only the current user message

app.py

@@ -7,6 +7,7 @@ from ollama import Client as OllamaClient
 LLM_MODEL = "llama3.2:3b"      # or "mistral:7b", "qwen2.5:3b"
 WHISPER_SIZE = "small"         # "base", "small", "medium"
 USE_SILERO = True              # set False to use Coqui XTTS v2
+USE_CONTEXT = False  # <— new: disable conversational memory
 
 import os
 USE_REMOTE_OLLAMA = bool(os.getenv("OLLAMA_HOST"))
@@ -43,24 +44,18 @@ Reply: <your friendly response to keep the conversation going>"""
 
 def chat_with_llm(history_messages, user_text):
     if USE_REMOTE_OLLAMA:
-        messages = [{"role": "system", "content": SYSTEM_PROMPT}]
-        
-        for m in (history_messages or []):
-            if m.get("role") in ("user", "assistant") and m.get("content"):
-                messages.append({"role": m["role"], "content": m["content"]})
-        messages.append({"role": "user", "content": user_text})
+        # Only system + current user
+        messages = [
+            {"role": "system", "content": SYSTEM_PROMPT},
+            {"role": "user", "content": user_text},
+        ]
         resp = ollama.chat(model=LLM_MODEL, messages=messages)
         return resp["message"]["content"]
     else:
-        # Simple prompt stitching for the fallback pipeline
-        history_text = "\n".join(
-            [f"User: {m['content']}" if m["role"]=="user" else f"Assistant: {m['content']}"
-             for m in (history_messages or [])]
-        )
-        prompt = f"{SYSTEM_PROMPT}\n{history_text}\nUser: {user_text}\nAssistant:"
-        out = gen(prompt)[0]["generated_text"]
-        # Return only the new assistant chunk after the prompt
-        return out.split("Assistant:", 1)[-1].strip()
+        # Only system + current user
+        prompt = f"{SYSTEM_PROMPT}\nUser: {user_text}\nAssistant:"
+        out = gen(prompt, return_full_text=False, max_new_tokens=80, temperature=0.7, repetition_penalty=1.1)[0]["generated_text"].strip()
+        return out