Update model_worker.py

model_worker.py

@@ -113,7 +113,7 @@ class ModelWorker:
         max_new_tokens = min(int(params.get("max_new_tokens", 256)), 1024)
         stop_str = params.get("stop", None)
         # do_sample = True if temperature > 0.001 else False
-        do_sample = False
+        do_sample = True
 
         input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to(self.device)
         keywords = [stop_str]
@@ -131,7 +131,7 @@ class ModelWorker:
             inputs=input_ids,
             do_sample=do_sample,
             temperature=temperature,
-            # top_p=top_p,
+            top_p=top_p,
             max_new_tokens=max_new_tokens,
             streamer=streamer,
             stopping_criteria=[stopping_criteria],