diff --git a/server/clip_server/executors/clip_onnx.py b/server/clip_server/executors/clip_onnx.py
index 67d2dd29f..b080118d8 100644
--- a/server/clip_server/executors/clip_onnx.py
+++ b/server/clip_server/executors/clip_onnx.py
@@ -5,8 +5,6 @@
 from typing import Optional, Dict
 
 import onnxruntime as ort
-from jina import Executor, requests, DocumentArray
-
 from clip_server.executors.helper import (
     split_img_txt_da,
     preproc_image,
@@ -15,6 +13,7 @@
 )
 from clip_server.model import clip
 from clip_server.model.clip_onnx import CLIPOnnxModel
+from jina import Executor, requests, DocumentArray
 
 
 class CLIPEncoder(Executor):
@@ -103,11 +102,14 @@ async def encode(self, docs: 'DocumentArray', **kwargs):
             ):
                 minibatch.embeddings = self._model.encode_image(minibatch.tensors)
                 # recover original content
-                if _contents:
+                try:
+                    _ = iter(_contents)
                     for _d, _ct in zip(minibatch, _contents):
                         _d.content = _ct
+                except TypeError:
+                    pass
 
-        # for text
+                    # for text
         if _txt_da:
             for minibatch, _contents in _txt_da.map_batch(
                 partial(preproc_text, return_np=True),
@@ -116,9 +118,12 @@ async def encode(self, docs: 'DocumentArray', **kwargs):
             ):
                 minibatch.embeddings = self._model.encode_text(minibatch.tensors)
                 # recover original content
-                if _contents:
+                try:
+                    _ = iter(_contents)
                     for _d, _ct in zip(minibatch, _contents):
                         _d.content = _ct
+                except TypeError:
+                    pass
 
         # drop tensors
         docs.tensors = None
diff --git a/server/clip_server/executors/clip_tensorrt.py b/server/clip_server/executors/clip_tensorrt.py
index f60c87dd6..fa1fc62f9 100644
--- a/server/clip_server/executors/clip_tensorrt.py
+++ b/server/clip_server/executors/clip_tensorrt.py
@@ -3,8 +3,6 @@
 from typing import Dict
 
 import numpy as np
-from jina import Executor, requests, DocumentArray
-
 from clip_server.executors.helper import (
     split_img_txt_da,
     preproc_image,
@@ -13,6 +11,7 @@
 )
 from clip_server.model import clip
 from clip_server.model.clip_trt import CLIPTensorRTModel
+from jina import Executor, requests, DocumentArray
 
 
 class CLIPEncoder(Executor):
@@ -80,11 +79,14 @@ async def encode(self, docs: 'DocumentArray', **kwargs):
                     .astype(np.float32)
                 )
                 # recover original content
-                if _contents:
+                try:
+                    _ = iter(_contents)
                     for _d, _ct in zip(minibatch, _contents):
                         _d.content = _ct
+                except TypeError:
+                    pass
 
-        # for text
+                    # for text
         if _txt_da:
             for minibatch, _contents in _txt_da.map_batch(
                 partial(preproc_text, device=self._device, return_np=False),
@@ -99,11 +101,14 @@ async def encode(self, docs: 'DocumentArray', **kwargs):
                     .astype(np.float32)
                 )
                 # recover original content
-                if _contents:
+                try:
+                    _ = iter(_contents)
                     for _d, _ct in zip(minibatch, _contents):
                         _d.content = _ct
+                except TypeError:
+                    pass
 
-        # drop tensors
+                    # drop tensors
         docs.tensors = None
 
         return docs
diff --git a/server/clip_server/executors/clip_torch.py b/server/clip_server/executors/clip_torch.py
index 2c1e05ad4..e356f448b 100644
--- a/server/clip_server/executors/clip_torch.py
+++ b/server/clip_server/executors/clip_torch.py
@@ -6,8 +6,6 @@
 
 import numpy as np
 import torch
-from jina import Executor, requests, DocumentArray
-
 from clip_server.executors.helper import (
     split_img_txt_da,
     preproc_image,
@@ -15,6 +13,7 @@
     set_rank,
 )
 from clip_server.model import clip
+from jina import Executor, requests, DocumentArray
 
 
 class CLIPEncoder(Executor):
@@ -95,9 +94,12 @@ async def encode(self, docs: 'DocumentArray', **kwargs):
                     )
 
                     # recover original content
-                    if _contents:
+                    try:
+                        _ = iter(_contents)
                         for _d, _ct in zip(minibatch, _contents):
                             _d.content = _ct
+                    except TypeError:
+                        pass
 
             # for text
             if _txt_da:
@@ -114,9 +116,12 @@ async def encode(self, docs: 'DocumentArray', **kwargs):
                     )
 
                     # recover original content
-                    if _contents:
+                    try:
+                        _ = iter(_contents)
                         for _d, _ct in zip(minibatch, _contents):
                             _d.content = _ct
+                    except TypeError:
+                        pass
 
         # drop tensors
         docs.tensors = None