diff --git a/paddle/scripts/conda_build.py b/paddle/scripts/conda_build.py
index 12004460a50360f0b62ef9e182a092e1f16c76ce..98692ff3df39c91a16e0d25ba564a5629ff535b1 100644
--- a/paddle/scripts/conda_build.py
+++ b/paddle/scripts/conda_build.py
@@ -55,7 +55,6 @@ requirements:
     - protobuf>=3.1.0
     - gast==0.3.3
     - Pillow
-    - six
     - decorator
     - astor
 """
@@ -67,7 +66,6 @@ requirements:
     - protobuf>=3.1.0
     - gast==0.3.3
     - Pillow
-    - six
     - decorator
     - astor
 """
diff --git a/python/paddle/distributed/fleet/elastic/manager.py b/python/paddle/distributed/fleet/elastic/manager.py
index 9397f6d10d4a693a0874dab3ecb977d4477dc534..b23a11642259073245148a198fe99ce63484ef2d 100644
--- a/python/paddle/distributed/fleet/elastic/manager.py
+++ b/python/paddle/distributed/fleet/elastic/manager.py
@@ -15,7 +15,6 @@
 import time
 import socket
 import os
-import six
 import copy
 import signal
 import random
@@ -244,8 +243,7 @@ class ElasticManager(object):
         # register callback
         def host_call_back(event):
             self.hosts = [
-                six.ensure_str(i[0])
-                for i in self.etcd.get_prefix(self.node_prefix)
+                i[0].decode() for i in self.etcd.get_prefix(self.node_prefix)
             ]
             self.hosts = list(set(self.hosts)) if self.hosts else self.hosts
             logger.info(
@@ -266,7 +264,7 @@ class ElasticManager(object):
                     host_lease.refresh()
 
                     hosts = [
-                        six.ensure_str(i[0])
+                        i[0].decode()
                         for i in self.etcd.get_prefix(self.node_prefix)
                     ]
                     hosts = list(set(hosts)) if hosts else hosts
@@ -311,7 +309,8 @@ class ElasticManager(object):
         def endpoints_call_back(event):
             if not self.dist_endpoints:
                 return
-            edps = six.ensure_str(self.etcd.get(self.endpoints_path)[0] or '')
+            value = self.etcd.get(self.endpoints_path)[0]
+            edps = value.decode() if value is not None else ''
             self.dist_endpoints, self.trainers = edps.split('|')
             logger.info(
                 "set DISTRIBUTED_TRAINER_ENDPOINTS {} ".format(
@@ -426,8 +425,7 @@ class ElasticManager(object):
             self.hosts = host_list
         else:
             self.hosts = [
-                six.ensure_str(i[0])
-                for i in self.etcd.get_prefix(self.node_prefix)
+                i[0].decode() for i in self.etcd.get_prefix(self.node_prefix)
             ]
         self.hosts = list(set(self.hosts)) if self.hosts else self.hosts
 
diff --git a/python/paddle/distributed/launch/controllers/master.py b/python/paddle/distributed/launch/controllers/master.py
index 128852e092cfbe3347cba7a613a1b32aadbf7964..9c3f0a8501a3ac2334452ad5152bb58986ecfa33 100644
--- a/python/paddle/distributed/launch/controllers/master.py
+++ b/python/paddle/distributed/launch/controllers/master.py
@@ -17,7 +17,6 @@ from paddle.distributed.launch.utils.kv_server import KVServer
 
 import time
 import sys
-import six
 import threading
 import copy
 import random
@@ -214,22 +213,22 @@ class ETCDMaster(Master):
 
             if len(result) == size:
                 if rank < 0:
-                    keys = [six.ensure_str(i[1].key) for i in result]
-                    sorted_keys = [six.ensure_str(i[1].key) for i in result]
+                    keys = [i[1].key.decode() for i in result]
+                    sorted_keys = [i[1].key.decode() for i in result]
                     sorted_keys.sort()
-                    values = [six.ensure_str(i[0]) for i in result]
+                    values = [i[0].decode() for i in result]
                     ret = [values[keys.index(k)] for k in sorted_keys]
                     idx = ret.index(value)
                     return ret, idx
                 else:
                     ret = [None] * size
                     for v, k in result:
-                        ii = int(six.ensure_str(k.key).split('/')[-1])
+                        ii = int(k.key.decode().split('/')[-1])
                         if ii < 0:
                             self.ctx.logger.error(
                                 "rank {} error in sync".format(ii)
                             )
-                        ret[ii] = six.ensure_str(v)
+                        ret[ii] = v.decode()
                     return ret, rank
             else:
                 time.sleep(0.5)
@@ -278,8 +277,7 @@ class ETCDMaster(Master):
 
     def fetch_peer_alive(self):
         peer_alive = [
-            six.ensure_str(i[0])
-            for i in self.client.get_prefix(self.heartbeat_prefix)
+            i[0].decode() for i in self.client.get_prefix(self.heartbeat_prefix)
         ]
         self.ctx.logger.debug("peer alive {}".format(peer_alive))
         return peer_alive
@@ -319,7 +317,8 @@ class ETCDMaster(Master):
         ), "set status failed {}".format(status)
 
     def get_status(self):
-        return six.ensure_str(self.client.get(self.job_prefix)[0] or '')
+        value = self.client.get(self.job_prefix)[0]
+        return value.decode() if value is not None else ''
 
     def stop(self):
         if hasattr(self, 'beat_thread'):
diff --git a/python/paddle/fluid/contrib/slim/quantization/imperative/ptq_quantizer.py b/python/paddle/fluid/contrib/slim/quantization/imperative/ptq_quantizer.py
index ba881f88efc39b0fa433d0ef48cfc6da77358f4c..3dfc95a8ac34600f7cd976b13071ae475ec44177 100644
--- a/python/paddle/fluid/contrib/slim/quantization/imperative/ptq_quantizer.py
+++ b/python/paddle/fluid/contrib/slim/quantization/imperative/ptq_quantizer.py
@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import six
 import abc
 import copy
 import math
diff --git a/python/paddle/fluid/dataloader/dataloader_iter.py b/python/paddle/fluid/dataloader/dataloader_iter.py
index 83d95c479250ef6548e45e25daa1fd4aa81977c7..aa5d71293fdabe2ed2ce6dd6765a9542457f906c 100644
--- a/python/paddle/fluid/dataloader/dataloader_iter.py
+++ b/python/paddle/fluid/dataloader/dataloader_iter.py
@@ -13,7 +13,6 @@
 # limitations under the License.
 
 import os
-import six
 import sys
 import time
 import signal
@@ -284,9 +283,9 @@ class _DataLoaderIterSingleProcess(_DataLoaderIterBase):
                 except:
                     self._exit_thread_expectedly()
 
-            except:
+            except Exception as e:
                 self._exit_thread_unexpectedly()
-                six.reraise(*sys.exc_info())
+                raise e
 
         self._exit_thread_expectedly()
 
@@ -334,7 +333,7 @@ class _DataLoaderIterSingleProcess(_DataLoaderIterBase):
         except StopIteration:
             self._reader.shutdown()
             self._try_shutdown_all()
-            six.reraise(*sys.exc_info())
+            raise
         finally:
             if in_profiler_mode():
                 trace_event.end()
@@ -629,7 +628,7 @@ class _DataLoaderIterMultiProcess(_DataLoaderIterBase):
                             self._blocking_queue.close()
                     except Exception as e:
                         self._exit_thread_unexpectedly()
-                        six.reraise(*sys.exc_info())
+                        raise e
                     finally:
                         self._rcvd_idx += 1
 
@@ -715,7 +714,7 @@ class _DataLoaderIterMultiProcess(_DataLoaderIterBase):
                     "DataLoader reader thread failed({}) to read data from "
                     "workers' result queue.".format(e)
                 )
-                six.reraise(*sys.exc_info())
+                raise e
             else:
                 if self._dataset_kind == _DatasetKind.ITER and isinstance(
                     data, _IterableDatasetStopIteration
@@ -850,7 +849,7 @@ class _DataLoaderIterMultiProcess(_DataLoaderIterBase):
             if not self._persistent_workers:
                 self._reader.shutdown()
                 self._try_shutdown_all()
-            six.reraise(*sys.exc_info())
+            raise
         finally:
             if in_profiler_mode():
                 trace_event.end()
diff --git a/python/paddle/fluid/dataloader/worker.py b/python/paddle/fluid/dataloader/worker.py
index f0aa32e774522eac2d6c31b30cf3eb03c1055201..fef07df92751ad172c9592891e901d9b427abb66 100644
--- a/python/paddle/fluid/dataloader/worker.py
+++ b/python/paddle/fluid/dataloader/worker.py
@@ -13,7 +13,6 @@
 # limitations under the License.
 
 import os
-import six
 import sys
 import paddle
 import numpy as np
@@ -395,7 +394,7 @@ def _worker_loop(
         # NOTE: Main process will raise KeyboardInterrupt anyways, ignore it in child process
         pass
     except:
-        six.reraise(*sys.exc_info())
+        raise
     finally:
         if use_shared_memory:
             _cleanup_mmap()
diff --git a/python/paddle/fluid/dygraph/dygraph_to_static/convert_call_func.py b/python/paddle/fluid/dygraph/dygraph_to_static/convert_call_func.py
index 5adf810eef4a388e3b667ac416f94d52ea72483d..b43c4e9d36d60190b551d0185993d98af4cbda9a 100644
--- a/python/paddle/fluid/dygraph/dygraph_to_static/convert_call_func.py
+++ b/python/paddle/fluid/dygraph/dygraph_to_static/convert_call_func.py
@@ -22,7 +22,6 @@ import re
 import types
 
 import numpy
-import six
 import builtins
 
 from paddle.fluid.dygraph.container import Sequential
@@ -58,7 +57,6 @@ BUILTIN_LIKELY_MODULES = [
     copy,
     inspect,
     re,
-    six,
     numpy,
     logging,
 ]
diff --git a/python/paddle/fluid/executor.py b/python/paddle/fluid/executor.py
index ba8221bb2f0fb2cca44f2e178b245072c09d4d75..55a0334c8e933808c053da1147340301dcad29e6 100755
--- a/python/paddle/fluid/executor.py
+++ b/python/paddle/fluid/executor.py
@@ -19,7 +19,6 @@ import sys
 import warnings
 import numpy as np
 from .wrapped_decorator import signature_safe_contextmanager
-import six
 from .data_feeder import convert_dtype
 from .framework import Program, default_main_program, Variable, Operator
 from .framework import convert_np_dtype_to_dtype_, _apply_pass
@@ -1574,23 +1573,20 @@ class Executor(object):
             ]
             self._log_force_set_program_cache(use_program_cache)
 
-        try:
-            res = self._run_impl(
-                program=program,
-                feed=feed,
-                fetch_list=fetch_list,
-                feed_var_name=feed_var_name,
-                fetch_var_name=fetch_var_name,
-                scope=scope,
-                return_numpy=return_numpy,
-                use_program_cache=use_program_cache,
-                use_prune=use_prune,
-                return_merged=return_merged,
-            )
-            core.update_autotune_status()
-            return res
-        except Exception as e:
-            six.reraise(*sys.exc_info())
+        res = self._run_impl(
+            program=program,
+            feed=feed,
+            fetch_list=fetch_list,
+            feed_var_name=feed_var_name,
+            fetch_var_name=fetch_var_name,
+            scope=scope,
+            return_numpy=return_numpy,
+            use_program_cache=use_program_cache,
+            use_prune=use_prune,
+            return_merged=return_merged,
+        )
+        core.update_autotune_status()
+        return res
 
     def _run_impl(
         self,
diff --git a/python/paddle/fluid/framework.py b/python/paddle/fluid/framework.py
index 4d8e356d63d8c482da3523f5a75205d48dac0e01..188ff9a8ea8d711644230fbe3381caa130c913c8 100644
--- a/python/paddle/fluid/framework.py
+++ b/python/paddle/fluid/framework.py
@@ -5777,10 +5777,10 @@ class Program(object):
 
             .. code-block:: python
 
-                import six
+                import paddle
 
                 def print_prog(prog):
-                    for name, value in sorted(six.iteritems(prog.block(0).vars)):
+                    for name, value in sorted(prog.block(0).vars.items()):
                         print(value)
                     for op in prog.block(0).ops:
                         print("op type is {}".format(op.type))
diff --git a/python/paddle/fluid/incubate/fleet/parameter_server/ir/public.py b/python/paddle/fluid/incubate/fleet/parameter_server/ir/public.py
index 9b762bb9ee7f865f842f2d78327d0cf2e2252b1d..865ffd2e0fd0c4c1184b56b844c4afbb25292d6c 100755
--- a/python/paddle/fluid/incubate/fleet/parameter_server/ir/public.py
+++ b/python/paddle/fluid/incubate/fleet/parameter_server/ir/public.py
@@ -19,7 +19,6 @@ import math
 import os
 import warnings
 import logging
-import six
 import paddle.fluid as fluid
 from paddle.fluid import core
 from paddle.fluid.core import CommContext
diff --git a/python/paddle/fluid/layers/io.py b/python/paddle/fluid/layers/io.py
index 86654bbf669587c71177af74d4b538b4f54acf31..758cd94c214211c325b4590e01e91cb4d99d4747 100644
--- a/python/paddle/fluid/layers/io.py
+++ b/python/paddle/fluid/layers/io.py
@@ -14,7 +14,6 @@
 
 import multiprocessing
 import os
-import six
 import sys
 import threading
 
@@ -523,10 +522,10 @@ def _py_reader(
                     if reader.exited:
                         break
                 feed_queue.close()
-            except Exception as ex:
+            except Exception as e:
                 feed_queue.kill()
                 logging.warn('Your decorated reader has raised an exception!')
-                six.reraise(*sys.exc_info())
+                raise e
 
         reader.thread = threading.Thread(
             target=__provider_thread__, args=(_current_expected_place(),)
diff --git a/python/paddle/fluid/layers/nn.py b/python/paddle/fluid/layers/nn.py
index 17bd7ff86dbf93ee2ae5a82236175cb17b846059..69c00b5dbb1f2816525aa7fdcb99a97e727d8c77 100755
--- a/python/paddle/fluid/layers/nn.py
+++ b/python/paddle/fluid/layers/nn.py
@@ -14998,7 +14998,6 @@ def py_func(func, x, out, backward_func=None, skip_vars_in_backward_input=None):
 
             # example 1:
             import paddle
-            import six
             import numpy as np
 
             paddle.enable_static()
@@ -15024,7 +15023,7 @@ def py_func(func, x, out, backward_func=None, skip_vars_in_backward_input=None):
 
             def simple_net(img, label):
                 hidden = img
-                for idx in six.moves.range(4):
+                for idx in range(4):
                     hidden = paddle.static.nn.fc(hidden, size=200)
                     new_hidden = create_tmp_var(name='hidden_{}'.format(idx),
                         dtype=hidden.dtype, shape=hidden.shape)
@@ -15042,13 +15041,13 @@ def py_func(func, x, out, backward_func=None, skip_vars_in_backward_input=None):
                 return ce_loss(prediction, label)
 
             x = paddle.static.data(name='x', shape=[1,4], dtype='float32')
-            y = paddle.static.data(name='y', shape=[1,10], dtype='int64')
+            y = paddle.static.data(name='y', shape=[1], dtype='int64')
             res = simple_net(x, y)
 
             exe = paddle.static.Executor(paddle.CPUPlace())
             exe.run(paddle.static.default_startup_program())
             input1 = np.random.random(size=[1,4]).astype('float32')
-            input2 = np.random.randint(1, 10, size=[1,10], dtype='int64')
+            input2 = np.random.randint(1, 10, size=[1], dtype='int64')
             out = exe.run(paddle.static.default_main_program(),
                           feed={'x':input1, 'y':input2},
                           fetch_list=[res.name])
diff --git a/python/paddle/fluid/layers/utils.py b/python/paddle/fluid/layers/utils.py
index e96e105c141df977af53c94072b6b987268e399d..66fc253bb097b95ab10a95c9d70eaa50d8ae0e65 100644
--- a/python/paddle/fluid/layers/utils.py
+++ b/python/paddle/fluid/layers/utils.py
@@ -14,7 +14,6 @@
 
 import collections
 import copy
-import six
 import numpy as np
 from ..framework import Block, Variable, _non_static_mode
 from ..data_feeder import (
diff --git a/python/paddle/fluid/reader.py b/python/paddle/fluid/reader.py
index 5c2e92ae458e79ba7857b03adb4ce3d81a66cd81..d71a4ab184a0fae155feee66f9c195ea66b81c1b 100644
--- a/python/paddle/fluid/reader.py
+++ b/python/paddle/fluid/reader.py
@@ -14,7 +14,6 @@
 
 from . import core
 import sys
-import six
 import numpy as np
 import threading
 import paddle
@@ -143,7 +142,7 @@ def _reader_process_loop(batch_reader, data_queue):
         # NOTE: Main process will raise KeyboardInterrupt anyways, ignore it in child process
         pass
     except:
-        six.reraise(*sys.exc_info())
+        raise
 
 
 class DataLoaderBase(object):
@@ -1202,7 +1201,7 @@ class DygraphGeneratorLoader(DataLoaderBase):
                 return self._reader.read_next_var_list()
         except StopIteration:
             self._reset()
-            six.reraise(*sys.exc_info())
+            raise
 
     def _exit_thread_expectedly(self):
         self._thread_done_event.set()
@@ -1232,7 +1231,7 @@ class DygraphGeneratorLoader(DataLoaderBase):
                 # start trying to get data from queue. At this time, the child thread needs
                 # to wait slightly longer
                 tensor_list = self._data_queue.get(timeout=QUEUE_GET_TIMEOUT)
-            except:
+            except Exception as e:
                 # NOTE [ avoid handing ] After adding the shared memory mechanism, not only
                 # the queue.Empty exception will occur here, but other exceptions will also
                 # occur, such as mmap failure. If it is not handled here, it will hang.
@@ -1240,7 +1239,7 @@ class DygraphGeneratorLoader(DataLoaderBase):
                 logging.error(
                     "DataLoader reader thread failed to read data from the multiprocessing.Queue."
                 )
-                six.reraise(*sys.exc_info())
+                raise e
 
             if not self._thread_done_event.is_set():
                 if tensor_list is not None:
@@ -1250,9 +1249,9 @@ class DygraphGeneratorLoader(DataLoaderBase):
                             array.append(tensor)
                         if not self._blocking_queue.push(array):
                             self._blocking_queue.close()
-                    except:
+                    except Exception as e:
                         self._exit_thread_unexpectedly()
-                        six.reraise(*sys.exc_info())
+                        raise e
                 else:
                     self._exit_thread_expectedly()
 
@@ -1278,13 +1277,13 @@ class DygraphGeneratorLoader(DataLoaderBase):
 
             self._blocking_queue.close()
             self._thread = None
-        except Exception:
+        except Exception as e:
             self._blocking_queue.kill()
             self._thread = None
             logging.warning(
                 "DygraphDataLoader reader thread raised an exception."
             )
-            six.reraise(*sys.exc_info())
+            raise e
 
     def set_sample_generator(
         self, reader, batch_size, drop_last=True, places=None
@@ -1510,7 +1509,7 @@ class GeneratorLoader(DataLoaderBase):
         except StopIteration:
             self._queue.close()
             self._reset()
-            six.reraise(*sys.exc_info())
+            raise
 
     def start(self):
         assert (
@@ -1551,11 +1550,11 @@ class GeneratorLoader(DataLoaderBase):
 
                 self._queue.close()
                 self._thread = None
-            except Exception as ex:
+            except Exception as e:
                 self._queue.kill()
                 self._thread = None
                 logging.warning('Your reader has raised an exception!')
-                six.reraise(*sys.exc_info())
+                raise e
 
         self._thread = threading.Thread(
             target=__thread_main__, args=(_current_expected_place(),)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_save_load_npu.py b/python/paddle/fluid/tests/unittests/npu/test_save_load_npu.py
index 09f25273696efd41d395cc76b6c6c6274ef5b754..0f86bafa9135277f062993278b74bbbc42fe1ecb 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_save_load_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_save_load_npu.py
@@ -26,7 +26,6 @@ from paddle.fluid.dygraph.base import to_variable
 from test_imperative_base import new_program_scope
 from paddle.fluid.executor import global_scope
 import numpy as np
-import six
 import pickle
 import os
 import errno
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_elastic_manager.py b/python/paddle/fluid/tests/unittests/test_fleet_elastic_manager.py
index a4ad3e1a7c4ea61f983d4d4c1a3d0c5d0950ecb6..0570ae1928c12440d6dad33b84b4fd62ab09f458 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_elastic_manager.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_elastic_manager.py
@@ -27,6 +27,16 @@ class MockLease:
         pass
 
 
+class MockKVMetadata:
+    def __init__(self, key):
+        self.key = key
+        self.create_revision = 2
+        self.mod_revision = 3
+        self.version = 2
+        self.lease_id = 0
+        self.response_header = None
+
+
 class MockEtcdClient:
     def __init__(self, lease=None):
         self._lease = lease
@@ -35,28 +45,30 @@ class MockEtcdClient:
         pass
 
     def get(self, key):
-        value = "0"
-        return value, value
+        return b'0', MockKVMetadata(b"/prefix")
 
     def delete_prefix(self, key):
         pass
 
     def get_prefix(self, key_prefix):
-        hosts = ["10.10.10.1:6001", "10.10.10.2:6001"]
-        return hosts
+        hosts = [
+            (b"/prefix/host1", b"10.10.10.1:6001"),
+            (b"/prefix/host2", b"10.10.10.2:6001"),
+        ]
+        return ((v, MockKVMetadata(k)) for k, v in hosts)
 
     def add_watch_callback(self, *args, **kwargs):
-        return "host_watch"
+        return 0
 
     def add_watch_prefix_callback(self, key_prefix, callback, **kwargs):
         callback(None)
-        return "host_watch"
+        return 0
 
     def cancel_watch(self, watch_id):
         pass
 
     def delete(self, key):
-        pass
+        return True
 
     def lease(self, ttl):
         if self._lease:
diff --git a/python/paddle/reader/decorator.py b/python/paddle/reader/decorator.py
index b2aa88cc810b4d46ea005aebc35323997d06b78b..1969c7ba11e9e87ae231de2badc9e57ca08eaa35 100644
--- a/python/paddle/reader/decorator.py
+++ b/python/paddle/reader/decorator.py
@@ -14,7 +14,6 @@
 
 from threading import Thread
 import multiprocessing
-import six
 import sys
 import warnings
 import logging
@@ -610,9 +609,9 @@ def multiprocess_reader(readers, use_pipe=True, queue_size=1000):
                     raise ValueError("sample has None")
                 queue.put(sample)
             queue.put(None)
-        except:
+        except Exception as e:
             queue.put("")
-            six.reraise(*sys.exc_info())
+            raise e
 
     def queue_reader():
         queue = fork_context.Queue(queue_size)
@@ -627,11 +626,11 @@ def multiprocess_reader(readers, use_pipe=True, queue_size=1000):
         while finish_num < reader_num:
             try:
                 sample = queue.get(timeout=QUEUE_GET_TIMEOUT)
-            except:
+            except Exception as e:
                 logging.error(
                     "multiprocess_reader failed to get data from the multiprocessing.Queue."
                 )
-                six.reraise(*sys.exc_info())
+                raise e
 
             if sample is None:
                 finish_num += 1
@@ -650,10 +649,10 @@ def multiprocess_reader(readers, use_pipe=True, queue_size=1000):
                 conn.send(json.dumps(sample))
             conn.send(json.dumps(None))
             conn.close()
-        except:
+        except Exception as e:
             conn.send(json.dumps(""))
             conn.close()
-            six.reraise(*sys.exc_info())
+            raise e
 
     def pipe_reader():
         conns = []
diff --git a/python/requirements.txt b/python/requirements.txt
index 74f2c2b9401aa9ebf5c1b30a96801e683bac417d..fcdfddc9e2eb69f16a7aafcc16ac5681e31da8ef 100644
--- a/python/requirements.txt
+++ b/python/requirements.txt
@@ -2,7 +2,6 @@ requests>=2.20.0
 numpy>=1.13
 protobuf>=3.1.0, <=3.20.0
 Pillow
-six
 decorator
 astor
 paddle_bfloat==0.1.7