Update model.py

model.py

@@ -33,12 +33,15 @@ def call_ai_model_for_insights(input_data: Dict, delay_risk: float) -> List[str]
     """
     model_name = "t5-small"
     max_retries = 3
-    retry_delay = 20  # Increased for network stability
+    retry_delay = 30  # Increased for network stability
     
     # Log system resources if psutil is available
     if psutil:
-        memory = psutil.virtual_memory()
-        logger.info(f"System memory - Total: {memory.total / 1e9:.2f} GB, Available: {memory.available / 1e9:.2f} GB, Used: {memory.percent}%")
+        try:
+            memory = psutil.virtual_memory()
+            logger.info(f"System memory - Total: {memory.total / 1e9:.2f} GB, Available: {memory.available / 1e9:.2f} GB, Used: {memory.percent}%")
+        except Exception as e:
+            logger.warning(f"Failed to log system memory: {str(e)}")
     else:
         logger.warning("psutil not available; cannot log system memory usage")
     
@@ -56,24 +59,17 @@ def call_ai_model_for_insights(input_data: Dict, delay_risk: float) -> List[str]
             
             logger.info("Model loaded successfully. Generating insights...")
             prompt = f"""
-            Summarize delay risk data into 2-3 concise insights:
-            Project: {input_data.get('project_name', 'Unnamed')}
-            Phase: {input_data.get('phase', '')}
-            Task: {input_data.get('task', '')}
-            Progress: {input_data.get('current_progress', 0)}%
-            Workforce Gap: {input_data.get('workforce_gap', 0)}%
-            Skill: {input_data.get('workforce_skill_level', '').lower()}
-            Weather Score: {input_data.get('weather_impact_score', 0)}
+            Summarize risk:
             Risk: {delay_risk:.1f}%
 
-            Format as a list, e.g., ["Insight 1", "Insight 2"].
+            Format: ["Insight 1"].
             """
             
             with torch.no_grad():
-                inputs = tokenizer(prompt, return_tensors="pt", max_length=32, truncation=True).to("cpu")
+                inputs = tokenizer(prompt, return_tensors="pt", max_length=8, truncation=True).to("cpu")
                 outputs = model.generate(
                     **inputs,
-                    max_new_tokens=15,
+                    max_new_tokens=5,
                     num_beams=1,
                     temperature=0.7,
                     do_sample=True
@@ -82,7 +78,7 @@ def call_ai_model_for_insights(input_data: Dict, delay_risk: float) -> List[str]
             
             insights = [line.strip() for line in response.split("\n") if line.strip() and line.strip() not in [prompt]]
             logger.info(f"Generated insights: {insights}")
-            return insights[:3] or ["No insights generated; review input data."]
+            return insights[:1] or ["No insights generated."]
         except Exception as e:
             logger.error(f"Attempt {attempt + 1}/{max_retries} - Model inference failed: {str(e)}")
             if attempt < max_retries - 1:
@@ -92,13 +88,9 @@ def call_ai_model_for_insights(input_data: Dict, delay_risk: float) -> List[str]
                 logger.error("Max retries reached. Using fallback insights.")
                 fallback_insights = []
                 if delay_risk > 75:
-                    fallback_insights.append("High risk; allocate resources urgently.")
+                    fallback_insights.append("High risk detected.")
                 elif delay_risk > 50:
-                    fallback_insights.append("Moderate risk; extend shift hours.")
-                if input_data.get('workforce_gap', 0) > 20:
-                    fallback_insights.append("Workforce gap; recruit workers.")
-                if input_data.get('weather_impact_score', 0) > 50:
-                    fallback_insights.append("Adverse weather; prioritize indoor tasks.")
+                    fallback_insights.append("Moderate risk found.")
                 return fallback_insights or ["AI model failed to generate insights; check system resources."]
 
 def predict_delay(input_data: Dict) -> Dict: