diff --git a/fastdeploy/model_executor/pre_and_post_process.py b/fastdeploy/model_executor/pre_and_post_process.py
index d2c82e2af..e25e9360f 100644
--- a/fastdeploy/model_executor/pre_and_post_process.py
+++ b/fastdeploy/model_executor/pre_and_post_process.py
@@ -292,8 +292,7 @@ def _build_stream_transfer_data(
                 decoder_state=DecoderState.TEXT, tokens=output_token_per_sample, batch_id=bid
             )
             if logprobs:
-                logprobs = logprobs.slice_rows(bid, bid + 1)
-                stream_transfer_data.logprobs = logprobs
+                stream_transfer_data.logprobs = logprobs.slice_rows(bid, bid + 1)
             if prompt_logprobs_list:
                 stream_transfer_data.prompt_logprobs = prompt_logprobs_list[bid]
             stream_transfer_datas.append(stream_transfer_data)
diff --git a/fastdeploy/worker/output.py b/fastdeploy/worker/output.py
index a674c3606..396fc1980 100644
--- a/fastdeploy/worker/output.py
+++ b/fastdeploy/worker/output.py
@@ -117,11 +117,12 @@ class LogprobsTensors(NamedTuple):
         Slice rows.
         Keeps the number of max_num_logprobs unchanged.
         """
-        return LogprobsTensors(
-            self.logprob_token_ids[start:end],
-            self.logprobs[start:end],
-            self.selected_token_ranks[start:end],
-        )
+        with paddle.no_grad():
+            return LogprobsTensors(
+                paddle.to_tensor(self.logprob_token_ids[start:end], place=self.logprob_token_ids.place),
+                paddle.to_tensor(self.logprobs[start:end], place=self.logprob_token_ids.place),
+                paddle.to_tensor(self.selected_token_ranks[start:end], place=self.logprob_token_ids.place),
+            )
 
 
 @dataclass