[CodeStyle][task 3] enable Ruff PLR1711 rule in python/paddle/base (P…

…addlePaddle#57444)

pyproject.toml

@@ -70,7 +70,7 @@ select = [
     "PLR0206",
     "PLR0402",
     "PLR1701",
-    # "PLR1711",  # Confirmation required
+    "PLR1711",
     "PLR1722",
     "PLW3301",
 ]

python/paddle/base/core.py

@@ -507,7 +507,6 @@ def _set_prim_forward_blacklist(*args):
             raise TypeError("ops set in forward_blacklist must belong to str")
         else:
             prim_config["forward_blacklist"].add(item)
-    return
 
 
 def _set_prim_backward_blacklist(*args):
@@ -516,7 +515,6 @@ def _set_prim_backward_blacklist(*args):
         if not isinstance(item, str):
             raise TypeError("all items in set must belong to string")
     _set_bwd_prim_blacklist(ops)
-    return
 
 
 def _set_prim_backward_enabled(value):

python/paddle/base/incubate/checkpoint/auto_checkpoint.py

@@ -633,8 +633,6 @@ def _normal_yield(max_epoch_num):
         max_epoch_num = sys.maxint
     yield from range(0, max_epoch_num)
 
-    return
-
 
 def train_epoch_range(max_epoch_num, save_checkpoint_inter=None):
     global g_acp_type

python/paddle/base/layers/math_op_patch.py

@@ -236,7 +236,6 @@ def place(self):
         warnings.warn(
             "Variable do not have 'place' interface for static graph mode, try not to use it. None will be returned."
         )
-        return None
 
     def astype(self, dtype):
         """

python/paddle/check_import_scipy.py

@@ -26,4 +26,3 @@ def check_import_scipy(OsName):
                     print_info
                     + "\nplease download Visual C++ Redistributable from https://support.microsoft.com/en-us/topic/the-latest-supported-visual-c-downloads-2647da03-1eea-4433-9aff-95f26a218cc0"
                 )
-    return

python/paddle/distributed/auto_parallel/static/operators/dist_default.py

@@ -92,8 +92,6 @@ def prim_operator_data_parallel_functor(ctx, src_op):
         op_attr.set_input_dims_mapping(grad_var.name, dims_mapping)
         ctx.set_op_dist_attr_for_program(allreduce_op, op_attr)
 
-    return
-
 
 class DistributedDefault(DistributedOperatorImplContainer):
     def __init__(self, op_type):

python/paddle/distributed/auto_tuner/recorder.py

@@ -46,7 +46,6 @@ def sort_metric(self, direction, metric_name) -> None:
                 else float('inf'),
                 reverse=False,
             )
-        return
 
     def get_best(self, metric, direction) -> Tuple[dict, bool]:
         self.sort_metric(direction=direction, metric_name=metric)

python/paddle/distributed/communication/stream/all_reduce.py

@@ -68,8 +68,6 @@ def _all_reduce_in_static_mode(tensor, op, group, sync_op, use_calc_stream):
         attrs={'ring_id': ring_id, 'use_calc_stream': sync_op},
     )
 
-    return None
-
 
 def all_reduce(
     tensor, op=ReduceOp.SUM, group=None, sync_op=True, use_calc_stream=False

python/paddle/distributed/communication/stream/all_to_all.py

@@ -123,8 +123,6 @@ def _all_to_all_in_static_mode(
                 paddle.split(out_tensor, nranks, 0)
             )
 
-    return None
-
 
 def alltoall(
     out_tensor_or_tensor_list,

python/paddle/distributed/communication/stream/broadcast.py

@@ -69,7 +69,6 @@ def _broadcast_in_static_mode(
             'ring_id': ring_id,
         },
     )
-    return None
 
 
 def broadcast(tensor, src, group=None, sync_op=True, use_calc_stream=False):

python/paddle/distributed/communication/stream/recv.py

@@ -59,7 +59,6 @@ def _recv_in_static_mode(
             'use_calc_stream': sync_op,
         },
     )
-    return None
 
 
 def recv(tensor, src=0, group=None, sync_op=True, use_calc_stream=False):

python/paddle/distributed/communication/stream/reduce.py

@@ -73,7 +73,6 @@ def _reduce_in_static_mode(
             'root_id': dst_rank_in_group,
         },
     )
-    return None
 
 
 def reduce(

python/paddle/distributed/communication/stream/reduce_scatter.py

@@ -99,7 +99,6 @@ def _reduce_scatter_in_static_mode(tensor, tensor_or_tensor_list, group):
             'nranks': nranks,
         },
     )
-    return None
 
 
 def reduce_scatter(

python/paddle/distributed/communication/stream/scatter.py

@@ -129,8 +129,6 @@ def _scatter_in_static_mode(
         },
     )
 
-    return None
-
 
 def scatter(
     tensor,

python/paddle/distributed/communication/stream/send.py

@@ -58,7 +58,6 @@ def _send_in_static_mode(
             'use_calc_stream': sync_op,
         },
     )
-    return None
 
 
 def send(tensor, dst=0, group=None, sync_op=True, use_calc_stream=False):

python/paddle/distributed/communicator.py

@@ -229,14 +229,12 @@ def start_coordinator(self, self_endpoint, trainer_endpoints):
             self.communicator_.start_coordinator(
                 self_endpoint, trainer_endpoints
             )
-        return
 
     def save_fl_strategy(self, mp):
         if self.communicator_ is not None:
             self.communicator_.save_fl_strategy(mp)
         else:
             raise ValueError("self.communicator_ is null")
-        return
 
     def query_fl_clients_info(self):
         info_mp = {}

python/paddle/distributed/fleet/base/role_maker.py

@@ -497,7 +497,6 @@ def to_string(self):
 
     def _all_gather(self, input, comm_world="worker"):
         print("warning: RoleMakerBase does not have all gather worker.")
-        return None
 
     def _all_reduce(self, input, mode="sum", comm_world="worker"):
         """
@@ -507,7 +506,6 @@ def _all_reduce(self, input, mode="sum", comm_world="worker"):
             mode(str): "sum" or "min" or "max"
         """
         print("warning: RoleMakerBase does not have all reduce worker.")
-        return None
 
     def _barrier(self, comm_world):
         """

python/paddle/distributed/fleet/launch_utils.py

@@ -843,7 +843,6 @@ def direct_start(args):
     ] + args.training_script_args
     proc = subprocess.Popen(cmd)
     proc.wait()
-    return
 
 
 def get_custom_endpoints(origin_endpoints, offset=0):

python/paddle/distributed/fleet/meta_optimizers/sharding/utils.py

@@ -93,7 +93,6 @@ def check_broadcast(block):
                     < last_sync_comm_op_idx
                 )
                 assert last_sync_comm_op_idx < idx
-    return
 
 
 def check_allreduce_sum(block, shard, sharding_ring_id, dp_ring_id=-1):
@@ -254,8 +253,6 @@ def check_allreduce_sum(block, shard, sharding_ring_id, dp_ring_id=-1):
     if idx_gradient_clip_allreduce != -1:
         assert idx_gradient_clip_allreduce > idx_last_grad_allreduce
 
-    return
-
 
 def get_valid_op_role(block, insert_idx):
     """
@@ -284,7 +281,6 @@ def insert_sync_calc_op(block, insert_idx, calc_dep_vars):
         outputs={'Out': calc_dep_vars},
         attrs={OP_ROLE_KEY: op_role},
     )
-    return
 
 
 def insert_sync_comm_op(block, insert_idx, ring_id, comm_dep_vars):
@@ -339,7 +335,6 @@ def insert_fill_constant_ops(block, insert_idx, fill_constant_vars):
                 OP_ROLE_KEY: op_role,
             },
         )
-    return
 
 
 def insert_cast_ops(block, insert_idx, cast_ops):
@@ -359,7 +354,6 @@ def insert_cast_ops(block, insert_idx, cast_ops):
                 OP_ROLE_KEY: op_role,
             },
         )
-    return
 
 
 def insert_allreduce_ops(
@@ -873,8 +867,6 @@ def insert_broadcast_ops(block, insert_idx, ring_id, broadcast2root):
             },
         )
 
-    return
-
 
 DtypeToSize = {
     core.VarDesc.VarType.FP16: 2,
@@ -994,7 +986,6 @@ def add_sync_comm(program, sharding_ring_id):
                 'op_role': core.op_proto_and_checker_maker.OpRole.Forward,
             },
         )
-    return
 
 
 def save_persistables(exe, dirname, main_program, filename=None):
@@ -1055,8 +1046,6 @@ def sharding_predicate(var):
             filename=None,
         )
 
-    return
-
 
 def append_naive_sync(block, sync_var, ring_id):
     # NOTE (JZ-LIANG) update this to use barrier sync for more elegent logic

python/paddle/distributed/fleet/meta_optimizers/sharding_optimizer.py

@@ -980,7 +980,6 @@ def _split_program(self, block):
                         ].desc.input_arg_names(),
                     )
                 )
-        return
 
     def _prune_main_program(self, block, shard, rings):
         """
@@ -1096,7 +1095,6 @@ def _prune_main_program(self, block, shard, rings):
                         reserved_x.append(var_name)
                 op.desc.set_input('X', reserved_x)
         block._sync_with_cpp()
-        return
 
     def _add_broadcast_allreduce(self, block):
         """
@@ -1665,8 +1663,6 @@ def _build_groups(self):
         logger.info(f"pure dp ring id: {self.dp_ring_id}")
         logger.info("#####" * 6)
 
-        return
-
     def _recreate_not_persist_param_as_var(self):
         def recreate_not_persist_param_as_var(program):
             block = program.global_block()

python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_stage3.py

@@ -959,7 +959,6 @@ def _release_param(
 
                 if offload:
                     param.fw_storage = _device2cpu(param.fw_storage)
-    return
 
 
 def _wait_layer(

python/paddle/distributed/fleet/utils/mix_precision_utils.py

@@ -67,7 +67,6 @@ def param_hook(tmp_grad):
                     param.main_grad.add_(tmp_grad)
 
                 tmp_grad._clear_data()
-            return None
 
         return param_hook
 

python/paddle/distributed/passes/auto_parallel_sharding.py

@@ -1430,7 +1430,6 @@ def _insert_init_and_broadcast_op(
             broadcast_var_dist_attr.dims_mapping,
             dist_context,
         )
-    return
 
 
 def _insert_reduce_op(

python/paddle/distributed/passes/ps_trainer_pass.py

@@ -1391,7 +1391,6 @@ def _insert_partA_communicate_op(self, block, idx):
                 RPC_OP_ROLE_ATTR_NAME: RPC_OP_ROLE_ATTR_VALUE,
             },
         )
-        return
 
     def _insert_partB_communicate_op(self, block, idx):
         comm_info = f"backward_joint_{2}_{1}@fl_ps"
@@ -1416,7 +1415,6 @@ def _insert_partB_communicate_op(self, block, idx):
                 RPC_OP_ROLE_ATTR_NAME: RPC_OP_ROLE_ATTR_VALUE,
             },
         )
-        return
 
     def _create_var_for_block(self, vars, block):
         for var in vars:

python/paddle/distributed/ps/coordinator.py

@@ -246,7 +246,6 @@ def train_loop(self):
     def push_fl_client_info_sync(self, state_info):
         str_msg = self.__build_fl_client_info_desc(state_info)
         self._client_ptr.push_fl_client_info_sync(str_msg)
-        return
 
     def pull_fl_strategy(self):
         strategy_dict = {}

python/paddle/distributed/ps/utils/ps_program_builder.py

@@ -125,7 +125,6 @@ def _build_pserver_programs(self):
         add_listen_and_serv_pass.apply(
             [self.attrs['_main_server']], [None], self.pass_ctx
         )
-        return
 
 
 class NuPsProgramBuilder(PsProgramBuilder):
@@ -174,8 +173,6 @@ def _build_trainer_programs(self):
         if self.launch_barrier and self.launch_barrier_flag:
             wait_server_ready(self.server_endpoints)
 
-        return
-
 
 class CpuSyncPsProgramBuilder(PsProgramBuilder):
     def __init__(self, pass_ctx):
@@ -226,8 +223,6 @@ def _build_trainer_programs(self):
         if self.launch_barrier and self.launch_barrier_flag:
             wait_server_ready(self.server_endpoints)
 
-        return
-
 
 class CpuAsyncPsProgramBuilder(CpuSyncPsProgramBuilder):
     def __init__(self, pass_ctx):
@@ -296,8 +291,6 @@ def _build_trainer_programs(self):
         if self.launch_barrier and self.launch_barrier_flag:
             wait_server_ready(self.server_endpoints)
 
-        return
-
 
 class HeterAsyncPsProgramBuilder(PsProgramBuilder):
     def __init__(self, pass_ctx):
@@ -355,8 +348,6 @@ def _build_trainer_programs(self):
         if self.launch_barrier and self.launch_barrier_flag:
             wait_server_ready(self.server_endpoints)
 
-        return
-
     def _build_programs(self):
         if self.attrs['is_worker'] or self.attrs['is_heter_worker']:
             self._build_trainer_programs()
@@ -458,8 +449,6 @@ def _build_trainer_programs(self):
                 ],
             )
 
-        return
-
     def _build_pserver_programs(self):
         self.loss.block.program = self.attrs['_main_server']
 

python/paddle/incubate/distributed/fleet/role_maker.py

@@ -160,7 +160,6 @@ def all_gather(self, input):
             return a list of values
         """
         print("warning: RoleMakerBase does not have all gather.")
-        return None
 
     def all_reduce_worker(self, input, output, mode="sum"):
         """

python/paddle/incubate/optimizer/recompute.py

@@ -274,8 +274,6 @@ def _append_fill_constant_ops(self, startup_program):
                 },
             )
 
-        return
-
     def _insert_async_memcpy_op(
         self, insert_idx, src_varname, dst_varname, op_role, dst_place_type
     ):
@@ -639,8 +637,6 @@ def _offload(self, loss, startup_program=None):
             # step 4. verify the correctness
             self._check_offload_fetch()
 
-        return
-
     def backward(
         self,
         loss,

python/paddle/jit/api.py

@@ -821,7 +821,6 @@ def set_property(meta, key, val):
                 meta.set_strings(key, val)
         else:
             raise ValueError(f"Note support val type: {type(val)}")
-        return
 
     with open(filename, 'wb') as f:
         meta = paddle.framework.core.Property()