Update app.py

app.py

@@ -116,50 +116,47 @@ def chat_with_musician(user_input, history, artist):
         history = []
     previous_artist = artist_history[-1]
     if artist != previous_artist:
-        # Reset history if the artist changes
         history.clear()
+
+    # Convert Gradio history tuples to HF message dicts
     messages = []
+    for user_msg, bot_msg in history[-5:]:  # last 5 exchanges
+        messages.append({"role": "user", "content": user_msg})
+        messages.append({"role": "assistant", "content": bot_msg})
 
-    for dict in history[-5*2:]:  # Keep only the last 5 exchanges, #I think this must be even to ensure that the last user message is included
-        messages.append(dict)
-    # Add the latest user message
+    # Add current user message
     messages.append({"role": "user", "content": system_message + "\n\n" + user_input})
+
     try:
         response = client.chat_completion(
             messages=messages,
-            #model="google/gemma-2-2b-it",
-            #model="openai/gpt-oss-120b",
-            #test without specifying model
             max_tokens=256,
             temperature=0.75,
-            #top_p=0.9
         )
-        gemma_response= response["choices"][0]["message"]["content"]
+        gemma_response = response["choices"][0]["message"]["content"]
     except Exception as e:
-        return f"Error: {str(e)}"
-    
+        gemma_response = f"Error: {str(e)}"
+
     lyric_response = artist_response(gemma_response, artist)
-    if len(messages)>1:
-        if lyric_response == messages[-2]["content"]:
-            # If the response is the same as the last one, get a new one:
-            messages[-1]={"role": "user", "content": system_message_repeated + "\n\n" + user_input}
-            try:
-                response = client.chat_completion(
-                    messages=messages,
-                    #model="google/gemma-2-2b-it",
-                    max_tokens=256,
-                    temperature=0.75,
-                    #top_p=0.9
-                )
-                gemma_response= response["choices"][0]["message"]["content"]
-            except Exception as e:
-                return f"Error: {str(e)}"
+
+    # Check for repeated response logic (optional)
+    if len(messages) > 1 and lyric_response == messages[-2]["content"]:
+        messages[-1] = {"role": "user", "content": system_message_repeated + "\n\n" + user_input}
+        try:
+            response = client.chat_completion(
+                messages=messages,
+                max_tokens=256,
+                temperature=0.75,
+            )
+            gemma_response = response["choices"][0]["message"]["content"]
+        except Exception as e:
+            gemma_response = f"Error: {str(e)}"
         lyric_response = artist_response(gemma_response, artist)
 
-    history.append({"role": "user", "content": user_input})
-    history.append({"role": "assistant", "content": lyric_response})
-    artist_history.append(artist)  # Store the previous artist for the next call
-    artist_history[:] = artist_history[-10:]  # Keep only last 10 entries
+    # Append new exchange to Gradio history format
+    history.append((user_input, lyric_response))
+    artist_history.append(artist)
+    artist_history[:] = artist_history[-10:]
     return lyric_response
 
 def cosine_similarity_int8(query, embeddings):