Update app.py

def generate_answer(question, context):
if context:
prompt = (
"You are a helpful assistant answering questions strictly from the provided context.\n"
"If the answer is not in the context, say you don't know.\n\n"
f"Context:\n{context}\n\n"
f"Question: {question}\nAnswer:"
)
else:
prompt = (
"You are a helpful assistant. The user asked a question but no document context is available.\n\n"
f"Question: {question}\nAnswer:"
)

response = llm.post(
"/generate",
json={
"model": GRANITE_MODEL,
"inputs": prompt,
"parameters": {
"max_new_tokens": 512,
"temperature": 0.2,
"top_p": 0.9,
"repetition_penalty": 1.05,
},
},
)

return response["generated_text"].strip()

app.py

@@ -87,7 +87,6 @@ def retrieve_context(question, state, k=4):
 # ============================================================
 
 def generate_answer(question, context):
-    """Generate an answer using IBM Granite."""
     if context:
         prompt = (
             "You are a helpful assistant answering questions strictly from the provided context.\n"
@@ -101,15 +100,21 @@ def generate_answer(question, context):
             f"Question: {question}\nAnswer:"
         )
 
-    response = llm.text_generation(
-        prompt=prompt,
-        max_new_tokens=512,
-        temperature=0.2,
-        top_p=0.9,
-        repetition_penalty=1.05,
+    response = llm.post(
+        "/generate",
+        json={
+            "model": GRANITE_MODEL,
+            "inputs": prompt,
+            "parameters": {
+                "max_new_tokens": 512,
+                "temperature": 0.2,
+                "top_p": 0.9,
+                "repetition_penalty": 1.05,
+            },
+        },
     )
 
-    return response.strip()
+    return response["generated_text"].strip()
 
 
 # ============================================================