Update huggingfaceModel.py

huggingfaceModel.py

@@ -1,71 +1,71 @@
-import os
-from huggingface_hub import InferenceClient
-
-def generate_dropout_insights(input_values, risk_factors):
-    """
-    Generates structured dropout risk insights using AI, ensuring Unicode-safe text for PDFs.
-    """
-    excluded_fields = {"roll_no", "semester", "degree"}  # Fields to exclude from analysis
-    
-    # Filter only relevant input values
-    filtered_input_values = {key: value for key, value in input_values.items() if key not in excluded_fields}
-    
-    # Map input values to their corresponding risk factors
-    risk_mapping = {key: risk for key, risk in zip(filtered_input_values.keys(), risk_factors)}
-    
-    # Construct message for AI model
-    formatted_input = "\n".join([f"{key}: {value} (Risk: {risk})" for key, (value, risk) in zip(filtered_input_values.keys(), zip(filtered_input_values.values(), risk_factors))])
-    
-    messages = [
-        {
-            "role": "user",
-            "content": f"""
-            Generate a **detailed dropout risk analysis** report based on the following data:
-            
-            {formatted_input}
-            
-            Format the response as:
-            
-            ## **Analysis of Student Dropout Risk**
-            
-            **Overall Risk Level: {{risk_level}}**
-            
-            ### **🟢 Strengths:**
-            List areas where the student is performing well:
-            {{strengths}}
-            
-            ### **🔴 Concerns:**
-            Identify factors that increase dropout risk:
-            {{concerns}}
-            
-            ### **📌 Action Plan for Improvement:**
-            Provide targeted **solutions** based on the student's **weak areas**:
-            {{recommendations}}
-            
-            ## **📝 Final Recommendation:**
-            {{overall_recommendation}}
-            """
-        }
-    ]
-    
-    client = InferenceClient(
-        provider="hyperbolic",
-        api_key=os.getenv("HUGGINGFACE_API_TOKEN")
-    )
-    
-    stream = client.chat.completions.create(
-        model="Qwen/QwQ-32B", 
-        messages=messages, 
-        temperature=0.5,
-        max_tokens=2048,
-        top_p=0.7,
-        stream=True
-    )
-    
-    # Collect and return the generated report
-    report = ""
-    for chunk in stream:
-        report += chunk.choices[0].delta.content
-    
-    return report
-
+import os
+from huggingface_hub import InferenceClient
+
+def generate_dropout_insights(input_values, risk_factors):
+    """
+    Generates structured dropout risk insights using AI, ensuring Unicode-safe text for PDFs.
+    """
+    excluded_fields = {"roll_no", "semester", "degree"}  # Fields to exclude from analysis
+    
+    # Filter only relevant input values
+    filtered_input_values = {key: value for key, value in input_values.items() if key not in excluded_fields}
+    
+    # Map input values to their corresponding risk factors
+    risk_mapping = {key: risk for key, risk in zip(filtered_input_values.keys(), risk_factors)}
+    
+    # Construct message for AI model
+    formatted_input = "\n".join([f"{key}: {value} (Risk: {risk})" for key, (value, risk) in zip(filtered_input_values.keys(), zip(filtered_input_values.values(), risk_factors))])
+    
+    messages = [
+        {
+            "role": "user",
+            "content": f"""
+            Generate a **detailed dropout risk analysis** report based on the following data:
+            
+            {formatted_input}
+            
+            Format the response as:
+            
+            ## **Analysis of Student Dropout Risk**
+            
+            **Overall Risk Level: {{risk_level}}**
+            
+            ### **🟢 Strengths:**
+            List areas where the student is performing well:
+            {{strengths}}
+            
+            ### **🔴 Concerns:**
+            Identify factors that increase dropout risk:
+            {{concerns}}
+            
+            ### **📌 Action Plan for Improvement:**
+            Provide targeted **solutions** based on the student's **weak areas**:
+            {{recommendations}}
+            
+            ## **📝 Final Recommendation:**
+            {{overall_recommendation}}
+            """
+        }
+    ]
+    
+    client = InferenceClient(
+        provider="nebius",
+        api_key=os.getenv("HUGGINGFACE_API_TOKEN")
+    )
+    
+    stream = client.chat.completions.create(
+        model="deepseek-ai/DeepSeek-R1", 
+        messages=messages, 
+        temperature=0.5,
+        max_tokens=2048,
+        top_p=0.7,
+        stream=True
+    )
+    
+    # Collect and return the generated report
+    report = ""
+    for chunk in stream:
+        report += chunk.choices[0].delta.content
+    
+    return report
+