hf_hub.py
@@ -8,6 +8,7 @@
 
 import pytest
 from huggingface_hub import HfApi
+from sklearn.datasets import load_iris
 from sklearn.linear_model import LogisticRegression
 
 from skops.hub_utils import download, get_config, get_requirements, init, push
@@ -29,10 +30,16 @@ def destination_path():
 
 
 @pytest.fixture(scope="session")
-def model_pickle(repo_path):
+def classification_data():
+    return load_iris(return_X_y=True, as_frame=True)
+
+
+@pytest.fixture(scope="session")
+def classifier_pickle(repo_path, classification_data):
     # Create a simple pickle file for the purpose of testing
-    clf = LogisticRegression()
-    clf.fit([[0, 1], [1, 0]], [0, 1])
+    X, y = classification_data
+    clf = LogisticRegression(solver="newton-cg")
+    clf.fit(X, y)
     path = repo_path / "model.pickle"
 
     try:
@@ -89,63 +96,91 @@ def test_validate_folder(config_json):
     _validate_folder(path=dir_path)
 
 
-def test_create_config():
+def test_create_config(classification_data):
     dir_path = tempfile.mkdtemp()
     _create_config(
         model_path="model.pkl",
         requirements=['scikit-learn="1.1.1"', "numpy"],
         dst=dir_path,
-        data=None,
+        task="tabular-classification",
+        data=classification_data[0],
     )
 
-    config_content = {
+    config_expected = {
         "sklearn": {
+            "columns": [
+                ["petal length (cm)", "float64"],
+                ["petal width (cm)", "float64"],
+                ["sepal length (cm)", "float64"],
+                ["sepal width (cm)", "float64"],
+            ],
             "environment": ['scikit-learn="1.1.1"', "numpy"],
+            "example_input": {
+                "petal length (cm)": [1.4, 1.4, 1.3],
+                "petal width (cm)": [0.2, 0.2, 0.2],
+                "sepal length (cm)": [5.1, 4.9, 4.7],
+                "sepal width (cm)": [3.5, 3.0, 3.2],
+            },
             "model": {"file": "model.pkl"},
+            "task": "tabular-classification",
         }
     }
 
     with open(Path(dir_path) / "config.json") as f:
         config = json.load(f)
-        assert config == config_content
+        for key in ["environment", "model", "task"]:
+            assert config["sklearn"][key] == config_expected["sklearn"][key]
+
+        for key in ["columns", "example_input"]:
+            assert sorted(config["sklearn"][key]) == sorted(
+                config_expected["sklearn"][key]
+            )
 
 
-def test_init(model_pickle, config_json):
+def test_init(classifier_pickle, classification_data, config_json):
     # create a temp directory and delete it, we just need a unique name.
     dir_path = tempfile.mkdtemp()
     shutil.rmtree(dir_path)
 
     version = metadata.version("scikit-learn")
     init(
-        model=model_pickle,
+        model=classifier_pickle,
         requirements=[f'scikit-learn="{version}"'],
         dst=dir_path,
-        data=None,
+        task="tabular-classification",
+        data=classification_data[0],
     )
     _validate_folder(path=dir_path)
 
     # it should fail a second time since the folder is no longer empty.
     with pytest.raises(OSError, match="None-empty dst path already exists!"):
         init(
-            model=model_pickle,
+            model=classifier_pickle,
             requirements=[f'scikit-learn="{version}"'],
             dst=dir_path,
-            data=None,
+            task="tabular-classification",
+            data=classification_data[0],
         )
 
 
 @pytest.mark.parametrize("explicit_create", [True, False])
 def test_push_download(
-    explicit_create, repo_path, destination_path, model_pickle, config_json
+    explicit_create,
+    repo_path,
+    destination_path,
+    classifier_pickle,
+    classification_data,
+    config_json,
 ):
     client = HfApi()
 
     version = metadata.version("scikit-learn")
     init(
-        model=model_pickle,
+        model=classifier_pickle,
         requirements=[f'scikit-learn="{version}"'],
         dst=destination_path,
-        data=None,
+        task="tabular-classification",
+        data=classification_data[0],
     )
 
     user = client.whoami(token=HF_HUB_TOKEN)["name"]
@@ -164,7 +199,7 @@ def test_push_download(
         download(repo_id=repo_id, dst=destination_path)
 
     files = client.list_repo_files(repo_id=repo_id, token=HF_HUB_TOKEN)
-    for f_name in [model_pickle.name, config_json.name]:
+    for f_name in [classifier_pickle.name, config_json.name]:
         assert f_name in files
 
     try:

From 01c507bddf0cdcaf37ef233f67ac13f657e8031d Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 13:59:09 +0200
Subject: [PATCH 05/16] fix model card tests

---
 skops/card/tests/test_card.py | 83 ++++++++++++++++++++++-------------
 1 file changed, 52 insertions(+), 31 deletions(-)

diff --git a/skops/card/tests/test_card.py b/skops/card/tests/test_card.py
index a5fb2fe9..24cb775b 100644
--- a/skops/card/tests/test_card.py
+++ b/skops/card/tests/test_card.py
@@ -1,26 +1,48 @@
-import os
+import pickle
 import tempfile
+from pathlib import Path
 
 import numpy as np
+import pytest
+import sklearn
 from modelcards import CardData
 from sklearn.linear_model import LinearRegression
 
+from skops import hub_utils
 from skops.card import create_model_card
 
 
-def fit_model():
+@pytest.fixture
+def temp_path():
+    with tempfile.TemporaryDirectory(prefix="skops-test") as dir_path:
+        yield Path(dir_path)
+
+
+@pytest.fixture
+def model_folder(temp_path):
     X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
     y = np.dot(X, np.array([1, 2])) + 3
     reg = LinearRegression().fit(X, y)
-    return reg
+    with tempfile.TemporaryDirectory(prefix="skops-test") as dir_path:
+        model_path = Path(dir_path) / "model.pkl"
+        with open(model_path, "wb") as f:
+            pickle.dump(reg, f)
+        hub_utils.init(
+            model=model_path,
+            requirements=[f"scikit-learn{sklearn.__version__}"],
+            task="tabular-classification",
+            data=X,
+            dst=temp_path,
+        )
+    yield Path(temp_path)
 
 
-def generate_card():
-    model = fit_model()
+@pytest.fixture
+def sample_card(model_folder):
     card_data = CardData(library_name="sklearn")
 
     model_card = create_model_card(
-        model,
+        model_folder,
         card_data,
         template_path="skops/card/default_template.md",
         model_description="sklearn FTW",
@@ -28,32 +50,31 @@ def generate_card():
     return model_card
 
 
-def test_write_model_card():
-    with tempfile.TemporaryDirectory(prefix="skops-test") as dir_path:
-        model = fit_model()
-        card_data = CardData(library_name="sklearn")
-        model_card = create_model_card(
-            model, card_data=card_data, model_description="sklearn FTW"
-        )
-        model_card.save(os.path.join(f"{dir_path}", "README.md"))
-        with open(os.path.join(f"{dir_path}", "README.md"), "r") as f:
-            model_card = f.read()
-        assert "sklearn FTW" in model_card
+def test_write_model_card(model_folder):
+    card_data = CardData(library_name="sklearn")
+    model_card = create_model_card(
+        model_folder,
+        card_data=card_data,
+        model_description="sklearn FTW",
+    )
+    card_path = model_folder / "README.md"
+    model_card.save(card_path)
+    with open(card_path, "r") as f:
+        model_card = f.read()
+    assert "sklearn FTW" in model_card
 
 
-def test_hyperparameter_table():
-    with tempfile.TemporaryDirectory(prefix="skops-test") as dir_path:
-        model_card = generate_card()
-        model_card.save(os.path.join(f"{dir_path}", "README.md"))
-        with open(os.path.join(f"{dir_path}", "README.md"), "r") as f:
-            model_card = f.read()
-        assert "fit_intercept" in model_card
+def test_hyperparameter_table(sample_card, temp_path):
+    card_path = temp_path / "README.md"
+    sample_card.save(card_path)
+    with open(card_path, "r") as f:
+        model_card = f.read()
+    assert "fit_intercept" in model_card
 
 
-def test_plot_model():
-    with tempfile.TemporaryDirectory(prefix="skops-test") as dir_path:
-        model_card = generate_card()
-        model_card.save(os.path.join(f"{dir_path}", "README.md"))
-        with open(os.path.join(f"{dir_path}", "README.md"), "r") as f:
-            model_card = f.read()
-        assert "<style>" in model_card
+def test_plot_model(sample_card, temp_path):
+    card_path = temp_path / "README.md"
+    sample_card.save(card_path)
+    with open(card_path, "r") as f:
+        model_card = f.read()
+    assert "<style>" in model_card

From 4b6b9c70128a069f95b03ff9b1934ea840dc0ebc Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 14:06:30 +0200
Subject: [PATCH 06/16] add typing_extensions to python 3.7

---
 skops/_min_dependencies.py | 1 +
 skops/hub_utils/_hf_hub.py | 4 +++-
 skops/utils/fixes.py       | 6 ++++++
 3 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/skops/_min_dependencies.py b/skops/_min_dependencies.py
index 4c7ce871..9b6e907f 100644
--- a/skops/_min_dependencies.py
+++ b/skops/_min_dependencies.py
@@ -24,6 +24,7 @@
     "sphinx-prompt": ("1.3.0", "docs", None),
     "matplotlib": ("3.3", "docs", None),
     "pandas": ("1", "docs", None),
+    "typing_extensions": ("3.7", "install", "python_full_version < '3.8'"),
 }
 
 
diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index d9cdb73e..1d9110f0 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -7,12 +7,14 @@
 import json
 import shutil
 from pathlib import Path
-from typing import List, Literal, Union
+from typing import List, Union
 
 import numpy as np
 from huggingface_hub import HfApi, snapshot_download
 from requests import HTTPError
 
+from ..utils.fixes import Literal
+
 SUPPORTED_TASKS = [
     "tabular-classification",
     "tabular-regression",
diff --git a/skops/utils/fixes.py b/skops/utils/fixes.py
index e386bbd8..9367c780 100644
--- a/skops/utils/fixes.py
+++ b/skops/utils/fixes.py
@@ -15,6 +15,12 @@
     # older pythons
     import importlib_metadata as metadata  # noqa
 
+try:
+    # py>=3.8
+    from typing import Literal  # noqa
+except ImportError:
+    from typing_extensions import Literal  # noqa
+
 
 def path_unlink(path: Path, missing_ok=False) -> None:
     """Remove this file or symbolic link

From a9bc358cc1e01848e089b37aae73e014ca5f3019 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 14:07:37 +0200
Subject: [PATCH 07/16] add comment

---
 skops/utils/fixes.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/skops/utils/fixes.py b/skops/utils/fixes.py
index 9367c780..e4b874be 100644
--- a/skops/utils/fixes.py
+++ b/skops/utils/fixes.py
@@ -19,6 +19,9 @@
     # py>=3.8
     from typing import Literal  # noqa
 except ImportError:
+    # older pythons, this requires typing_extensions to be installed.
+    # if you're removing this, you should also remove the dependency from
+    # _min_dependencies.py
     from typing_extensions import Literal  # noqa
 
 

From 95f47a2dc6b4337f303d90aed1456140616d1ef0 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 14:19:03 +0200
Subject: [PATCH 08/16] fix model card example

---
 examples/plot_model_card.py | 26 +++++++++++++++-----------
 1 file changed, 15 insertions(+), 11 deletions(-)

diff --git a/examples/plot_model_card.py b/examples/plot_model_card.py
index 5531d80b..7c88e560 100644
--- a/examples/plot_model_card.py
+++ b/examples/plot_model_card.py
@@ -56,6 +56,20 @@
 ).fit(X_train, y_train)
 model.score(X_test, y_test)
 
+_, pkl_name = mkstemp(prefix="skops-", suffix=".pkl")
+
+with open(pkl_name, mode="bw") as f:
+    pickle.dump(model, file=f)
+
+local_repo = mkdtemp(prefix="skops-")
+hub_utils.init(
+    model=pkl_name,
+    requirements=[f"scikit-learn={sklearn.__version__}"],
+    dst=local_repo,
+    task="tabular-classification",
+    data=X_test,
+)
+
 # %%
 # Create a model card
 # ====================
@@ -86,7 +100,7 @@
 citation = "bibtex\n@inproceedings{...,year={2020}}"
 
 model_card = card.create_model_card(
-    model,
+    local_repo,
     card_data=card_data,
     limitations=limitations,
     model_description=model_description,
@@ -95,14 +109,4 @@
     get_started_code=get_started_code,
 )
 
-_, pkl_name = mkstemp(prefix="skops-", suffix=".pkl")
-
-with open(pkl_name, mode="bw") as f:
-    pickle.dump(model, file=f)
-
-local_repo = mkdtemp(prefix="skops-")
-hub_utils.init(
-    model=pkl_name, requirements=[f"scikit-learn={sklearn.__version__}"], dst=local_repo
-)
-
 model_card.save(os.path.join(f"{local_repo}", "README.md"))

From a5c02f6dc7edc91e15ef4e0b88c5608e4f37a655 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 15:34:41 +0200
Subject: [PATCH 09/16] add dtype and get_example tests

---
 skops/hub_utils/_hf_hub.py           |  2 +-
 skops/hub_utils/tests/test_hf_hub.py | 46 +++++++++++++++++++++++++++-
 2 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index 1d9110f0..08b16413 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -95,7 +95,7 @@ def _get_example_input(data):
     # here we convert the first three rows of the numpy array to a dict of lists
     # to be stored in the config file
     if isinstance(data, np.ndarray):
-        return {x: data[:3, x].tolist() for x in range(data.shape[1])}
+        return {f"x{x}": data[:3, x].tolist() for x in range(data.shape[1])}
 
     raise ValueError("The data is not a pandas.DataFrame or a numpy.ndarray.")
 
diff --git a/skops/hub_utils/tests/test_hf_hub.py b/skops/hub_utils/tests/test_hf_hub.py
index a283123b..2555d251 100644
--- a/skops/hub_utils/tests/test_hf_hub.py
+++ b/skops/hub_utils/tests/test_hf_hub.py
@@ -6,13 +6,20 @@
 from pathlib import Path
 from uuid import uuid4
 
+import numpy as np
+import pandas as pd
 import pytest
 from huggingface_hub import HfApi
 from sklearn.datasets import load_iris
 from sklearn.linear_model import LogisticRegression
 
 from skops.hub_utils import download, get_config, get_requirements, init, push
-from skops.hub_utils._hf_hub import _create_config, _validate_folder
+from skops.hub_utils._hf_hub import (
+    _create_config,
+    _get_column_dtypes,
+    _get_example_input,
+    _validate_folder,
+)
 from skops.hub_utils.tests.common import HF_HUB_TOKEN
 from skops.utils.fixes import metadata, path_unlink
 
@@ -215,3 +222,40 @@ def test_get_config(repo_path):
     config = get_config(repo_path)
     assert config == CONFIG
     assert get_requirements(repo_path) == ['scikit-learn="1.1.1"']
+
+
+def test_get_example_input():
+    """Test the _get_example_input function."""
+    with pytest.raises(
+        ValueError, match="The data is not a pandas.DataFrame or a numpy.ndarray."
+    ):
+        _get_example_input(["a", "b", "c"])
+
+    examples = _get_example_input(np.ones((5, 10)))
+    # the result if a dictionary of column name: list of values
+    assert len(examples) == 10
+    assert len(examples["x0"]) == 3
+
+    examples = _get_example_input(
+        pd.DataFrame(np.ones((5, 10)), columns=[f"column{x}" for x in range(10)])
+    )
+    # the result if a dictionary of column name: list of values
+    assert len(examples) == 10
+    assert len(examples["column0"]) == 3
+
+
+def test_get_column_dtypes():
+    with pytest.raises(
+        ValueError, match="The data is not a pandas.DataFrame or a numpy.ndarray."
+    ):
+        _get_column_dtypes(["a", "b", "c"])
+
+    X_array = np.ones((5, 10), dtype=np.float32)
+    expected_columns = [f"x{x}" for x in range(10)]
+    expected_dtypes = list(zip(expected_columns, [np.float32] * 10))
+    assert _get_column_dtypes(X_array) == expected_dtypes
+
+    expected_columns = [f"column{x}" for x in range(10)]
+    X_df = pd.DataFrame(X_array, columns=expected_columns)
+    expected_dtypes = zip(*(expected_columns, ["float32"] * 10))
+    assert _get_column_dtypes(X_df) == expected_dtypes

From b20b9394a4ec24a4eb123b9d70c38269b56e2b97 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Mon, 25 Jul 2022 16:07:14 +0200
Subject: [PATCH 10/16] remove dtypes

---
 skops/hub_utils/_hf_hub.py           | 18 +++++++++---------
 skops/hub_utils/tests/test_hf_hub.py | 20 +++++++++-----------
 2 files changed, 18 insertions(+), 20 deletions(-)

diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index 08b16413..a0e0fb3a 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -100,27 +100,27 @@ def _get_example_input(data):
     raise ValueError("The data is not a pandas.DataFrame or a numpy.ndarray.")
 
 
-def _get_column_dtypes(data):
-    """Returns the dtype of the columns of the data.
+def _get_column_names(data):
+    """Returns the column names of the input.
 
-    If data is a numpy.ndarray, column names are assumed to be ``X0`` to
-    ``Xn-1``, where ``n`` is the number of columns.
+    If data is a ``numpy.ndarray``, column names are assumed to be ``x0`` to
+    ``xn-1``, where ``n`` is the number of columns.
 
     Parameters
     ----------
     data: pandas.DataFrame or numpy.ndarray
-        The data whose columns along with their dtypes are to be returned.
+        The data whose columns names are to be returned.
 
     Returns
     -------
     columns: list of tuples
-        A list of tuples of the form (column name, dtype).
+        A list of strings. Each string is a column name.
     """
     try:
         import pandas as pd
 
         if isinstance(data, pd.DataFrame):
-            return list(zip(data.dtypes.index, data.dtypes.astype("str")))
+            return list(data.columns)
     except ImportError:
         # pandas is not installed, the data cannot be a pandas DataFrame
         pass
@@ -128,7 +128,7 @@ def _get_column_dtypes(data):
     # TODO: this is going to fail for Structured Arrays. We can add support for
     # them later if we see need for it.
     if isinstance(data, np.ndarray):
-        return [(f"x{x}", str(data.dtype)) for x in range(data.shape[1])]
+        return [f"x{x}" for x in range(data.shape[1])]
 
     raise ValueError("The data is not a pandas.DataFrame or a numpy.ndarray.")
 
@@ -195,7 +195,7 @@ def recursively_default_dict():
 
     if "tabular" in task:
         config["sklearn"]["example_input"] = _get_example_input(data)
-        config["sklearn"]["columns"] = _get_column_dtypes(data)
+        config["sklearn"]["columns"] = _get_column_names(data)
     elif "text" in task:
         if isinstance(data, list) and all(isinstance(x, str) for x in data):
             config["sklearn"]["example_input"] = {"data": data[:3]}
diff --git a/skops/hub_utils/tests/test_hf_hub.py b/skops/hub_utils/tests/test_hf_hub.py
index 2555d251..3bd3c740 100644
--- a/skops/hub_utils/tests/test_hf_hub.py
+++ b/skops/hub_utils/tests/test_hf_hub.py
@@ -16,7 +16,7 @@
 from skops.hub_utils import download, get_config, get_requirements, init, push
 from skops.hub_utils._hf_hub import (
     _create_config,
-    _get_column_dtypes,
+    _get_column_names,
     _get_example_input,
     _validate_folder,
 )
@@ -116,10 +116,10 @@ def test_create_config(classification_data):
     config_expected = {
         "sklearn": {
             "columns": [
-                ["petal length (cm)", "float64"],
-                ["petal width (cm)", "float64"],
-                ["sepal length (cm)", "float64"],
-                ["sepal width (cm)", "float64"],
+                "petal length (cm)",
+                "petal width (cm)",
+                "sepal length (cm)",
+                "sepal width (cm)",
             ],
             "environment": ['scikit-learn="1.1.1"', "numpy"],
             "example_input": {
@@ -244,18 +244,16 @@ def test_get_example_input():
     assert len(examples["column0"]) == 3
 
 
-def test_get_column_dtypes():
+def test_get_column_names():
     with pytest.raises(
         ValueError, match="The data is not a pandas.DataFrame or a numpy.ndarray."
     ):
-        _get_column_dtypes(["a", "b", "c"])
+        _get_column_names(["a", "b", "c"])
 
     X_array = np.ones((5, 10), dtype=np.float32)
     expected_columns = [f"x{x}" for x in range(10)]
-    expected_dtypes = list(zip(expected_columns, [np.float32] * 10))
-    assert _get_column_dtypes(X_array) == expected_dtypes
+    assert _get_column_names(X_array) == expected_columns
 
     expected_columns = [f"column{x}" for x in range(10)]
     X_df = pd.DataFrame(X_array, columns=expected_columns)
-    expected_dtypes = zip(*(expected_columns, ["float32"] * 10))
-    assert _get_column_dtypes(X_df) == expected_dtypes
+    assert _get_column_names(X_df) == expected_columns

From 080af4e0520dbd79dd41bec5ea60640e35dd0548 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Tue, 26 Jul 2022 12:29:44 +0200
Subject: [PATCH 11/16] improve coverage

---
 skops/hub_utils/tests/test_hf_hub.py | 126 +++++++++++++++++++--------
 1 file changed, 88 insertions(+), 38 deletions(-)

diff --git a/skops/hub_utils/tests/test_hf_hub.py b/skops/hub_utils/tests/test_hf_hub.py
index 3bd3c740..abb54920 100644
--- a/skops/hub_utils/tests/test_hf_hub.py
+++ b/skops/hub_utils/tests/test_hf_hub.py
@@ -23,6 +23,14 @@
 from skops.hub_utils.tests.common import HF_HUB_TOKEN
 from skops.utils.fixes import metadata, path_unlink
 
+iris = load_iris(as_frame=True, return_X_y=False)
+
+
+@pytest.fixture
+def temp_path():
+    with tempfile.TemporaryDirectory(prefix="skops-test-temp-path") as temp_path:
+        yield temp_path
+
 
 @pytest.fixture(scope="session")
 def repo_path():
@@ -37,14 +45,9 @@ def destination_path():
 
 
 @pytest.fixture(scope="session")
-def classification_data():
-    return load_iris(return_X_y=True, as_frame=True)
-
-
-@pytest.fixture(scope="session")
-def classifier_pickle(repo_path, classification_data):
+def classifier_pickle(repo_path):
     # Create a simple pickle file for the purpose of testing
-    X, y = classification_data
+    X, y = iris.data, iris.target
     clf = LogisticRegression(solver="newton-cg")
     clf.fit(X, y)
     path = repo_path / "model.pickle"
@@ -103,48 +106,96 @@ def test_validate_folder(config_json):
     _validate_folder(path=dir_path)
 
 
-def test_create_config(classification_data):
+@pytest.mark.parametrize(
+    "data, task, expected_config",
+    [
+        (
+            iris.data,
+            "tabular-classification",
+            {
+                "sklearn": {
+                    "columns": [
+                        "petal length (cm)",
+                        "petal width (cm)",
+                        "sepal length (cm)",
+                        "sepal width (cm)",
+                    ],
+                    "environment": ['scikit-learn="1.1.1"', "numpy"],
+                    "example_input": {
+                        "petal length (cm)": [1.4, 1.4, 1.3],
+                        "petal width (cm)": [0.2, 0.2, 0.2],
+                        "sepal length (cm)": [5.1, 4.9, 4.7],
+                        "sepal width (cm)": [3.5, 3.0, 3.2],
+                    },
+                    "model": {"file": "model.pkl"},
+                    "task": "tabular-classification",
+                }
+            },
+        ),
+        (
+            ["test", "text", "problem", "random"],
+            "text-classification",
+            {
+                "sklearn": {
+                    "environment": ['scikit-learn="1.1.1"', "numpy"],
+                    "example_input": {"data": ["test", "text", "problem"]},
+                    "model": {"file": "model.pkl"},
+                    "task": "text-classification",
+                }
+            },
+        ),
+    ],
+)
+def test_create_config(data, task, expected_config):
     dir_path = tempfile.mkdtemp()
     _create_config(
         model_path="model.pkl",
         requirements=['scikit-learn="1.1.1"', "numpy"],
         dst=dir_path,
-        task="tabular-classification",
-        data=classification_data[0],
+        task=task,
+        data=data,
     )
 
-    config_expected = {
-        "sklearn": {
-            "columns": [
-                "petal length (cm)",
-                "petal width (cm)",
-                "sepal length (cm)",
-                "sepal width (cm)",
-            ],
-            "environment": ['scikit-learn="1.1.1"', "numpy"],
-            "example_input": {
-                "petal length (cm)": [1.4, 1.4, 1.3],
-                "petal width (cm)": [0.2, 0.2, 0.2],
-                "sepal length (cm)": [5.1, 4.9, 4.7],
-                "sepal width (cm)": [3.5, 3.0, 3.2],
-            },
-            "model": {"file": "model.pkl"},
-            "task": "tabular-classification",
-        }
-    }
-
     with open(Path(dir_path) / "config.json") as f:
         config = json.load(f)
         for key in ["environment", "model", "task"]:
-            assert config["sklearn"][key] == config_expected["sklearn"][key]
+            assert config["sklearn"][key] == expected_config["sklearn"][key]
 
-        for key in ["columns", "example_input"]:
+        keys = ["example_input"]
+        if "tabular" in task:
+            # text data doesn't introduce any "columns" in the configuration
+            keys += ["columns"]
+        for key in keys:
             assert sorted(config["sklearn"][key]) == sorted(
-                config_expected["sklearn"][key]
+                expected_config["sklearn"][key]
             )
 
 
-def test_init(classifier_pickle, classification_data, config_json):
+def test_create_config_invalid_data(temp_path):
+    with pytest.raises(ValueError, match="The data needs to be a list of strings."):
+        _create_config(
+            model_path="model.pkl",
+            requirements=['scikit-learn="1.1.1"', "numpy"],
+            task="text-classification",
+            data=[1, 2, 3],
+            dst=temp_path,
+        )
+
+
+def test_init_invalid_task(classifier_pickle, temp_path):
+    with pytest.raises(
+        ValueError, match="Task invalid not supported. Supported tasks are"
+    ):
+        init(
+            model=classifier_pickle,
+            requirements=["scikit-learn"],
+            dst=temp_path,
+            task="invalid",
+            data=iris.data,
+        )
+
+
+def test_init(classifier_pickle, config_json):
     # create a temp directory and delete it, we just need a unique name.
     dir_path = tempfile.mkdtemp()
     shutil.rmtree(dir_path)
@@ -155,7 +206,7 @@ def test_init(classifier_pickle, classification_data, config_json):
         requirements=[f'scikit-learn="{version}"'],
         dst=dir_path,
         task="tabular-classification",
-        data=classification_data[0],
+        data=iris.data,
     )
     _validate_folder(path=dir_path)
 
@@ -166,7 +217,7 @@ def test_init(classifier_pickle, classification_data, config_json):
             requirements=[f'scikit-learn="{version}"'],
             dst=dir_path,
             task="tabular-classification",
-            data=classification_data[0],
+            data=iris.data,
         )
 
 
@@ -176,7 +227,6 @@ def test_push_download(
     repo_path,
     destination_path,
     classifier_pickle,
-    classification_data,
     config_json,
 ):
     client = HfApi()
@@ -187,7 +237,7 @@ def test_push_download(
         requirements=[f'scikit-learn="{version}"'],
         dst=destination_path,
         task="tabular-classification",
-        data=classification_data[0],
+        data=iris.data,
     )
 
     user = client.whoami(token=HF_HUB_TOKEN)["name"]

From a35f8a842f0e2d781a240d0eca76874ffc952275 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Tue, 26 Jul 2022 12:40:39 +0200
Subject: [PATCH 12/16] remove dtype from docstring

---
 skops/hub_utils/_hf_hub.py | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index a0e0fb3a..618a3b8f 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -173,8 +173,9 @@ def _create_config(
             1. Save an example input to the model, which is used by
                HuggingFace's backend and shown in the widget of the model's
                page.
-            2. Store the dtype of the input, which is used by HuggingFace's
-               backend to pass the data with the right dtype to the model.
+            2. Store the columns and their order of the input, which is used by
+               HuggingFace's backend to pass the data in the right form to the
+               model.
 
         The first 3 input values are used as example inputs.
 
@@ -239,10 +240,12 @@ def init(
     data: array-like
         The input to the model. This is used for two purposes:
 
-            1. Save an example input to the model, which is used by HuggingFace's
-                backend and shown in the widget of the model's page.
-            2. Store the dtype of the input, which is used by HugfingFace's backend
-                to pass the data with the right dtype to the model.
+            1. Save an example input to the model, which is used by
+               HuggingFace's backend and shown in the widget of the model's
+               page.
+            2. Store the columns and their order of the input, which is used by
+               HuggingFace's backend to pass the data in the right form to the
+               model.
 
         The first 3 input values are used as example inputs.
 

From 1d63e3542472e5448618a1e69064c14724f979de Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Thu, 28 Jul 2022 15:15:51 +0200
Subject: [PATCH 13/16] fix sphinx import

---
 docs/modules/classes.rst | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/modules/classes.rst b/docs/modules/classes.rst
index 840a9e61..89d1c21b 100644
--- a/docs/modules/classes.rst
+++ b/docs/modules/classes.rst
@@ -9,5 +9,5 @@ This is the class and function reference of skops.
 :mod:`skops.hf_hub`: HuggingFace Hub Integration
 ================================================
 
-.. automodule:: skops.hf_hub
+.. automodule:: skops.hub_utils
     :members:

From cee60ebe477308caeecde654ec8f8e90232a3339 Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Thu, 28 Jul 2022 15:19:31 +0200
Subject: [PATCH 14/16] fix plot_model_card init call

---
 examples/plot_model_card.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/examples/plot_model_card.py b/examples/plot_model_card.py
index 1d504f81..36e8f8a6 100644
--- a/examples/plot_model_card.py
+++ b/examples/plot_model_card.py
@@ -69,7 +69,11 @@
 local_repo = mkdtemp(prefix="skops-")
 
 hub_utils.init(
-    model=pkl_name, requirements=[f"scikit-learn={sklearn.__version__}"], dst=local_repo
+    model=pkl_name,
+    requirements=[f"scikit-learn={sklearn.__version__}"],
+    dst=local_repo,
+    task="tabular-classification",
+    data=X_test,
 )
 
 # %%

From b056cb82ef432850a74498ea8b3da92a2db33fac Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Fri, 29 Jul 2022 08:16:40 +0200
Subject: [PATCH 15/16] make mypy shut up

---
 skops/hub_utils/_hf_hub.py | 2 +-
 skops/utils/fixes.py       | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/skops/hub_utils/_hf_hub.py b/skops/hub_utils/_hf_hub.py
index 520f879e..902ba54f 100644
--- a/skops/hub_utils/_hf_hub.py
+++ b/skops/hub_utils/_hf_hub.py
@@ -8,7 +8,7 @@
 import json
 import shutil
 from pathlib import Path
-from typing import Any, MutableMapping, Union
+from typing import Any, List, MutableMapping, Union
 
 import numpy as np
 from huggingface_hub import HfApi, snapshot_download
diff --git a/skops/utils/fixes.py b/skops/utils/fixes.py
index 24e926d4..e9d83558 100644
--- a/skops/utils/fixes.py
+++ b/skops/utils/fixes.py
@@ -12,10 +12,10 @@
     # older pythons
     import importlib_metadata as metadata  # noqa
 
-try:
+if sys.version_info >= (3, 8):
     # py>=3.8
     from typing import Literal  # noqa
-except ImportError:
+else:
     # older pythons, this requires typing_extensions to be installed.
     # if you're removing this, you should also remove the dependency from
     # _min_dependencies.py

From 2c174c301cad99f96e5b356f9c7f98bd91b2b2ac Mon Sep 17 00:00:00 2001
From: adrinjalali <adrin.jalali@gmail.com>
Date: Fri, 29 Jul 2022 17:59:46 +0200
Subject: [PATCH 16/16] rename test

---
 skops/hub_utils/tests/test_hf_hub.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/skops/hub_utils/tests/test_hf_hub.py b/skops/hub_utils/tests/test_hf_hub.py
index 04d2c6c6..d7736752 100644
--- a/skops/hub_utils/tests/test_hf_hub.py
+++ b/skops/hub_utils/tests/test_hf_hub.py
@@ -172,7 +172,7 @@ def test_create_config(data, task, expected_config):
             )
 
 
-def test_create_config_invalid_data(temp_path):
+def test_create_config_invalid_text_data(temp_path):
     with pytest.raises(ValueError, match="The data needs to be a list of strings."):
         _create_config(
             model_path="model.pkl",