feat: return output_scores as well

app.py

@@ -575,6 +575,7 @@ def generate(
     top_p = float(top_p)
     generate_kwargs = dict(
         max_new_tokens=max_new_tokens,
+        output_scores=True,
         temperature=None if temperature==1 else temperature,
         top_p=None if top_p==0 else top_p,
         repetition_penalty=None if repetition_penalty==1 else repetition_penalty,