Added lang tabs and filters

app.py

@@ -31,6 +31,9 @@ columns = [
     "CO₂ cost (kg)",
 ]
 
+# languages
+lang_list = ["Spanish", "Galician", "Basque", "Argentinian", "Chilean"]
+
 scheduler = CommitScheduler(
     repo_id="iberbench/ivace-user-request",
     repo_type="dataset",
@@ -83,30 +86,69 @@ def load_data() -> pd.DataFrame:
 
 
 # functions to create filter
-def active_data() -> pd.DataFrame:
+def active_data(lang) -> pd.DataFrame:
     """Change all records as active"""
-    return st.session_state["leaderboard_data"][st.session_state["leaderboard_data"]["Active"] == True].copy()
+    return st.session_state[f"leaderboard_data_{lang}"][
+        st.session_state[f"leaderboard_data_{lang}"]["Active"] == True
+    ].copy()
 
 
-def get_index(row) -> pd.Series:
+def get_index(lang, row) -> pd.Series:
     """Get index of the row"""
-    return active_data().iloc[row].name
+    return active_data(lang).iloc[row].name
 
 
-def commit() -> None:
+def commit(lang) -> None:
     """Commit changes to the session state"""
-    for row in st.session_state.editor["edited_rows"]:
-        row_index = get_index(row)
-        for key, value in st.session_state.editor["edited_rows"][row].items():
-            st.session_state["leaderboard_data"].at[row_index, key] = value
+    for row in st.session_state[f"edited_data_{lang}"]["edited_rows"]:
+        row_index = get_index(lang, row)
+        for key, value in st.session_state[f"edited_data_{lang}"][
+            "edited_rows"
+        ][row].items():
+            st.session_state[f"leaderboard_data_{lang}"].at[
+                row_index, key
+            ] = value
+
+
+def create_search_per_language(lang: str, search_dict: dict):
+    if not st.session_state[f"leaderboard_data_{lang}"].empty:
+        search_dict[lang] = st.text_input(
+            "Search for ...",
+            key=f"search_input_{lang}",
+            on_change=commit,
+            kwargs={"lang": lang},
+        )
+        if search_dict[lang] == "":
+            st.session_state[f"leaderboard_data_{lang}"].Active = True
+        else:
+            st.session_state[f"leaderboard_data_{lang}"].Active = False
+            st.session_state[f"leaderboard_data_{lang}"].loc[
+                st.session_state[f"leaderboard_data_{lang}"][
+                    "Model"
+                ].str.contains(search_dict[lang], case=False),
+                "Active",
+            ] = True
+
+        edited_data = st.data_editor(
+            active_data(lang),
+            column_order=columns,
+            key=f"edited_data_{lang}",
+            hide_index=False,
+            column_config={"Model": st.column_config.LinkColumn("Model")},
+        )
+    else:
+        st.write("No data found to display on leaderboard.")
 
 
 # streamlit UI
-leaderboard_data = load_data()
-if "leaderboard_data" not in st.session_state:
-    st.session_state["leaderboard_data"] = leaderboard_data
+for lang in lang_list:
+    # todo: load a different dataset per language
+    leaderboard_data = load_data()
+    if f"leaderboard_data_{lang}" not in st.session_state:
+        st.session_state[f"leaderboard_data_{lang}"] = leaderboard_data
 
 tabs = st.tabs(["Leaderboard", "Submit model"])
+search_dict = {}
 
 with tabs[0]:
     # logo image
@@ -128,26 +170,16 @@ with tabs[0]:
         unsafe_allow_html=True,
     )
 
-    if not st.session_state["leaderboard_data"].empty:
-        name = st.text_input("Search for ...", on_change=commit)
-        if name == "":
-            st.session_state["leaderboard_data"].Active = True
-        else:
-            st.session_state["leaderboard_data"].Active = False
-            st.session_state["leaderboard_data"].loc[
-                st.session_state["leaderboard_data"]["Model"].str.contains(name, case=False), "Active"
-            ] = True
+    # create tabs per language
+    lang_tabs = st.tabs(lang_list)
+
+    for lang, lt in zip(lang_list, lang_tabs):
+        with lt:
+            create_search_per_language(lang, search_dict)
 
-        edited_data = st.data_editor(
-            active_data(), column_order=columns, key="editor", hide_index=False,
-            column_config={"Model": st.column_config.LinkColumn("Model")},
-        )
-    else:
-        st.write("No data found to display on leaderboard.")
 
 with tabs[1]:
     st.header("Submit model")
-    import streamlit as st
 
     def get_id_number(id_val):
         html_template = f"""
@@ -212,10 +244,9 @@ with tabs[1]:
         )
         model_type = st.selectbox(
             "Choose model type:",
-            help="🟢 Pretrained: Base models trained on text using masked modeling 🟩 Continuously Pretrained: Extended training on additional corpus 🔶 Fine-tuned: Domain-specific optimization 💬 Chat: Models using RLHF, DPO, or IFT for conversation 🤝 Merge: Combined weights without additional training",
+            help="🟢 Pretrained: Base models trained on text using masked modeling 🔶 Fine-tuned: Domain-specific optimization 💬 Chat: Models using RLHF, DPO, or IFT for conversation 🤝 Merge: Combined weights without additional training",
             options=[
                 "🟢 Pretrained",
-                "🟩 Continuously Pretrained",
                 "🔶 Fine-tuned",
                 "💬 Chat",
                 "🤝 Merge",