Refactored Multi-Agent XAI Demo to hide prompts, limit response length, and enable a natural conversational flow between the Engineer and Analyst. Finalized with a cohesive plan summarizing the dialogue.

app.py

@@ -36,15 +36,10 @@ tokenizerA, modelA = load_model_analyst()
 #                     ENGINEER / ANALYST GENERATION
 ##############################################################################
 
-def generate_engineer_response(user_text, tokenizer, model):
+def generate_response(prompt, tokenizer, model, max_sentences=2):
     """
-    Generate a concise technical response from the Engineer based on the user's input.
+    Generate a concise response based on the provided prompt.
     """
-    prompt = f"""
-User: {user_text}
-
-Engineer: Provide a technical solution or approach that directly addresses the user's problem.
-"""
     inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True)
     outputs = model.generate(
         inputs["input_ids"],
@@ -57,43 +52,19 @@ Engineer: Provide a technical solution or approach that directly addresses the u
         no_repeat_ngram_size=4,
         pad_token_id=tokenizer.pad_token_id
     )
-    return tokenizer.decode(outputs[0], skip_special_tokens=True).strip()
-
-def generate_analyst_response(engineer_output, tokenizer, model):
-    """
-    Generate a conversational response from the Analyst based on the Engineer's response.
-    """
-    prompt = f"""
-Engineer: {engineer_output}
-
-Analyst: Respond with actionable insights or complementary data-driven recommendations.
-"""
-    inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True)
-    outputs = model.generate(
-        inputs["input_ids"],
-        attention_mask=inputs["attention_mask"],
-        max_new_tokens=50,  # Restrict length
-        temperature=0.6,
-        do_sample=True,
-        top_p=0.8,
-        repetition_penalty=2.0,
-        no_repeat_ngram_size=4,
-        pad_token_id=tokenizer.pad_token_id
-    )
-    return tokenizer.decode(outputs[0], skip_special_tokens=True).strip()
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True).strip()
+    # Limit to max_sentences by splitting and rejoining
+    return " ".join(response.split(".")[:max_sentences]) + "."
 
 def summarize_conversation(conversation):
     """
     Summarize the entire conversation to produce a cohesive and actionable plan.
     """
     summary = "### Final Plan\n"
-    engineer_response = next((text for speaker, text in conversation if speaker == "Engineer"), "")
-    analyst_response = next((text for speaker, text in conversation if speaker == "Analyst"), "")
-
-    summary += f"- **Engineer Perspective:**\n  {engineer_response}\n\n"
-    summary += f"- **Analyst Perspective:**\n  {analyst_response}\n\n"
-    summary += "This collaborative plan ensures a balance of technical and analytical insights."
-
+    for speaker, text in conversation:
+        if speaker == "Engineer" or speaker == "Analyst":
+            summary += f"- **{speaker}:** {text}\n"
+    summary += "\nThis collaborative plan integrates technical and analytical insights."
     return summary
 
 ##############################################################################
@@ -115,29 +86,34 @@ if st.button("Generate Responses"):
         user_text = st.session_state.user_input
         st.session_state.conversation = [("User", user_text)]  # Clear and restart conversation
 
-        # Engineer generates a response
-        with st.spinner("Engineer is formulating a solution..."):
-            engineer_resp = generate_engineer_response(
-                user_text=user_text,
-                tokenizer=tokenizerE,
-                model=modelE
-            )
-            st.session_state.conversation.append(("Engineer", engineer_resp))
-
-        # Display Engineer response immediately
-        st.markdown(f"### Engineer Response\n{engineer_resp}")
-
-        # Analyst generates a response based on engineer's output
-        with st.spinner("Analyst is analyzing data and providing insights..."):
-            analyst_resp = generate_analyst_response(
-                engineer_output=engineer_resp,
-                tokenizer=tokenizerA,
-                model=modelA
-            )
-            st.session_state.conversation.append(("Analyst", analyst_resp))
-
-        # Display Analyst response immediately
-        st.markdown(f"### Analyst Response\n{analyst_resp}")
+        engineer_prompt = f"User: {user_text}\nEngineer: Provide a technical solution or approach that directly addresses the user's problem."
+        analyst_prompt = ""
+
+        for turn in range(3):
+            # Engineer generates a response
+            with st.spinner(f"Engineer is formulating response {turn + 1}..."):
+                engineer_resp = generate_response(
+                    prompt=engineer_prompt,
+                    tokenizer=tokenizerE,
+                    model=modelE
+                )
+                st.session_state.conversation.append(("Engineer", engineer_resp))
+
+            # Display Engineer response
+            st.markdown(f"### Engineer Response ({turn + 1})\n{engineer_resp}")
+
+            # Analyst generates a response based on engineer's output
+            analyst_prompt = f"Engineer: {engineer_resp}\nAnalyst: Respond with actionable insights or complementary data-driven recommendations."
+            with st.spinner(f"Analyst is formulating response {turn + 1}..."):
+                analyst_resp = generate_response(
+                    prompt=analyst_prompt,
+                    tokenizer=tokenizerA,
+                    model=modelA
+                )
+                st.session_state.conversation.append(("Analyst", analyst_resp))
+
+            # Display Analyst response
+            st.markdown(f"### Analyst Response ({turn + 1})\n{analyst_resp}")
 
         # Summarize the final plan
         with st.spinner("Generating the final plan..."):