Upload 10 files

scripts/get_scores.py

@@ -93,6 +93,8 @@ def load_negative_rejection_scores(config):
     return df
 
 def load_counterfactual_robustness_scores(config):
+    #hard code noise rate to 0.4
+    config['noise_rate'] = 0.4
     if not os.path.exists(Counterfactual_Robustness_DIR):
         return pd.DataFrame(columns=["Model", "Accuracy with factual docs (%)", "Error Detection Rate", "Correction Rate (%)"])
 
@@ -109,14 +111,14 @@ def load_counterfactual_robustness_scores(config):
             with open(filepath, "r") as f:
                 score = json.load(f)
                 score_data[model] = {
-                    "Accuracy with factual docs (%)": int(score.get("all_rate", 0) * 100),  # No decimal
-                    "Error Detection Rate": int(score.get("reject_rate", 0) * 10),  # Multiply by 10
+                    "Accuracy with factual docs (%)": round(score.get("all_rate", 0) * 100, 2),  # No decimal
+                    "Error Detection Rate (%)": round(score.get("reject_rate", 0) * 100, 2), 
                     "Correction Rate (%)": round(score.get("correct_rate", 0) * 100, 2)  # 2 decimal places
                 }
         else:
             score_data[model] = {  # Populate with "N/A" if file not found
                 "Accuracy with factual docs (%)": "N/A",
-                "Error Detection Rate": "N/A",
+                "Error Detection Rate (%)": "N/A",
                 "Correction Rate (%)": "N/A"
             }
 
@@ -124,8 +126,8 @@ def load_counterfactual_robustness_scores(config):
     df = pd.DataFrame([
         {
             "Model": model,
-            "Accuracy with factual docs (%)": score_data[model]["Accuracy with factual docs (%)"],
-            "Error Detection Rate": score_data[model]["Error Detection Rate"],
+            "Accuracy with factual docs (%)": f"{score_data[model]['Accuracy with factual docs (%)']:.2f}" if score_data[model]["Accuracy with factual docs (%)"] != "N/A" else "N/A",
+            "Error Detection Rate": f"{score_data[model]['Error Detection Rate (%)']:.2f}" if score_data[model]["Error Detection Rate (%)"] != "N/A" else "N/A",
             "Correction Rate (%)": f"{score_data[model]['Correction Rate (%)']:.2f}" if score_data[model]["Correction Rate (%)"] != "N/A" else "N/A"
         }
         for model in config["models"]

scripts/helper.py

@@ -81,7 +81,7 @@ def initialize_logging():
 
 def get_logs():
     """Retrieve logs for display."""
-    return "\n".join(logs[-50:])
+    return "\n".join(logs[-1000:])
 
 def load_used_data(filepath):
         """Loads existing processed data to avoid redundant evaluations."""

scripts/process_data.py

@@ -3,7 +3,6 @@ import random
 import math
 
 def process_data(instance, noise_rate, passage_num, filename, correct_rate=0):
-    print(filename)
     """Process the data for generating a noisy document set."""
     query = instance['query']
     ans = instance['answer']