diff --git a/python/paddle/amp/auto_cast.py b/python/paddle/amp/auto_cast.py
index 02b2c3c12a881..b8a40df7160c3 100644
--- a/python/paddle/amp/auto_cast.py
+++ b/python/paddle/amp/auto_cast.py
@@ -516,8 +516,7 @@ def amp_guard(
             or tracer._expected_place.is_custom_place()
         ):
             warnings.warn(
-                'amp_guard can only be enabled on CUDAPlace, XPUPlace, and CustomPlace, current place is %s, so it makes no effect.'
-                % tracer._expected_place
+                f'amp_guard can only be enabled on CUDAPlace, XPUPlace, and CustomPlace, current place is {tracer._expected_place}, so it makes no effect.'
             )
             enable = False
         if enable:
@@ -873,8 +872,7 @@ def amp_decorate(
     if save_dtype is not None:
         if save_dtype not in ['float16', 'bfloat16', 'float32', 'float64']:
             raise ValueError(
-                "save_dtype can only be float16 float32 or float64, but your input save_dtype is %s."
-                % save_dtype
+                f"save_dtype can only be float16 float32 or float64, but your input save_dtype is {save_dtype}."
             )
         for idx in range(len(models)):
             for layer in models[idx].sublayers(include_self=True):
diff --git a/python/paddle/amp/grad_scaler.py b/python/paddle/amp/grad_scaler.py
index 020c13d6a337a..d0c1b021d9956 100644
--- a/python/paddle/amp/grad_scaler.py
+++ b/python/paddle/amp/grad_scaler.py
@@ -110,8 +110,7 @@ def __init__(
                 or tracer._expected_place.is_custom_place()
             ):
                 warnings.warn(
-                    'AmpScaler can only be enabled on CUDAPlace, XPUPlace and CustomPlace, current place is %s, so it makes no effect.'
-                    % tracer._expected_place
+                    f'AmpScaler can only be enabled on CUDAPlace, XPUPlace and CustomPlace, current place is {tracer._expected_place}, so it makes no effect.'
                 )
                 enable = False
 
@@ -210,8 +209,7 @@ def scale(self, var):
             self._use_dynamic_loss_scaling = False
             self._init_loss_scaling = 1.0
             warnings.warn(
-                'It is not recommended to use dynamic loss scaling for %s, so GradScaler is disable by default.'
-                % (amp_global_state().amp_dtype)
+                f'It is not recommended to use dynamic loss scaling for {amp_global_state().amp_dtype}, so GradScaler is disable by default.'
             )
 
         if in_pir_mode():
diff --git a/python/paddle/audio/functional/window.py b/python/paddle/audio/functional/window.py
index 60eb9626ac693..1422702d97f8b 100644
--- a/python/paddle/audio/functional/window.py
+++ b/python/paddle/audio/functional/window.py
@@ -376,9 +376,7 @@ def get_window(
         else:
             winstr = window
     else:
-        raise ValueError(
-            "%s as window type is not supported." % str(type(window))
-        )
+        raise ValueError(f"{type(window)} as window type is not supported.")
 
     try:
         winfunc = window_function_register.get('_' + winstr)
diff --git a/python/paddle/base/multiprocess_utils.py b/python/paddle/base/multiprocess_utils.py
index 1445e985773cd..76800acfa84fb 100644
--- a/python/paddle/base/multiprocess_utils.py
+++ b/python/paddle/base/multiprocess_utils.py
@@ -71,7 +71,7 @@ def _func_executor():
 
         def _func_register(function):
             if not callable(function):
-                raise TypeError("%s is not callable object." % (function))
+                raise TypeError(f"{function} is not callable object.")
             # check function object whether hash-able
             if function not in cls._registered_func_set:
                 atexit.register(_func_executor)
diff --git a/python/paddle/dataset/common.py b/python/paddle/dataset/common.py
index 35155a2de2d22..79628cd045974 100644
--- a/python/paddle/dataset/common.py
+++ b/python/paddle/dataset/common.py
@@ -144,9 +144,9 @@ def fetch_all():
         x for x in dir(paddle.dataset) if not x.startswith("__")
     ]:
         if "fetch" in dir(
-            importlib.import_module("paddle.dataset.%s" % module_name)
+            importlib.import_module(f"paddle.dataset.{module_name}")
         ):
-            importlib.import_module('paddle.dataset.%s' % module_name).fetch()
+            importlib.import_module(f'paddle.dataset.{module_name}').fetch()
 
 
 def split(reader, line_count, suffix="%05d.pickle", dumper=pickle.dump):
@@ -210,7 +210,7 @@ def reader():
         my_file_list = []
         for idx, fn in enumerate(file_list):
             if idx % trainer_count == trainer_id:
-                print("append file: %s" % fn)
+                print(f"append file: {fn}")
                 my_file_list.append(fn)
         for fn in my_file_list:
             with open(fn, "r") as f:
diff --git a/python/paddle/dataset/conll05.py b/python/paddle/dataset/conll05.py
index 2295e9b1bd447..1e85fc1c46d5e 100644
--- a/python/paddle/dataset/conll05.py
+++ b/python/paddle/dataset/conll05.py
@@ -126,9 +126,7 @@ def reader():
                                     lbl_seq.append('B-' + cur_tag)
                                     is_in_bracket = True
                                 else:
-                                    raise RuntimeError(
-                                        'Unexpected label: %s' % l
-                                    )
+                                    raise RuntimeError(f'Unexpected label: {l}')
 
                             yield sentences, verb_list[i], lbl_seq
 
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_embedding.py b/python/paddle/distributed/auto_parallel/static/operators/dist_embedding.py
index 588a0f30ebb0b..33ddb3e53de32 100644
--- a/python/paddle/distributed/auto_parallel/static/operators/dist_embedding.py
+++ b/python/paddle/distributed/auto_parallel/static/operators/dist_embedding.py
@@ -422,7 +422,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"forward op [{str(src_op)}] don't have dist attribute !"
+        ), f"forward op [{src_op}] don't have dist attribute !"
 
         # check validation of inputs / outputs
         assert 'Ids' in kwargs, "input [{}] is not given".format('Ids')
@@ -578,7 +578,7 @@ def backward(ctx, *args, **kwargs):
         dist_attr = ctx.get_op_dist_attr_for_program(backward_op)
         assert (
             dist_attr is not None
-        ), f"backward op [{str(backward_op)}] don't have dist attribute !"
+        ), f"backward op [{backward_op}] don't have dist attribute !"
 
         # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
         if rank_id not in dist_attr.process_mesh.process_ids:
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_flash_attn.py b/python/paddle/distributed/auto_parallel/static/operators/dist_flash_attn.py
index 841dc0a587044..cfbe575ec1f2d 100644
--- a/python/paddle/distributed/auto_parallel/static/operators/dist_flash_attn.py
+++ b/python/paddle/distributed/auto_parallel/static/operators/dist_flash_attn.py
@@ -62,7 +62,7 @@ def forward(ctx, *args, **kwargs):
         ):
             assert (
                 op_dist_attr is not None
-            ), f"forward op [{str(src_op)}] don't have dist attribute !"
+            ), f"forward op [{src_op}] don't have dist attribute !"
 
             if (
                 len(kwargs.get('fixed_seed_offset', [])) > 0
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_fused_dropout_add.py b/python/paddle/distributed/auto_parallel/static/operators/dist_fused_dropout_add.py
index fbe74fe6c9503..03d83eacc3c5b 100644
--- a/python/paddle/distributed/auto_parallel/static/operators/dist_fused_dropout_add.py
+++ b/python/paddle/distributed/auto_parallel/static/operators/dist_fused_dropout_add.py
@@ -74,7 +74,7 @@ def forward(ctx, *args, **kwargs):
         if is_enable_auto_rand_ctrl() and not op_dist_attr.is_recompute:
             assert (
                 op_dist_attr is not None
-            ), f"forward op [{str(src_op)}] don't have dist attribute !"
+            ), f"forward op [{src_op}] don't have dist attribute !"
 
             assert 'seed_tensor' in kwargs, "input [{}] is not given".format(
                 'seed_tensor'
@@ -101,7 +101,7 @@ def forward(ctx, *args, **kwargs):
                     assert (
                         pre_op.type == "seed"
                         and len(pre_op.attr("rng_name")) == 0
-                    ), f"found exception op {str(pre_op)}"
+                    ), f"found exception op {pre_op}"
 
                     # determinate rng
                     X_var = main_block._var_recursive(kwargs['x'][0])
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_matmul.py b/python/paddle/distributed/auto_parallel/static/operators/dist_matmul.py
index 4b44e17dea210..d60ea6a640865 100644
--- a/python/paddle/distributed/auto_parallel/static/operators/dist_matmul.py
+++ b/python/paddle/distributed/auto_parallel/static/operators/dist_matmul.py
@@ -315,7 +315,7 @@ def _right_operand_parameter_matmul_backward(ctx, *args, **kwargs):
     dist_attr = ctx.get_op_dist_attr_for_program(backward_op)
     assert (
         dist_attr is not None
-    ), f"backward op [{str(backward_op)}] don't have dist attribute !"
+    ), f"backward op [{backward_op}] don't have dist attribute !"
 
     # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
     if rank_id not in dist_attr.process_mesh.process_ids:
@@ -782,7 +782,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
         if rank_id not in op_dist_attr.process_mesh.process_ids:
@@ -1037,7 +1037,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
         if rank_id not in op_dist_attr.process_mesh.process_ids:
@@ -1475,7 +1475,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
         if rank_id not in op_dist_attr.process_mesh.process_ids:
@@ -1724,7 +1724,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
         if rank_id not in op_dist_attr.process_mesh.process_ids:
@@ -2155,7 +2155,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
         if rank_id not in op_dist_attr.process_mesh.process_ids:
@@ -2398,7 +2398,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # FIXME (JZ-LIANG) Remove this hack to support any op mesh group for Pipeline Parallelism
         if rank_id not in op_dist_attr.process_mesh.process_ids:
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_reduce_sum_p.py b/python/paddle/distributed/auto_parallel/static/operators/dist_reduce_sum_p.py
index e99b57f8f97d8..e9bfa53bb0a9f 100644
--- a/python/paddle/distributed/auto_parallel/static/operators/dist_reduce_sum_p.py
+++ b/python/paddle/distributed/auto_parallel/static/operators/dist_reduce_sum_p.py
@@ -230,9 +230,7 @@ def forward(ctx, *args, **kwargs):
 
     @staticmethod
     def backward(ctx, *args, **kwargs):
-        raise RuntimeError(
-            f"primitive operator does NOT have backward function, op type: {str(op.type)}"  # noqa: F821
-        )
+        raise RuntimeError("primitive operator does NOT have backward function")
 
 
 register_distributed_operator_impl(
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_reshape.py b/python/paddle/distributed/auto_parallel/static/operators/dist_reshape.py
index ec1472896f5b2..bc3c565ad0e45 100644
--- a/python/paddle/distributed/auto_parallel/static/operators/dist_reshape.py
+++ b/python/paddle/distributed/auto_parallel/static/operators/dist_reshape.py
@@ -295,7 +295,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # check validation of inputs / outputs
         for input_name in src_op.desc.input_names():
@@ -551,7 +551,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # check validation of inputs / outputs
         for input_name in src_op.desc.input_names():
@@ -800,7 +800,7 @@ def forward(ctx, *args, **kwargs):
         op_dist_attr = ctx.get_op_dist_attr_for_program(src_op)
         assert (
             op_dist_attr is not None
-        ), f"backward op [{str(src_op)}] don't have dist attribute !"
+        ), f"backward op [{src_op}] don't have dist attribute !"
 
         # check validation of inputs / outputs
         for input_name in src_op.desc.input_names():
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_update_loss_scaling.py b/python/paddle/distributed/auto_parallel/static/operators/dist_update_loss_scaling.py
index 8ff358d14b1db..39d4fdfef974a 100644
--- a/python/paddle/distributed/auto_parallel/static/operators/dist_update_loss_scaling.py
+++ b/python/paddle/distributed/auto_parallel/static/operators/dist_update_loss_scaling.py
@@ -74,7 +74,7 @@ def backward(ctx, *args, **kwargs):
         dist_attr = ctx.get_op_dist_attr_for_program(backward_op)
         assert (
             dist_attr is not None
-        ), f"backward op [{str(backward_op)}] don't have dist attribute !"
+        ), f"backward op [{backward_op}] don't have dist attribute !"
 
         assert rank_id in dist_attr.process_mesh.process_ids
 
diff --git a/python/paddle/distributed/auto_parallel/static/tuner/tunable_variable.py b/python/paddle/distributed/auto_parallel/static/tuner/tunable_variable.py
index 83ed42c3fe1c0..fc30b377e7dff 100644
--- a/python/paddle/distributed/auto_parallel/static/tuner/tunable_variable.py
+++ b/python/paddle/distributed/auto_parallel/static/tuner/tunable_variable.py
@@ -87,7 +87,7 @@ def __init__(self, name, values, default=None):
         types = {type(v) for v in values}
         if len(types) > 1:
             raise TypeError(
-                f"Choice can contain only one type of value, but found values: {str(values)} with types: {str(types)}."
+                f"Choice can contain only one type of value, but found values: {values} with types: {types}."
             )
         self._is_unknown_type = False
 
@@ -185,7 +185,7 @@ def get_state(self):
     def _check_int(self, val):
         int_val = int(val)
         if int_val != val:
-            raise ValueError(f"Expects val is an int, but found: {str(val)}.")
+            raise ValueError(f"Expects val is an int, but found: {val}.")
         return int_val
 
     def __repr__(self):