Fix another bug in the results handling

app.py

@@ -41,7 +41,7 @@ with tab1:
         )["train"]
 
     except Exception as e:
-        st.error(f"An error occurred while loading the predictions: {e}")
+        st.info(f"Error in loading the results:\n{e}")
         model_predictions_rows = []
 
     if model_predictions_rows:
@@ -97,9 +97,12 @@ with tab1:
                 f"https://huggingface.co/{row['model_name']}"
                 for row in model_predictions_rows if row["status"] == "completed"
             ]
+            results_df["Commit ID"] = [
+                row["commit_id"] for row in model_predictions_rows if row["status"] == "completed"
+            ]
 
             results_df = results_df[
-                ["Rank", "Model Name", "Accuracy", "Recall", "Precision", "F1 score", "URL"]
+                ["Rank", "Model Name", "Accuracy", "Recall", "Precision", "F1 score", "URL", "Commit ID"]
             ]
             st.data_editor(
                 results_df,

eval_utils.py

@@ -16,6 +16,8 @@ def predict_top_p(model, tokenizer, text, P=0.9):
     predictions = [0 for _ in range(18)]
     total_prob = 0
 
+    # TODO: Assert that the list length is just 18
+
     for i in range(18):
         total_prob += probabilities[topk_predictions[i]]
         predictions[topk_predictions[i]] = 1

utils.py

@@ -21,7 +21,7 @@ def update_model_queue(repo_id, model_name, commit_id, inference_function, statu
 
     predictions_object = {
         "model_name": model_name,
-        "predictions": [],
+        "predictions": [[""]],
         "commit_id": commit_id,
         "inference_function": inference_function,
         "last_updated_timestamp": timestamp,