update

app.py

@@ -34,6 +34,8 @@ if torch.cuda.is_available():
         task="text-generation",
         model=model,
         tokenizer=tokenizer,
+        do_sample=True,
+        num_beams=1,
     )
 
 @spaces.GPU
@@ -54,6 +56,7 @@ def generate(
 
     output = my_pipeline(
         messages,
+        temperature=temperature,
     )
     print(output)
     yield output[-1]["generated_text"][-1]["content"]