Update server.py

server.py

@@ -178,9 +178,8 @@ def generate_full_response(model, tokenizer, prompt, max_new_tokens=256, temp=0.
             if any(punct in new_text for punct in ['.', '!', '?', '\n']):
                 break
                 
-    # Return the newly generated text (excluding the prompt)
-    full_output = tokenizer.decode(gen_ids)
-    return full_output[len(prompt):].strip()
+    # 🚀 FIXED: Return the full, unbroken string (Prompt + Continuation)
+    return tokenizer.decode(gen_ids)
 
 # ==============================================================================
 # 4. LOAD MODEL & TOKENIZER ON STARTUP