Oracle weight assigning update

Oracle/deepfundingoracle.py

@@ -26,6 +26,7 @@ from tqdm import tqdm
 import sys
 import re
 import json
+import time
 
 from sklearn.model_selection import train_test_split, RandomizedSearchCV
 from sklearn.ensemble import RandomForestRegressor
@@ -234,7 +235,7 @@ def timeout_handler(signum, frame):
 
 
 
-def assign_base_weight(df, max_workers=32):
+def assign_base_weight(df, max_workers=32, llm_retries=3,llm_delay=2):
     """
     Assign base weights using a single LLM call to determine feature weights,
     and programmatically calculate repository weights.
@@ -258,25 +259,30 @@ def assign_base_weight(df, max_workers=32):
         "the influence of a repository. Provide the weights as a JSON object with "
         "keys as feature names and values as their weights. Ensure the response is strictly in valid JSON format."
     )
-    try:
-        response = llama.predict(prompt)
-        if not response.strip():
-            raise ValueError("Empty response from LLM.")
-        feature_weights = json.loads(response)  # Safely parse JSON
-        print(f"[INFO] Feature weights from LLM: {feature_weights}", flush=True)
-    except (json.JSONDecodeError, Exception) as e:
-        print(f"[ERROR] Failed to fetch feature weights from LLM: {e}", flush=True)
-        logging.error(f"[ERROR] Failed to fetch feature weights from LLM: {e}")
-        # Fallback to default weights
-        feature_weights = {
-            "stars": 0.3,
-            "forks": 0.2,
-            "watchers": 0.2,
-            "open_issues": 0.1,
-            "pulls": 0.1,
-            "activity": 0.05,
-            "contributors": 0.05
-        }
+    fearure_weights= None
+    for attempt in range(llm_retries):
+        try:
+            response = llama.predict(prompt)
+            if not response or response.strip():
+                raise ValueError("Empty response from LLM.")
+            feature_weights = json.loads(response)  # Safely parse JSON
+            print(f"[INFO] Feature weights from LLM: {feature_weights}", flush=True)
+            break
+        except Exception as e:
+            print(f"[ERROR] LLM attempt {attempt+1} failed: {e}", flush=True)
+            logging.error(f"[ERROR] LLM attempt {attempt+1} failed: {e}")
+            time.sleep(llm_delay)
+            # Fallback to default weights
+        if fearure_weights is None:
+            feature_weights = {
+                "stars": 0.3,
+                "forks": 0.2,
+                "watchers": 0.2,
+                "open_issues": 0.1,
+                "pulls": 0.1,
+                "activity": 0.05,
+                "contributors": 0.05
+            }
         print(f"[INFO] Using default feature weights: {feature_weights}", flush=True)
 
     # Step 2: Ensure all feature columns are numeric