diff --git a/python/paddle/reader/decorator.py b/python/paddle/reader/decorator.py
index 49e699aa94523a..cbda11e0375ba7 100644
--- a/python/paddle/reader/decorator.py
+++ b/python/paddle/reader/decorator.py
@@ -12,6 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from __future__ import annotations
+
 import itertools
 import logging
 import multiprocessing
@@ -21,9 +23,27 @@
 from itertools import zip_longest
 from queue import Queue
 from threading import Thread
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Callable,
+    Generator,
+    Sequence,
+    TypedDict,
+    TypeVar,
+    overload,
+)
+
+from typing_extensions import NotRequired, TypeAlias, Unpack
 
 from paddle.base.reader import QUEUE_GET_TIMEOUT
 
+if TYPE_CHECKING:
+
+    class _ComposeOptions(TypedDict):
+        check_alignment: NotRequired[bool]
+
+
 __all__ = []
 
 # On macOS, the 'spawn' start method is now the default in Python3.8 multiprocessing,
@@ -41,8 +61,18 @@
 else:
     fork_context = multiprocessing
 
+_T = TypeVar('_T')
+_T1 = TypeVar('_T1')
+_T2 = TypeVar('_T2')
+_T3 = TypeVar('_T3')
+_T4 = TypeVar('_T4')
+_U = TypeVar('_U')
+
 
-def cache(reader):
+_Reader: TypeAlias = Callable[[], Generator[_T, None, None]]
+
+
+def cache(reader: _Reader[_T]) -> _Reader[_T]:
     """
     Cache the reader data into memory.
 
@@ -77,12 +107,60 @@ def cache(reader):
     """
     all_data = tuple(reader())
 
-    def __impl__():
+    def __impl__() -> Generator[_T, None, None]:
         yield from all_data
 
     return __impl__
 
 
+# A temporary solution like builtin map function.
+# `Map` maybe the final solution in the future.
+# See https://github.com/python/typing/issues/1383
+@overload
+def map_readers(
+    func: Callable[[_T1], _U], reader1: _Reader[_T1], /
+) -> _Reader[_U]:
+    ...
+
+
+@overload
+def map_readers(
+    func: Callable[[_T1, _T2], _U],
+    reader1: _Reader[_T1],
+    reader2: _Reader[_T2],
+    /,
+) -> _Reader[_U]:
+    ...
+
+
+@overload
+def map_readers(
+    func: Callable[[_T1, _T2, _T3], _U],
+    reader1: _Reader[_T1],
+    reader2: _Reader[_T2],
+    reader3: _Reader[_T3],
+    /,
+) -> _Reader[_U]:
+    ...
+
+
+@overload
+def map_readers(
+    func: Callable[[_T1, _T2, _T3, _T4], _U],
+    reader1: _Reader[_T1],
+    reader2: _Reader[_T2],
+    reader3: _Reader[_T3],
+    reader4: _Reader[_T4],
+    /,
+) -> _Reader[_U]:
+    ...
+
+
+@overload
+def map_readers(func: Callable[..., _U], *readers: _Reader[Any]) -> _Reader[_U]:
+    ...
+
+
 def map_readers(func, *readers):
     """
     Creates a data reader that outputs return value of function using
@@ -124,7 +202,7 @@ def reader():
     return reader
 
 
-def shuffle(reader, buf_size):
+def shuffle(reader: _Reader[_T], buf_size: int) -> _Reader[_T]:
     """
     This API creates a decorated reader that outputs the shuffled data.
 
@@ -151,7 +229,7 @@ def shuffle(reader, buf_size):
             >>> # outputs are 0~4 unordered arrangement
     """
 
-    def data_reader():
+    def data_reader() -> Generator[_T, None, None]:
         buf = []
         for e in reader():
             buf.append(e)
@@ -169,7 +247,7 @@ def data_reader():
     return data_reader
 
 
-def chain(*readers):
+def chain(*readers: _Reader[_T]) -> _Reader[_T]:
     """
     Use the input data readers to create a chained data reader. The new created reader
     chains the outputs of input readers together as its output, and it do not change
@@ -218,8 +296,8 @@ def chain(*readers):
 
     """
 
-    def reader():
-        rs = []
+    def reader() -> Generator[_T, None, None]:
+        rs: list[Generator[_T, None, None]] = []
         for r in readers:
             rs.append(r())
 
@@ -232,7 +310,9 @@ class ComposeNotAligned(ValueError):
     pass
 
 
-def compose(*readers, **kwargs):
+def compose(
+    *readers: _Reader[Any], **kwargs: Unpack[_ComposeOptions]
+) -> _Reader[Any]:
     """
     Creates a data reader whose output is the combination of input readers.
 
@@ -289,7 +369,7 @@ def reader():
     return reader
 
 
-def buffered(reader, size):
+def buffered(reader: _Reader[_T], size: int) -> _Reader[_T]:
     """
     Creates a buffered data reader.
 
@@ -339,10 +419,7 @@ def data_reader():
         q = Queue(maxsize=size)
         t = Thread(
             target=read_worker,
-            args=(
-                r,
-                q,
-            ),
+            args=(r, q),
         )
         t.daemon = True
         t.start()
@@ -354,7 +431,7 @@ def data_reader():
     return data_reader
 
 
-def firstn(reader, n):
+def firstn(reader: _Reader[_T], n: int) -> _Reader[_T]:
     """
 
     This API creates a decorated reader, and limits the max number of
@@ -399,7 +476,13 @@ class XmapEndSignal:
     pass
 
 
-def xmap_readers(mapper, reader, process_num, buffer_size, order=False):
+def xmap_readers(
+    mapper: Callable[[_T], _U],
+    reader: _Reader[_T],
+    process_num: int,
+    buffer_size: int,
+    order: bool = False,
+) -> _Reader[_U]:
     """
     Use multi-threads to map samples from reader by a mapper defined by user.
 
@@ -495,7 +578,11 @@ def xreader():
     return xreader
 
 
-def multiprocess_reader(readers, use_pipe=True, queue_size=1000):
+def multiprocess_reader(
+    readers: Sequence[_Reader[_T]],
+    use_pipe: bool = True,
+    queue_size: int = 1000,
+) -> _Reader[list[_T]]:
     """
     This API use python ``multiprocessing`` to read data from ``readers`` parallelly,
     and then ``multiprocess.Queue`` or ``multiprocess.Pipe`` is used to merge
@@ -508,13 +595,13 @@ def multiprocess_reader(readers, use_pipe=True, queue_size=1000):
     in some platforms.
 
     Parameters:
-       readers (list( ``generator`` ) | tuple( ``generator`` )): a python ``generator`` list
-           used to read input data
-       use_pipe (bool, optional): control the inner API used to implement the multi-processing,
-           default True - use ``multiprocess.Pipe`` which is recommended
-       queue_size (int, optional): only useful when ``use_pipe`` is False - ``multiprocess.Queue``
-           is used, default 1000. Increase this value can speed up the data reading, and more memory
-           will be consumed.
+        readers (list( ``generator`` ) | tuple( ``generator`` )): a python ``generator`` list
+            used to read input data
+        use_pipe (bool, optional): control the inner API used to implement the multi-processing,
+            default True - use ``multiprocess.Pipe`` which is recommended
+        queue_size (int, optional): only useful when ``use_pipe`` is False - ``multiprocess.Queue``
+            is used, default 1000. Increase this value can speed up the data reading, and more memory
+            will be consumed.
 
     Returns:
         ``generator``: a new reader which can be run parallelly