[CodeStyle][py2][U004] unecessary explicit `object` inheritance in class definition (#47642)

* [CodeStyle][py2][U004] unecessary explicit `object` inheritance in class definition

* fix an increment

paddle/infrt/tests/models/efficientnet-b4/net/utils.py

@@ -239,7 +239,7 @@ def efficientnet_params(model_name):
     return params_dict[model_name]
 
 
-class BlockDecoder(object):
+class BlockDecoder:
     """Block Decoder for readability, straight from the official TensorFlow repository"""
 
     @staticmethod

paddle/phi/api/yaml/generator/api_base.py

@@ -19,7 +19,7 @@ PREFIX_TENSOR_NAME = 'input_'
 PREFIX_META_TENSOR_NAME = 'meta_'
 
 
-class BaseAPI(object):
+class BaseAPI:
     def __init__(self, api_item_yaml):
         self.api = self.get_api_name(api_item_yaml)
 

python/paddle/audio/functional/window.py

@@ -19,7 +19,7 @@ import paddle
 from paddle import Tensor
 
 
-class WindowFunctionRegister(object):
+class WindowFunctionRegister:
     def __init__(self):
         self._functions_dict = dict()
 

python/paddle/autograd/py_layer.py

@@ -21,7 +21,7 @@ from paddle.fluid import core
 __all__ = []
 
 
-class LegacyPyLayerContext(object):
+class LegacyPyLayerContext:
     """
     The object of this class is a context that is used in PyLayer to enhance the function.
 
@@ -131,7 +131,7 @@ def with_mateclass(meta, *bases):
     return type.__new__(impl, "impl", (), {})
 
 
-class CPyLayer(object):
+class CPyLayer:
     @classmethod
     @dygraph_only
     def apply(cls, *args, **kwargs):
@@ -336,7 +336,7 @@ class LegacyPyLayer(with_mateclass(LayerMeta, CPyLayer)):
         )
 
 
-class EagerPyLayerContext(object):
+class EagerPyLayerContext:
     def save_for_backward(self, *tensors):
         """
         Saves given tensors that backward need. Use ``saved_tensor`` in the `backward` to get the saved tensors.

python/paddle/dataset/imikolov.py

@@ -31,7 +31,7 @@ URL = 'https://dataset.bj.bcebos.com/imikolov%2Fsimple-examples.tgz'
 MD5 = '30177ea32e27c525793142b6bf2c8e2d'
 
 
-class DataType(object):
+class DataType:
     NGRAM = 1
     SEQ = 2
 

python/paddle/dataset/movielens.py

@@ -38,7 +38,7 @@ URL = 'https://dataset.bj.bcebos.com/movielens%2Fml-1m.zip'
 MD5 = 'c4d9eecfca2ab87c1945afe126590906'
 
 
-class MovieInfo(object):
+class MovieInfo:
     """
     Movie id, title and categories information are stored in MovieInfo.
     """
@@ -69,7 +69,7 @@ class MovieInfo(object):
         return self.__str__()
 
 
-class UserInfo(object):
+class UserInfo:
     """
     User id, gender, age, and job information are stored in UserInfo.
     """

python/paddle/distributed/auto_parallel/cluster_v2.py

@@ -116,7 +116,7 @@ class DeviceMesh(core.DeviceMesh):
         return self._mesh
 
 
-# class Cluster(object):
+# class Cluster:
 #     """
 #     The cluster represents the hardware resource.
 #     """

python/paddle/distributed/auto_parallel/converter.py

@@ -19,7 +19,7 @@ import numpy as np
 from ..utils.log_utils import get_logger
 
 
-class Converter(object):
+class Converter:
     """
     Converter is a class object for auto parallel to convert tensors from
     one parallel strategy to another one. Tensors will merge and slice value

python/paddle/distributed/auto_parallel/cost_model.py

@@ -35,7 +35,7 @@ class CostNodeType(Enum):
     NOP = 5
 
 
-class Cost(object):
+class Cost:
     def __init__(self):
         self.runtime = None
         self.static_mem = None
@@ -49,7 +49,7 @@ class CostModelMode(Enum):
     MIXED = 3
 
 
-class CostNode(object):
+class CostNode:
     def __init__(self, node, node_type, id=None):
         self.id = id
         self.node = node
@@ -172,7 +172,7 @@ class CompOpCostNode(CostNode):
             self.cost = 0.0
 
 
-class PipeEvent(object):
+class PipeEvent:
     def __init__(self, stage_id, event_name, duration, start_time=-1):
         self.stage_id = stage_id
         self.name = event_name
@@ -181,7 +181,7 @@ class PipeEvent(object):
         self.e_time = -1
 
 
-class CostModel(object):
+class CostModel:
     def __init__(
         self,
         mode=CostModelMode.BENCHMARKING,

python/paddle/distributed/auto_parallel/dist_context.py

@@ -1146,7 +1146,7 @@ class DistributedOperatorContext:
         return kinputs, koutputs
 
 
-class BlockState(object):
+class BlockState:
     def __init__(self):
         self.nblock = 0
         self.forward_indices = []

python/paddle/distributed/auto_parallel/helper.py

@@ -192,7 +192,7 @@ class BuildInfo:
         self.states = defaultdict(bool)
 
 
-class ProgramHelper(object):
+class ProgramHelper:
     """
     A Helper class for Engine to provides different Program IR according specified 'mode'.
     """

python/paddle/distributed/auto_parallel/interface.py

@@ -220,7 +220,7 @@ def recompute(op):
 _g_collections = {}
 
 
-class CollectionNames(object):
+class CollectionNames:
     FETCHES = "fetches"
     LOGGING = "logging"
 

python/paddle/distributed/auto_parallel/partitioner.py

@@ -32,7 +32,7 @@ __not_shape_var_type__ = [
 ]
 
 
-class Partitioner(object):
+class Partitioner:
     """
     warning:: Partitioner is experimental and subject to change.
 

python/paddle/distributed/auto_parallel/process_mesh.py

@@ -39,7 +39,7 @@ def reset_current_process_mesh():
     _g_current_process_mesh = _g_previous_process_mesh
 
 
-class ProcessMesh(object):
+class ProcessMesh:
     """
     The `Processmesh` object describes the topology of the used processes.
 

python/paddle/distributed/auto_parallel/strategy.py

@@ -16,7 +16,7 @@ import copy
 from . import constants
 
 
-class BaseConfig(object):
+class BaseConfig:
     def __init__(self, category, config_dict=None):
         self._category = category
         self._config_dict = None

python/paddle/distributed/auto_parallel/tuner/config.py

@@ -25,7 +25,7 @@ def _get_pass_config(strategy, pass_name):
     return config
 
 
-class TuningConfig(object):
+class TuningConfig:
     """
     A uniform config wrap:
     distributed strategy: the user defined configuration for optimization pass

python/paddle/distributed/auto_parallel/tuner/recorder.py

@@ -18,7 +18,7 @@
 import numpy as np
 
 
-class MetricRecord(object):
+class MetricRecord:
     """
     One record for a single metric at a given execution step.
     """
@@ -62,7 +62,7 @@ class MetricRecord(object):
         return "MetricRecord(value={}, step={})".format(self.value, self.step)
 
 
-class MetricRecords(object):
+class MetricRecords:
     """
     Records of a single metric across different executions.
     """
@@ -143,7 +143,7 @@ class MetricRecords(object):
         return records
 
 
-class MetricsRecorder(object):
+class MetricsRecorder:
     """
     Record the values for all metrics.
     """

python/paddle/distributed/auto_parallel/tuner/storable.py

@@ -18,7 +18,7 @@
 import json
 
 
-class Storable(object):
+class Storable:
     def get_state(self):
         raise NotImplementedError
 

python/paddle/distributed/auto_parallel/tuner/tunable_space.py

@@ -22,7 +22,7 @@ from .tunable_variable import IntRange
 from .tunable_variable import FloatRange
 
 
-class TunableSpace(object):
+class TunableSpace:
     """
     A TunableSpace is constructed by the tunable variables.
     """

python/paddle/distributed/auto_parallel/tuner/tunable_variable.py

@@ -18,7 +18,7 @@
 import numpy as np
 
 
-class TunableVariable(object):
+class TunableVariable:
     """
     Tunablevariable base class.
     """

python/paddle/distributed/communication/batch_isend_irecv.py

@@ -22,7 +22,7 @@ from paddle.distributed.communication.group import (
 )
 
 
-class P2POp(object):
+class P2POp:
     """
     A class that makes point-to-point operations for "batch_isend_irecv".
 

python/paddle/distributed/elastic.py

@@ -16,7 +16,7 @@ import argparse
 import os
 
 
-class Command(object):
+class Command:
     def __init__(self, server, name):
         import etcd3
 

python/paddle/distributed/entry_attr.py

@@ -15,7 +15,7 @@
 __all__ = []
 
 
-class EntryAttr(object):
+class EntryAttr:
     """
     Entry Config for paddle.static.nn.sparse_embedding with Parameter Server.
 

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

@@ -65,7 +65,7 @@ def check_configs_key(msg, config, field_name):
         assert key in key_list, "key:{} not in {}".format(key, field_name)
 
 
-class DistributedJobInfo(object):
+class DistributedJobInfo:
     """
     DistributedJobInfo will serialize all distributed training information
     Just for inner use: 1) debug 2) replicate experiments
@@ -106,7 +106,7 @@ ReduceStrategyFluid = paddle.fluid.BuildStrategy.ReduceStrategy
 ReduceStrategyFleet = int
 
 
-class DistributedStrategy(object):
+class DistributedStrategy:
     __lock_attr = False
 
     def __init__(self):

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

@@ -26,7 +26,7 @@ meta_optimizer_names.remove("HybridParallelOptimizer")
 meta_optimizer_names.remove("HeterParallelOptimizer")
 
 
-class MetaOptimizerFactory(object):
+class MetaOptimizerFactory:
     def __init__(self):
         pass
 

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

@@ -35,7 +35,7 @@ class Role:
     COORDINATOR = 5
 
 
-class Gloo(object):
+class Gloo:
     """
     Gloo is a universal class for barrier and collective communication
     """
@@ -383,7 +383,7 @@ class Gloo(object):
         return output
 
 
-class RoleMakerBase(object):
+class RoleMakerBase:
     """
     RoleMakerBase is a base class for assigning a role to current process
     in distributed training.

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

@@ -17,7 +17,7 @@ from ...ps.the_one_ps import TheOnePSRuntime
 __all__ = []
 
 
-class RuntimeFactory(object):
+class RuntimeFactory:
     def __init__(self):
         pass
 

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

@@ -106,7 +106,7 @@ def maximum_path_len_algo(optimizer_list):
     return candidate
 
 
-class StrategyCompilerBase(object):
+class StrategyCompilerBase:
     def __init__(self):
         pass
 

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

@@ -23,7 +23,7 @@ __all__ = ['CommunicateTopology', 'HybridCommunicateGroup']
 _HYBRID_PARALLEL_GROUP = None
 
 
-class ParallelMode(object):
+class ParallelMode:
     """
     There are all the parallel modes currently supported:
     - DATA_PARALLEL: Distribute input data to different devices.
@@ -47,7 +47,7 @@ class ParallelMode(object):
     SHARDING_PARALLEL = 3
 
 
-class CommunicateTopology(object):
+class CommunicateTopology:
     def __init__(
         self,
         hybrid_group_names=["data", "pipe", "sharding", "model"],
@@ -133,7 +133,7 @@ class CommunicateTopology(object):
         return self.get_rank(**tf)
 
 
-class HybridCommunicateGroup(object):
+class HybridCommunicateGroup:
     def __init__(self, topology):
         self.nranks = paddle.distributed.get_world_size()
         self.global_rank = paddle.distributed.get_rank()
@@ -410,7 +410,7 @@ class HybridCommunicateGroup(object):
         )
 
 
-class _CommunicateGroup(object):
+class _CommunicateGroup:
     """tmp for static"""
 
     def __init__(self):

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

@@ -31,7 +31,7 @@ import numpy as np
 __all__ = []
 
 
-class UtilFactory(object):
+class UtilFactory:
     def _create_util(self, context=None):
         util = UtilBase()
         if context is not None and "valid_strategy" in context:
@@ -41,7 +41,7 @@ class UtilFactory(object):
         return util
 
 
-class UtilBase(object):
+class UtilBase:
     def __init__(self):
         self.role_maker = None
         self.dist_strategy = None

python/paddle/distributed/fleet/data_generator/data_generator.py

@@ -17,7 +17,7 @@ import sys
 __all__ = []
 
 
-class DataGenerator(object):
+class DataGenerator:
     """
     DataGenerator is a general Base class for user to inherit
     A user who wants to define his/her own python processing logic

python/paddle/distributed/fleet/dataset/dataset.py

@@ -20,7 +20,7 @@ import paddle.fluid.core as core
 __all__ = []
 
 
-class DatasetBase(object):
+class DatasetBase:
     """Base dataset class."""
 
     def __init__(self):

python/paddle/distributed/fleet/dataset/index_dataset.py

@@ -16,7 +16,7 @@ from paddle.fluid import core
 __all__ = []
 
 
-class Index(object):
+class Index:
     def __init__(self, name):
         self._name = name
 

python/paddle/distributed/fleet/elastic/manager.py

@@ -52,7 +52,7 @@ class ElasticStatus:
     EXIT = "exit"
 
 
-class LauncherInterface(object):
+class LauncherInterface:
     def __init__(self, args):
         self.args = args
         self.procs = []
@@ -124,7 +124,7 @@ class LauncherInterface(object):
         raise NotImplementedError
 
 
-class ElasticManager(object):
+class ElasticManager:
     def __init__(self, args, etcd_client):
 
         self.args = args

python/paddle/distributed/fleet/fleet.py

@@ -95,7 +95,7 @@ inited_runtime_handler = wrap_decorator(_inited_runtime_handler_)
 is_non_distributed_check = wrap_decorator(_is_non_distributed_check_)
 
 
-class Fleet(object):
+class Fleet:
     """
     Unified API for distributed training of PaddlePaddle
     Please reference the https://github.com/PaddlePaddle/PaddleFleetX for details

python/paddle/distributed/fleet/launch_utils.py

@@ -60,7 +60,7 @@ class DeviceMode:
     MLU = 4
 
 
-class Cluster(object):
+class Cluster:
     def __init__(self, hdfs):
         self.job_server = None
         self.pods = []
@@ -133,7 +133,7 @@ class Cluster(object):
         return None
 
 
-class JobServer(object):
+class JobServer:
     def __init__(self):
         self.endpoint = None
 
@@ -147,7 +147,7 @@ class JobServer(object):
         return not self == j
 
 
-class Trainer(object):
+class Trainer:
     def __init__(self):
         self.accelerators = []
         self.endpoint = None
@@ -179,7 +179,7 @@ class Trainer(object):
         return self.rank
 
 
-class Pod(object):
+class Pod:
     def __init__(self):
         self.rank = None
         self.id = None
@@ -483,7 +483,7 @@ def pretty_print_envs(envs, header=None):
     return _str
 
 
-class TrainerProc(object):
+class TrainerProc:
     def __init__(self):
         self.proc = None
         self.log_fn = None
@@ -1278,7 +1278,7 @@ def get_mapped_cluster_from_args_with_rank_mapping(args, device_mode):
     )
 
 
-class ParameterServerLauncher(object):
+class ParameterServerLauncher:
     def __init__(self, args, distribute_mode):
         self.args = args
         self.distribute_mode = distribute_mode

python/paddle/distributed/fleet/meta_optimizers/ascend/ascend_optimizer.py

@@ -24,7 +24,7 @@ HcomGroupConfig = namedtuple('HcomGroupConfig', ['name', 'nranks', 'rank_ids'])
 __all__ = []
 
 
-class AscendIRParser(object):
+class AscendIRParser:
     def __init__(self, auto_dp=False, world_rank_size=1):
         self.graph_idx = 0
         self.hcom_endpoints = {}

python/paddle/distributed/fleet/meta_optimizers/ascend/ascend_parser.py

@@ -101,7 +101,7 @@ global_cnt = -1
 global_input_cnt = -1
 
 
-class AscendHelper(object):
+class AscendHelper:
     def __init__(self):
         self.dtype2ge_map = {
             0: core.GEDataType.DT_BOOL,
@@ -136,7 +136,7 @@ class AscendHelper(object):
         return self.dtype2np_map[index]
 
 
-class AscendParserFactory(object):
+class AscendParserFactory:
     def __init__(self, graph, var2geop):
         self.graph = graph
         self.var2geop = var2geop
@@ -149,7 +149,7 @@ class AscendParserFactory(object):
             raise ValueError("parser class %s does not exist" % parser_class)
 
 
-class AscendParserBase(object):
+class AscendParserBase:
     def __init__(self, graph, var2geop):
         self.graph = graph
         self.var2geop = var2geop

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

@@ -53,7 +53,7 @@ def is_optimizer_op(op):
     ) & int(OpRole.Optimize)
 
 
-class CollectiveHelper(object):
+class CollectiveHelper:
     def __init__(self, role_maker, nrings=1, wait_port=True):
         self.nrings = nrings
         self.wait_port = wait_port

python/paddle/distributed/fleet/meta_optimizers/dygraph_optimizer/dygraph_sharding_optimizer.py

@@ -25,7 +25,7 @@ def _is_trainable(param):
     return not param.stop_gradient
 
 
-class DygraphShardingOptimizer(object):
+class DygraphShardingOptimizer:
     """
     A wrapper for Sharding Optimizer in Dygraph.
 

python/paddle/distributed/fleet/meta_optimizers/sharding/fp16_helper.py

@@ -23,7 +23,7 @@ from paddle.fluid import core
 __all__ = []
 
 
-class FP16Utils(object):
+class FP16Utils:
     def __init__(self):
         pass
 

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

@@ -17,7 +17,7 @@ from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY, OpRole
 __all__ = []
 
 
-class GradientClipHelper(object):
+class GradientClipHelper:
     def __init__(self, mp_ring_id):
         self.mp_ring_id = mp_ring_id
 

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

@@ -44,7 +44,7 @@ class PlaceType:
         return PlaceType.CPU
 
 
-class OffloadHelper(object):
+class OffloadHelper:
     cpu_place_type = 0
     cuda_place_type = PlaceType.default_device()
     cuda_pinned_place_type = PlaceType.default_pinned()

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

@@ -15,7 +15,7 @@
 __all__ = []
 
 
-class ProgramDeps(object):
+class ProgramDeps:
     def __init__(self, block, start_vars, end_vars):
         self._block = block
         # vars where to start to build the deps

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

@@ -22,7 +22,7 @@ from paddle.distributed.fleet.meta_optimizers.sharding.fp16_helper import (
 __all__ = []
 
 
-class Shard(object):
+class Shard:
     def __init__(
         self,
     ):
@@ -155,7 +155,7 @@ class Shard(object):
         return grads_in_shard
 
 
-class ProgramSegment(object):
+class ProgramSegment:
     def __init__(self, block):
         self._block = block
         self._allreduce_vars = []

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

@@ -408,7 +408,7 @@ def insert_allreduce_ops(
     return
 
 
-class FuseHelper(object):
+class FuseHelper:
     @staticmethod
     def sort_vars_by_dtype(block, vars_name):
         fp32_vars = []

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

@@ -17,7 +17,7 @@ from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_VAR_KEY
 __all__ = []
 
 
-class WeightDecayHelper(object):
+class WeightDecayHelper:
     def __init__(self):
         pass
 

python/paddle/distributed/fleet/meta_parallel/parallel_layers/pp_layers.py

@@ -53,7 +53,7 @@ from paddle.incubate.distributed.fleet import recompute_hybrid
 __all__ = []
 
 
-class LayerDesc(object):
+class LayerDesc:
     def __init__(self, layer_func, *inputs, **kwargs):
         self.layer_func = layer_func
         self.inputs = inputs
@@ -89,7 +89,7 @@ class SharedLayerDesc(LayerDesc):
         self.shared_weight_attr = shared_weight_attr
 
 
-class SegmentLayers(object):
+class SegmentLayers:
     def __init__(
         self,
         layers_desc,

python/paddle/distributed/fleet/runtime/runtime_base.py

@@ -15,7 +15,7 @@
 __all__ = []
 
 
-class RuntimeBase(object):
+class RuntimeBase:
     def __init__(self):
         pass
 

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

@@ -46,7 +46,7 @@ class FSShellCmdAborted(ExecuteError):
     pass
 
 
-class FS(object):
+class FS:
     @abc.abstractmethod
     def ls_dir(self, fs_path):
         raise NotImplementedError

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

@@ -128,7 +128,7 @@ class KVHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
         self.end_headers()
 
 
-class KVHTTPServer(HTTPServer, object):
+class KVHTTPServer(HTTPServer):
     """
     it is a http server storing kv pairs.
     """

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

@@ -20,7 +20,7 @@ import paddle.distributed.fleet as fleet
 import numpy as np
 
 
-class HybridParallelInferenceHelper(object):
+class HybridParallelInferenceHelper:
     """
     A helper class to split program for inference with hybrid parallelism.
 

python/paddle/distributed/launch/context/__init__.py

@@ -21,7 +21,7 @@ from .args_envs import parse_args, fetch_envs, env_args_mapping
 import logging
 
 
-class Context(object):
+class Context:
     def __init__(self, enable_plugin=True):
         self.args, self.unknown_args = parse_args()
         self.envs = fetch_envs()

python/paddle/distributed/launch/context/device.py

@@ -27,7 +27,7 @@ class DeviceType:
     CUSTOM_DEVICE = 'custom_device'
 
 
-class Device(object):
+class Device:
     def __init__(self, dtype=None, memory="", labels=""):
         self._dtype = dtype
         self._memory = memory

python/paddle/distributed/launch/context/event.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 
-class Event(object):
+class Event:
     def __init__(self, kind="status", message="", fatal=False):
         self.kind = kind
         self.message = message

python/paddle/distributed/launch/context/node.py

@@ -21,7 +21,7 @@ import struct
 from contextlib import closing
 
 
-class Node(object):
+class Node:
     def __init__(self):
         # self.device = Device.detect_device()
         self.device = Device.parse_device()

python/paddle/distributed/launch/context/resource.py

@@ -13,6 +13,6 @@
 # limitations under the License.
 
 
-class Resource(object):
+class Resource:
     def __init__(self):
         self.devices = []

python/paddle/distributed/launch/context/status.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 
-class Status(object):
+class Status:
     UNINIT = "uninit"
     READY = "ready"
     RUNNING = "running"

python/paddle/distributed/launch/controllers/controller.py

@@ -31,7 +31,7 @@ class ControleMode:
     RPC = "rpc"
 
 
-class ControllerBase(object):
+class ControllerBase:
     def __init__(self, ctx):
         signal.signal(signal.SIGTERM, self.signal_handler)
         signal.signal(signal.SIGABRT, self.signal_handler)

python/paddle/distributed/launch/controllers/master.py

@@ -24,7 +24,7 @@ import random
 ETCD_PROTOCAL = 'etcd://'
 
 
-class Master(object):
+class Master:
     '''
     Master is a distributed store design to exchange info among nodes
     '''

python/paddle/distributed/launch/controllers/watcher.py

@@ -19,7 +19,7 @@ import os
 from threading import Thread
 
 
-class Watcher(object):
+class Watcher:
     def __init__(self, ctx):
         self.ctx = ctx
 

python/paddle/distributed/launch/job/container.py

@@ -20,7 +20,7 @@ import os
 import sys
 
 
-class Container(object):
+class Container:
     '''
     TODO(kuizhiqing) A container can be run by process/thread or just a callable function
     '''

python/paddle/distributed/launch/job/job.py

@@ -19,7 +19,7 @@ class JobMode:
     HETER = 'heter'
 
 
-class Job(object):
+class Job:
     def __init__(self, jid='default', mode=JobMode.COLLECTIVE, nnodes="1"):
         self._mode = mode
         self._id = jid

python/paddle/distributed/launch/job/pod.py

@@ -20,7 +20,7 @@ import random
 import time
 
 
-class PodSepc(object):
+class PodSepc:
     def __init__(self):
         self._name = ''.join(
             random.choice('abcdefghijklmnopqrstuvwxyz') for _ in range(6)

python/paddle/distributed/launch/job/status.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 
-class Status(object):
+class Status:
     UNINIT = "uninit"
     READY = "ready"
     RUNNING = "running"

python/paddle/distributed/launch/utils/kv_client.py

@@ -16,7 +16,7 @@ import requests
 import time
 
 
-class KVClient(object):
+class KVClient:
     def __init__(self, endpoint='localhost:2379'):
         self.endpoint = (
             endpoint

python/paddle/distributed/launch/utils/kv_server.py

@@ -67,7 +67,7 @@ class KVHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
         return
 
 
-class KVServer(HTTPServer, object):
+class KVServer(HTTPServer):
     def __init__(self, port):
         super().__init__(('', port), KVHandler)
         self.kv_lock = threading.Lock()

python/paddle/distributed/launch/utils/nvsmi.py

@@ -18,7 +18,7 @@ import json
 import shutil
 
 
-class Info(object):
+class Info:
     def __repr__(self):
         return str(self.__dict__)
 

python/paddle/distributed/launch/utils/process_context.py

@@ -16,7 +16,7 @@ import subprocess
 import os, sys, signal, time
 
 
-class ProcessContext(object):
+class ProcessContext:
     def __init__(
         self,
         cmd,

python/paddle/distributed/passes/auto_parallel_amp.py

@@ -54,7 +54,7 @@ from ..auto_parallel.utils import is_forward_op, is_backward_op, is_loss_op
 world_process_group = get_world_process_group()
 
 
-class AMPState(object):
+class AMPState:
     def __init__(self, block):
         self._block = block
         self._op_fp16_dict = (

python/paddle/distributed/passes/auto_parallel_data_parallel_optimization.py

@@ -572,7 +572,7 @@ class DataParallelOptimizationPass(PassBase):
             self._logger.info("individual gradient {}".format(individual_grads))
 
 
-class GradientsGroup(object):
+class GradientsGroup:
     def __init__(self, ops, max_group_size):
         self.max_group_size = max_group_size
         self.ops = ops

python/paddle/distributed/passes/auto_parallel_fp16.py

@@ -126,7 +126,7 @@ def _keep_fp32_output(op, out_name):
     return False
 
 
-class FP16State(object):
+class FP16State:
     def __init__(
         self,
         program,

python/paddle/distributed/passes/auto_parallel_grad_clip.py

@@ -143,7 +143,7 @@ def _is_about_global_norm(
     return rank_id in complete_param_ranks
 
 
-class ClipHelper(object):
+class ClipHelper:
     def __init__(self, params_grads, rank_id, block, dist_context):
         params, _ = zip(*params_grads)
         self.params = list(params)

python/paddle/distributed/passes/auto_parallel_sharding.py

@@ -850,7 +850,7 @@ def shard_parameters(params, group_size):
     return mapping
 
 
-class ShardingInfo(object):
+class ShardingInfo:
     def __init__(self, group, rank, params_grads):
         self.group = group
         self.params_grads = dict([(p.name, (p, g)) for p, g in params_grads])

python/paddle/distributed/ps/coordinator.py

@@ -353,7 +353,7 @@ class FLClient(FLClientBase):
             f.write(str(self.train_statical_info))
 
 
-class Coordinator(object):
+class Coordinator:
     def __init__(self, ps_hosts):
         self._communicator = FLCommunicator(ps_hosts)
         self._client_selector = None

python/paddle/distributed/ps/the_one_ps.py

@@ -914,7 +914,7 @@ class fsClient:
         proto.hadoop_bin = self.fs_client_param.hadoop_bin
 
 
-class PsDescBuilder(object):
+class PsDescBuilder:
     def __init__(self, context):
         self.context = context
         self.is_sync = context['is_sync']

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

@@ -27,7 +27,7 @@ __all__ = [
 ]
 
 
-class PsProgramBuilderFactory(object):
+class PsProgramBuilderFactory:
     def __init__(self):
         pass
 

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

@@ -19,7 +19,7 @@ from paddle.distributed.fleet.base.private_helper_function import (
 from paddle.distributed.passes import new_pass
 
 
-class PsProgramBuilder(object):
+class PsProgramBuilder:
     def __init__(self, pass_ctx):
         self.pass_ctx = pass_ctx
         self.attrs = self.pass_ctx._attrs

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

@@ -88,7 +88,7 @@ class DistributedMode:
     NU = 5
 
 
-class TrainerRuntimeConfig(object):
+class TrainerRuntimeConfig:
     def __init__(self, valid_strategy):
         self.mode = None
         num_threads = os.getenv("CPU_NUM", "1")

python/paddle/distributed/spawn.py

@@ -43,7 +43,7 @@ from paddle.fluid.framework import set_flags
 __all__ = []
 
 
-class ParallelEnvArgs(object):
+class ParallelEnvArgs:
     def __init__(self):
         # Paddle cluster nodes ips, such as 192.168.0.16,192.168.0.17..
         self.cluster_node_ips = None
@@ -412,7 +412,7 @@ def _func_wrapper(func, args, error_queue, return_queue, env_dict, backend):
         sys.exit(1)
 
 
-class MultiprocessContext(object):
+class MultiprocessContext:
     def __init__(self, processes, error_queues, return_queues):
         _py_supported_check()
         self.error_queues = error_queues

python/paddle/distributed/utils/launch_utils.py

@@ -99,7 +99,7 @@ def get_gpus(selected_gpus):
     return gpus
 
 
-class Hdfs(object):
+class Hdfs:
     def __init__(self):
         self.hdfs_ugi = None
         self.hdfs_name = None
@@ -128,7 +128,7 @@ class Hdfs(object):
         return not self == n
 
 
-class Cluster(object):
+class Cluster:
     def __init__(self, hdfs):
         self.job_server = None
         self.pods = []
@@ -194,7 +194,7 @@ class Cluster(object):
         return None
 
 
-class JobServer(object):
+class JobServer:
     def __init__(self):
         self.endpoint = None
 
@@ -208,7 +208,7 @@ class JobServer(object):
         return not self == j
 
 
-class Trainer(object):
+class Trainer:
     def __init__(self):
         self.gpus = []
         self.endpoint = None
@@ -239,7 +239,7 @@ class Trainer(object):
         return self.rank
 
 
-class Pod(object):
+class Pod:
     def __init__(self):
         self.rank = None
         self.id = None
@@ -454,7 +454,7 @@ def _prepare_trainer_env(cluster, trainer, backend=None):
     return proc_env
 
 
-class TrainerProc(object):
+class TrainerProc:
     def __init__(self):
         self.proc = None
         self.log_fn = None

python/paddle/distribution/constraint.py

@@ -14,7 +14,7 @@
 import paddle
 
 
-class Constraint(object):
+class Constraint:
     """Constraint condition for random variable."""
 
     def __call__(self, value):

python/paddle/distribution/distribution.py

@@ -33,7 +33,7 @@ from paddle.fluid.framework import (
 from paddle.fluid.layers import tensor
 
 
-class Distribution(object):
+class Distribution:
     """
     The abstract base class for probability distributions. Functions are
     implemented in specific distributions.

python/paddle/distribution/kl.py

@@ -127,7 +127,7 @@ def _dispatch(cls_p, cls_q):
 
 
 @functools.total_ordering
-class _Compare(object):
+class _Compare:
     def __init__(self, *classes):
         self.classes = classes
 

python/paddle/distribution/transform.py

@@ -58,7 +58,7 @@ class Type(enum.Enum):
         return _type in (cls.BIJECTION, cls.INJECTION)
 
 
-class Transform(object):
+class Transform:
     r"""Base class for the transformations of random variables.
 
     ``Transform`` can be used to represent any differentiable and injective

python/paddle/distribution/variable.py

@@ -15,7 +15,7 @@
 from paddle.distribution import constraint
 
 
-class Variable(object):
+class Variable:
     """Random variable of probability distribution.
 
     Args:

python/paddle/fluid/average.py

@@ -39,7 +39,7 @@ def _is_number_or_matrix_(var):
     return _is_number_(var) or isinstance(var, np.ndarray)
 
 
-class WeightedAverage(object):
+class WeightedAverage:
     """
     Calculate weighted average.
 

python/paddle/fluid/backward.py

@@ -41,7 +41,7 @@ _logger = log_helper.get_logger(
 )
 
 
-class ProgramStats(object):
+class ProgramStats:
     def __init__(self, block, ops):
         self.block = block
         self.ops = ops
@@ -789,7 +789,7 @@ def _find_not_need_ops(grad_op_descs, forward_ops, input_grad_names_set):
         (set[core.OpDesc]): A set of OpDescs which should be pruned.
     """
 
-    class Var(object):
+    class Var:
         def __init__(self, var_name):
             self.var_name = var_name
             self.gen_op = None
@@ -804,7 +804,7 @@ def _find_not_need_ops(grad_op_descs, forward_ops, input_grad_names_set):
             assert isinstance(op, Op)
             self.pendding_ops.append(op)
 
-    class Op(object):
+    class Op:
         def __init__(self, op_desc):
             self.op_desc = op_desc
             self.inputs = []

python/paddle/fluid/clip.py

@@ -92,7 +92,7 @@ def _squared_l2_norm(x):
     return out
 
 
-class BaseErrorClipAttr(object):
+class BaseErrorClipAttr:
     def __str__(self):
         raise NotImplementedError()
 
@@ -177,7 +177,7 @@ def error_clip_callback(block, context):
             error_clip._append_clip_op(block, grad_n)
 
 
-class ClipGradBase(object):
+class ClipGradBase:
     def __init__(self):
         super().__init__()
 

python/paddle/fluid/communicator.py

@@ -38,7 +38,7 @@ from paddle.fluid.incubate.fleet.parameter_server.mode import DistributedMode
 __all__ = ['Communicator', 'FLCommunicator', 'LargeScaleKV']
 
 
-class Communicator(object):
+class Communicator:
     def __init__(self, mode, kwargs=None, envs=None):
         """
         Communicator is used for async distribute training in distribute_transpiler mode.
@@ -246,7 +246,7 @@ class FLCommunicator(Communicator):  ## only for coordinator
         return info_mp
 
 
-class LargeScaleKV(object):
+class LargeScaleKV:
     def __init__(self):
         self.scale_kv = core.LargeScaleKV()
 
@@ -260,7 +260,7 @@ class LargeScaleKV(object):
         return self.scale_kv.size(varname)
 
 
-class HeterClient(object):
+class HeterClient:
     def __init__(self, endpoint, previous_endpoint, trainer_id):
         self.heter_client_ = core.HeterClient(
             endpoint, previous_endpoint, trainer_id

python/paddle/fluid/compiler.py

@@ -105,7 +105,7 @@ def _should_broadcast_or_not_exists(program, var_name):
     return not is_distributed
 
 
-class CompiledProgram(object):
+class CompiledProgram:
     """
     :api_attr: Static Graph
 
@@ -567,7 +567,7 @@ class CompiledProgram(object):
         return place_list
 
 
-class IpuDynamicPatcher(object):
+class IpuDynamicPatcher:
     """
     Patcher for IPU dynamic2static support.
     """
@@ -777,7 +777,7 @@ class IpuDynamicPatcher(object):
             setattr(module, key, attr)
 
 
-class IpuStrategy(object):
+class IpuStrategy:
     """
     Help users precisely control the graph building in :code:`paddle.static.IpuCompiledProgram` .
 
@@ -1237,7 +1237,7 @@ class IpuStrategy(object):
         return self.get_option('enable_fp16')
 
 
-class IpuCompiledProgram(object):
+class IpuCompiledProgram:
     """
     The IpuCompiledProgram is used to transform a program to a ipu-target program,
     such as forward graph extraction, computing graph transformation, useless scale Ops clean, etc.

python/paddle/fluid/contrib/decoder/beam_search_decoder.py

@@ -37,7 +37,7 @@ class _DecoderType:
     BEAM_SEARCH = 2
 
 
-class InitState(object):
+class InitState:
     """
     The initial hidden state object. The state objects holds a variable, and may
     use it to initialize the hidden state cell of RNN. Usually used as input to
@@ -98,7 +98,7 @@ class InitState(object):
         return self._need_reorder
 
 
-class _MemoryState(object):
+class _MemoryState:
     def __init__(self, state_name, rnn_obj, init_state):
         self._state_name = state_name  # each is a rnn.memory
         self._rnn_obj = rnn_obj
@@ -113,7 +113,7 @@ class _MemoryState(object):
         self._rnn_obj.update_memory(self._state_mem, state)
 
 
-class _ArrayState(object):
+class _ArrayState:
     def __init__(self, state_name, block, init_state):
         self._state_name = state_name
         self._block = block
@@ -161,7 +161,7 @@ class _ArrayState(object):
         layers.array_write(state, array=self._state_array, i=self._counter)
 
 
-class StateCell(object):
+class StateCell:
     """
     The state cell class stores the hidden state of the RNN cell. A typical RNN
     cell has one or more hidden states, and one or more step inputs. This class
@@ -401,7 +401,7 @@ class StateCell(object):
         return self._cur_states[self._out_state]
 
 
-class TrainingDecoder(object):
+class TrainingDecoder:
     """
     A decoder that can only be used for training. The decoder could be
     initialized with a `StateCell` object. The computation within the RNN cell
@@ -547,7 +547,7 @@ class TrainingDecoder(object):
             )
 
 
-class BeamSearchDecoder(object):
+class BeamSearchDecoder:
     """
     A beam search decoder that can be used for inference. The decoder should be
     initialized with a `StateCell` object. The decode process can be defined

python/paddle/fluid/contrib/extend_optimizer/extend_optimizer_with_weight_decay.py

@@ -17,7 +17,7 @@ from paddle.fluid import framework as framework
 __all__ = ["extend_with_decoupled_weight_decay"]
 
 
-class DecoupledWeightDecay(object):
+class DecoupledWeightDecay:
     def __init__(self, coeff=0.0, apply_decay_param_fun=None, **kwargs):
         if not isinstance(coeff, float) and not isinstance(
             coeff, framework.Variable

python/paddle/fluid/contrib/mixed_precision/bf16/amp_lists.py

@@ -24,7 +24,7 @@ from ..fp16_lists import (
 __all__ = ["AutoMixedPrecisionListsBF16"]
 
 
-class AutoMixedPrecisionListsBF16(object):
+class AutoMixedPrecisionListsBF16:
     """
     AutoMixedPrecisionListsBF16 is a class for fp32/bf16 op types list. The lists are used for an
     algorithm which determines op's execution mode (fp32 or bf16).It can update pre-defined

python/paddle/fluid/contrib/mixed_precision/bf16/decorator.py

@@ -31,7 +31,7 @@ import warnings
 __all__ = ["decorate_bf16"]
 
 
-class OptimizerWithMixedPrecision(object):
+class OptimizerWithMixedPrecision:
     """
     Optimizer with mixed-precision (MP) training. This is a wrapper of a common
     optimizer, plus the support of mixed-precision pre-training. The object

python/paddle/fluid/contrib/mixed_precision/decorator.py

@@ -34,7 +34,7 @@ import paddle
 __all__ = ["decorate"]
 
 
-class OptimizerWithMixedPrecision(object):
+class OptimizerWithMixedPrecision:
     """
     Optimizer with mixed-precision (MP) training. This is a wrapper of a common
     optimizer, plus the support of mixed-precision pre-training. The object

python/paddle/fluid/contrib/mixed_precision/fp16_lists.py

@@ -26,7 +26,7 @@ _extra_unsupported_fp16_list = {
 }
 
 
-class AutoMixedPrecisionLists(object):
+class AutoMixedPrecisionLists:
     """
     AutoMixedPrecisionLists is a class for black/white list. It can update
     pre-defined black list and white list according to users' custom black

python/paddle/fluid/contrib/quantize/quantize_transpiler.py

@@ -81,7 +81,7 @@ def quant(x, scale, num_bits):
     return y
 
 
-class QuantizeTranspiler(object):
+class QuantizeTranspiler:
     def __init__(
         self,
         weight_bits=8,

python/paddle/fluid/contrib/slim/quantization/adaround.py

@@ -51,7 +51,7 @@ def compute_soft_rounding_np(alpha_v):
     )
 
 
-class AdaRoundLoss(object):
+class AdaRoundLoss:
     def __init__(self, reg_param=0.01, default_beta_range=(20, 2)):
         self.default_reg_param = reg_param
         self.default_beta_range = default_beta_range
@@ -111,7 +111,7 @@ class AdaRoundLoss(object):
         return beta
 
 
-class AdaRound(object):
+class AdaRound:
     def __init__(
         self,
         scale,

python/paddle/fluid/contrib/slim/quantization/imperative/ptq.py

@@ -36,7 +36,7 @@ _logger = get_logger(
 )
 
 
-class ImperativePTQ(object):
+class ImperativePTQ:
     """
     Static post training quantization.
     """