Merge branch 'main' of https://huggingface.co/spaces/huggingface/data-measurements-tool-2 into main

data_measurements/streamlit_utils.py

@@ -251,24 +251,27 @@ def expander_text_embeddings(
         else:
             show_node_default = len(node_list) // 2
         st.markdown("---\n")
-        show_node = st.selectbox(
-            f"Choose a leaf node to explore in the{column_id} dataset:",
-            range(len(node_list)),
-            index=show_node_default,
-        )
-        node = node_list[show_node]
-        start_id = st.slider(
-            f"Show closest sentences in cluster to the centroid{column_id} starting at index:",
-            0,
-            len(node["sorted_examples_centroid"]) - 5,
-            value=0,
-            step=5,
-        )
-        for sid, sim in node["sorted_examples_centroid"][start_id : start_id + 5]:
-            # only show the first 4 lines and the first 10000 characters
-            show_text = text_dset[sid][text_field][:10000]
-            show_text = "\n".join(show_text.split("\n")[:4])
-            st.text(f"{sim:.3f} \t {show_text}")
+        if text_dset is None:
+            st.markdown("Missing source text to show, check back later!")
+        else:
+            show_node = st.selectbox(
+                f"Choose a leaf node to explore in the{column_id} dataset:",
+                range(len(node_list)),
+                index=show_node_default,
+            )
+            node = node_list[show_node]
+            start_id = st.slider(
+                f"Show closest sentences in cluster to the centroid{column_id} starting at index:",
+                0,
+                len(node["sorted_examples_centroid"]) - 5,
+                value=0,
+                step=5,
+            )
+            for sid, sim in node["sorted_examples_centroid"][start_id : start_id + 5]:
+                # only show the first 4 lines and the first 10000 characters
+                show_text = text_dset[sid][text_field][:10000]
+                show_text = "\n".join(show_text.split("\n")[:4])
+                st.text(f"{sim:.3f} \t {show_text}")
 
 
 ### Then, show duplicates