fix sampling in chat stream

modeling_qwen.py

@@ -1079,6 +1079,7 @@ class QWenLMHeadModel(QWenPreTrainedModel):
                     return_dict_in_generate=False,
                     generation_config=stream_config,
                     logits_processor=logits_processor,
+                    seed=-1,
                     **kwargs):
                 outputs.append(token.item())
                 yield tokenizer.decode(outputs, skip_special_tokens=True, errors='ignore')