Add system prompt in '/generate' endpoint

app.py

@@ -57,6 +57,7 @@ def check_rate_limit():
 
 # ── Request schema ─────────────────────────────────────────────────────────────
 class GenerateRequest(BaseModel):
+    system_prompt: str = "" # added system prompt
     prompt: str
     model: str = "gpt2"
     max_tokens: int = 50
@@ -86,9 +87,15 @@ def generate(req: GenerateRequest):
     logger.info(f"Request: model={model_id}, max_tokens={max_tokens}, prompt_len={len(req.prompt)}")
 
     try:
+        messages = []
+        # add system prompt to messages if provided
+        if req.system_prompt:
+            messages.append({"role": "system", "content": req.system_prompt})
+        messages.append({"role": "user", "content": req.prompt})
+        
         response = client.chat.completions.create(
             model=model_id,
-            messages=[{"role": "user", "content": req.prompt}],
+            messages=messages,
             max_tokens=max_tokens,
             logprobs=True,
         )