Add possibility to load existing model card

app.py

@@ -15,6 +15,7 @@ from tempfile import mkdtemp
 import pandas as pd
 import sklearn
 import streamlit as st
+from huggingface_hub import hf_hub_download
 from sklearn.base import BaseEstimator
 
 import skops.io as sio
@@ -31,10 +32,10 @@ PLOT_PREFIX = "__plot__:"
 if "custom_sections" not in st.session_state:
     st.session_state.custom_sections = {}
 
-# the tmp_repo is used to upload the sklearn model to
-tmp_repo = Path(mkdtemp(prefix="skops-"))
-# the hf_repo is the actual repo used for init()
-hf_repo = Path(mkdtemp(prefix="skops-"))
+# the tmp_path is used to upload the sklearn model to
+tmp_path = Path(mkdtemp(prefix="skops-"))
+# the hf_path is the actual repo used for init()
+hf_path = Path(mkdtemp(prefix="skops-"))
 
 # a hacky way to "persist" custom sections
 CUSTOM_SECTIONS_CACHE_FILE = ".custom-sections.json"
@@ -69,16 +70,16 @@ def _write_plot(plot_name, plot_file):
 
 
 def init_repo():
-    _clear_repo(hf_repo)
+    _clear_repo(hf_path)
 
     try:
-        file_name = tmp_repo / "model.skops"
+        file_name = tmp_path / "model.skops"
         sio.dump(model, file_name)
 
         reqs = [r.strip().rstrip(",") for r in requirements.splitlines()]
         hub_utils.init(
             model=file_name,
-            dst=hf_repo,
+            dst=hf_path,
             task=task,
             data=data,
             requirements=reqs,
@@ -120,10 +121,18 @@ def _parse_metrics(metrics):
     return metrics_table
 
 
+def _load_model_card_from_repo(repo_id: str) -> Card:
+    path = hf_hub_download(repo_id, "README.md")
+    return card.parse_modelcard(path)
+
+
 def _create_model_card():
     init_repo()
-    metadata = card.metadata_from_config(hf_repo)
-    model_card = card.Card(model=model, metadata=metadata)
+    if model_card_repo:  # load existing model card
+        model_card = _load_model_card_from_repo(model_card_repo)
+    else:  # create new model card
+        metadata = card.metadata_from_config(hf_path)
+        model_card = card.Card(model=model, metadata=metadata)
 
     if model_description:
         model_card.add(**{"Model description": model_description})
@@ -256,7 +265,7 @@ def add_custom_plot():
 
     # store plot in temp repo
     file_name = plot_file.name.replace(" ", "_")
-    file_path = str(tmp_repo / file_name)
+    file_path = str(tmp_path / file_name)
     with open(file_path, "wb") as f:
         f.write(plot_file.getvalue())
 
@@ -303,28 +312,43 @@ with st.sidebar:
     if model is not None and data is not None:
         init_repo()
 
-    model_description = st.text_input("Model description", value=PLACEHOLDER)
-    intended_uses = st.text_area(
-        "Intended uses & limitations", height=2, value=PLACEHOLDER
-    )
-    metrics = st.text_area("Metrics (e.g. 'accuracy = 0.95'), one metric per line")
-    authors = st.text_area(
-        "Authors",
-        value="This model card is written by following authors:\n\n" + PLACEHOLDER,
-    )
-    contact = st.text_area(
-        "Contact",
-        value="You can contact the model card authors through following channels:\n\n"
-        + PLACEHOLDER,
-    )
-    citation = st.text_area(
-        "Citation",
-        value="Below you can find information related to citation.\n\nBibTex:\n\n```\n"
-        + PLACEHOLDER
-        + "\n```",
-        height=5,
+    model_card_repo = st.text_input(
+        "Optional: HF repo to load model card from (e.g. 'gpt2'), "
+        "leave empty to use default skops template",
+        value="",
     )
 
+    # DEFAULT SKOPS SECTIONS
+    if not model_card_repo:
+        model_description = st.text_input("Model description", value=PLACEHOLDER)
+        intended_uses = st.text_area(
+            "Intended uses & limitations", height=2, value=PLACEHOLDER
+        )
+        metrics = st.text_area("Metrics (e.g. 'accuracy = 0.95'), one metric per line")
+        authors = st.text_area(
+            "Authors",
+            value="This model card is written by following authors:\n\n" + PLACEHOLDER,
+        )
+        contact = st.text_area(
+            "Contact",
+            value="You can contact the model card authors through following channels:\n\n"
+            + PLACEHOLDER,
+        )
+        citation = st.text_area(
+            "Citation",
+            value="Below you can find information related to citation.\n\nBibTex:\n\n```\n"
+            + PLACEHOLDER
+            + "\n```",
+            height=5,
+        )
+    else:
+        model_description = None
+        intended_uses = None
+        metrics = None
+        authors = None
+        contact = None
+        citation = None
+
     # ADD A CUSTOM SECTIONS
     with st.form("custom-section", clear_on_submit=True):
         section_name = st.text_input(

requirements.txt

@@ -1,3 +1,4 @@
+huggingface_hub
 pandas
 scikit-learn
 skops