[Auto Parallel] Support Iterable dataset for auto parallel (#45518)

* support iterable dataset for auto parallel

* add split_data proto

* fix unittest bug

* fix recompute bug

* update cmake

paddle/fluid/framework/distributed_strategy.proto

@@ -343,6 +343,7 @@ message DistributedStrategy {
   optional bool is_fl_ps_mode = 39 [ default = false ];
   optional bool with_coordinator = 40 [ default = false ];
   optional bool qat = 41 [ default = false ];
+  optional bool split_data = 42 [ default = true ];
 
   optional RecomputeConfig recompute_configs = 101;
   optional AMPConfig amp_configs = 102;

python/paddle/distributed/auto_parallel/dist_loader.py

@@ -14,10 +14,13 @@
 
 import abc
 import numpy as np
+from functools import wraps
+
 import paddle
 from .utils import to_list
 from paddle.fluid.layers.utils import flatten
 from paddle.io import DataLoader, BatchSampler, IterableDataset
+from paddle.fluid.dataloader.batch_sampler import _InfiniteIterableSampler
 from paddle.fluid.dataloader.dataloader_iter import _DatasetKind, default_collate_fn, default_convert_fn
 
 
@@ -29,33 +32,41 @@ class DistributedDataLoader(metaclass=abc.ABCMeta):
                  epochs=1,
                  data_parallel_world_size=None,
                  data_parallel_rank=None,
-                 drop_last=False):
+                 drop_last=False,
+                 split_data=True):
         if isinstance(dataset, IterableDataset):
-            raise TypeError("IterableDataset is not supported.")
+            self.dataset_kind = _DatasetKind.ITER
         else:
             self.dataset_kind = _DatasetKind.MAP
 
         self.dataset = dataset
         self.epochs = epochs
         self.drop_lost = drop_last
+        self.data_parallel_world_size = data_parallel_world_size
+        self.data_parallel_rank = data_parallel_rank
+        self.split_data = split_data
 
         if batch_size is None:
             self.batch_size = None
             self.batch_sampler = None
         else:
             if data_parallel_world_size is not None:
-                assert batch_size % data_parallel_world_size == 0, \
-                    "'batch_size' must be divisible by data parallel size"
+                for dp_world_size in data_parallel_world_size:
+                    if dp_world_size is not None:
+                        assert batch_size % dp_world_size == 0, \
+                            "batch_size must be divisible by dp_world_size value {}".format(str(dp_world_size))
             self.batch_size = batch_size
-            self.batch_sampler = BatchSampler(dataset,
-                                              batch_size=batch_size,
-                                              shuffle=False,
-                                              drop_last=drop_last)
+            if isinstance(dataset, IterableDataset):
+                self.batch_sampler = _InfiniteIterableSampler(
+                    dataset, batch_size)
+            else:
+                self.batch_sampler = BatchSampler(dataset,
+                                                  batch_size=batch_size,
+                                                  shuffle=False,
+                                                  drop_last=drop_last)
 
         self.auto_collate_batch = self.batch_sampler is not None
         self.sampler_iter = iter(self.index_sampler)
-        self.dp_world_size = 1 if data_parallel_world_size is None else data_parallel_world_size
-        self.dp_rank = 0 if data_parallel_rank is None else data_parallel_rank
 
     @abc.abstractmethod
     def __iter__(self):
@@ -73,7 +84,7 @@ class DistributedDataLoader(metaclass=abc.ABCMeta):
             if self.dataset_kind == _DatasetKind.MAP:
                 return list(range(len(self.dataset)))
             else:
-                raise TypeError("Only support datasets in map-style.")
+                return _InfiniteIterableSampler(self.dataset, 1)
 
 
 class NonIterableGeneratorLoader(DistributedDataLoader):
@@ -88,7 +99,8 @@ class NonIterableGeneratorLoader(DistributedDataLoader):
                  collate_fn=None,
                  data_parallel_world_size=None,
                  data_parallel_rank=None,
-                 drop_last=False):
+                 drop_last=False,
+                 split_data=True):
         self.feed_list = feed_list
         self.places = places
         self.steps_per_epoch = steps_per_epoch
@@ -96,7 +108,7 @@ class NonIterableGeneratorLoader(DistributedDataLoader):
         super(NonIterableGeneratorLoader,
               self).__init__(dataset, batch_size, epochs,
                              data_parallel_world_size, data_parallel_rank,
-                             drop_last)
+                             drop_last, split_data)
 
         if self.auto_collate_batch:
             self.collate_fn = collate_fn or default_collate_fn
@@ -115,17 +127,22 @@ class NonIterableGeneratorLoader(DistributedDataLoader):
         return self
 
     def __next__(self):
-        if self._cur_step < self._steps:
+        if not self._steps:
+            self._cur_step += 1
+        elif self._cur_step < self._steps:
             self._cur_step += 1
         else:
             self._inner_dataloader.reset()
+            self.sampler_iter = iter(self.index_sampler)
             raise StopIteration
 
     def _infer_steps(self):
         if self.steps_per_epoch is not None:
             return self.steps_per_epoch
         try:
-            if self.batch_size is None:
+            if isinstance(self.dataset, IterableDataset):
+                steps_per_epoch = None
+            elif self.batch_size is None:
                 steps_per_epoch = len(self.dataset)
             else:
                 steps_per_epoch = len(self.dataset) // self.batch_size
@@ -138,26 +155,61 @@ class NonIterableGeneratorLoader(DistributedDataLoader):
     def _create_inner_dataloader(self):
 
         def sample_data_generator():
-            for indices in self.sampler_iter:
-                assert len(indices) % self.dp_world_size == 0, \
-                    "Please set batch_size to be divisible by data parallel size"
-                n = len(indices) // self.dp_world_size
-                cur_indices = [
-                    indices[i:i + n] for i in range(0, len(indices), n)
-                ]
-                batch = self.dataset_fetcher.fetch(cur_indices[self.dp_rank])
-                yield batch[:len(self.feed_list)]
+            while True:
+                try:
+                    indices = next(self.sampler_iter)
+                    batch = self.dataset_fetcher.fetch(indices)
+                    if batch is None: break
+
+                except StopIteration:
+                    self.dataset_fetcher = _DatasetKind.create_fetcher(
+                        self.dataset_kind, self.dataset,
+                        self.auto_collate_batch, self.collate_fn,
+                        self.drop_lost)
+                    break
+
+                partial_data = []
+                for i, d in enumerate(batch[:len(self.feed_list)]):
+                    array = np.array(d)
+                    if not self.split_data:
+                        partial_data.append(array)
+                    elif self.dp_world_sizes[i] is not None:
+                        partial_data.append(
+                            np.split(array,
+                                     self.dp_world_sizes[i])[self.dp_ranks[i]])
+                    else:
+                        partial_data.append(array)
+                yield partial_data
 
         def batch_data_generator():
-            for indices in self.sampler_iter:
+            while True:
+                try:
+                    indices = next(self.sampler_iter)
+
+                    batch = self.dataset_fetcher.fetch(indices)
+                    if batch is None: break
+                except StopIteration:
+                    break
+
                 partial_data = []
-                batch = self.dataset_fetcher.fetch(indices)
-                for data in batch:
-                    assert data.shape[0] % self.dp_world_size == 0, \
-                        "Please padding dataset's batch_size to be divisible by data parallel size"
-                    partial_data.append(
-                        np.split(data, self.dp_world_size)[self.dp_rank])
-                yield partial_data[:len(self.feed_list)]
+                for i, d in enumerate(batch[:len(self.feed_list)]):
+                    array = np.array(d)
+                    if not self.split_data:
+                        partial_data.append(array)
+                    elif self.dp_world_sizes[i] is not None:
+                        partial_data.append(
+                            np.split(array,
+                                     self.dp_world_sizes[i])[self.dp_ranks[i]])
+                    else:
+                        partial_data.append(array)
+                yield partial_data
+
+        self.dp_world_sizes = [
+            1 for _ in range(len(self.feed_list))
+        ] if self.data_parallel_world_size is None else self.data_parallel_world_size
+        self.dp_ranks = [
+            0 for _ in range(len(self.feed_list))
+        ] if self.data_parallel_rank is None else self.data_parallel_rank
 
         dataloader = paddle.fluid.io.DataLoader.from_generator(
             feed_list=self.feed_list, capacity=70, iterable=False)

python/paddle/distributed/auto_parallel/engine.py

@@ -233,8 +233,8 @@ class Engine:
         assert "dataset" in self._user_tuning_config, "Optimization Tuning should provide with dataset."
         batch_size = self._user_tuning_config["batch_size"]
         dataset = self._user_tuning_config["dataset"]
-        dataset.dp_world_size = self._input_split_size
-        dataset.dp_rank = self._input_split_rank
+        dataset.dp_world_size = self.dp_world_sizes
+        dataset.dp_rank = self.dp_ranks
 
         from .tuner.optimization_tuner import OptimizationTuner
         self._optimization_tuner = OptimizationTuner(self._user_tuning_config,
@@ -272,8 +272,13 @@ class Engine:
             if var.name in block.vars:
                 feed_list.append(block.vars[var.name])
 
-        self._input_split_size, self._input_split_rank = self._get_input_split_info(
-            feed_list[0], self._dist_contexts[mode])
+        self.dp_world_sizes = []
+        self.dp_ranks = []
+        for feed_var in feed_list:
+            dp_world_size, dp_rank = self._get_input_split_info(
+                feed_var, self._dist_contexts[mode])
+            self.dp_world_sizes.append(dp_world_size)
+            self.dp_ranks.append(dp_rank)
 
     def _parallel(self, mode, all_ranks):
         # Parallelize program based on the planner's results
@@ -440,15 +445,23 @@ class Engine:
         for epoch in range(epochs):
             train_logs = {"epoch: {:d} ": epoch}
             for step, _ in enumerate(train_dataloader):
+                try:
+                    outs = self._executor.run(self.main_program,
+                                              fetch_list=fetch_list,
+                                              use_program_cache=use_cache,
+                                              return_numpy=return_numpy)
+                except fluid.core.EOFException:
+                    break
 
-                outs = self._executor.run(self.main_program,
-                                          fetch_list=fetch_list,
-                                          use_program_cache=use_cache,
-                                          return_numpy=return_numpy)
                 train_logs["step: {:d} "] = step
                 if lr_scheduler is not None:
                     lr_scheduler.step()
-                    train_logs["lr: {:5e} "] = self._lr_optimizer.get_lr()
+                    try:
+                        train_logs["lr: {:5e} "] = self._lr_optimizer.get_lr()
+                    except:
+                        train_logs[
+                            "lr: {:5e} "] = self._lr_optimizer._learning_rate.get_lr(
+                            )
                 # inner fetches
                 if fetch_loss:
                     train_logs["loss: {:9f} "] = outs[0][0]
@@ -486,10 +499,13 @@ class Engine:
 
         for step, _ in enumerate(eval_dataloader):
             eval_logs = {"step: {:d} ": step}
-            outs = self._executor.run(self.main_program,
-                                      fetch_list=fetch_list,
-                                      use_program_cache=use_cache,
-                                      return_numpy=return_numpy)
+            try:
+                outs = self._executor.run(self.main_program,
+                                          fetch_list=fetch_list,
+                                          use_program_cache=use_cache,
+                                          return_numpy=return_numpy)
+            except fluid.core.EOFException:
+                break
             # inner fetches
             if fetch_loss:
                 eval_logs["loss: {:9f} "] = outs[0][0]
@@ -534,10 +550,13 @@ class Engine:
         outputs = []
         for step, _ in enumerate(test_dataloader):
             predict_logs = {"step: {:d} ": step}
-            outs = self._executor.run(self.main_program,
-                                      fetch_list=fetch_list,
-                                      use_program_cache=use_cache,
-                                      return_numpy=return_numpy)
+            try:
+                outs = self._executor.run(self.main_program,
+                                          fetch_list=fetch_list,
+                                          use_program_cache=use_cache,
+                                          return_numpy=return_numpy)
+            except fluid.core.EOFException:
+                break
             outputs.append(outs[:len(fetch_outputs)])
             for i, out in enumerate(outs):
                 predict_logs[fetch_map[fetch_list[i]] + ": {}"] = out
@@ -586,8 +605,9 @@ class Engine:
                 epochs,
                 steps_per_epoch,
                 collate_fn,
-                data_parallel_world_size=self._input_split_size,
-                data_parallel_rank=self._input_split_rank)
+                data_parallel_world_size=self.dp_world_sizes,
+                data_parallel_rank=self.dp_ranks,
+                split_data=self.strategy.split_data)
 
         # move read op from the end of program to the start of program
         new_op_size = len(dist_main_block.ops)
@@ -682,6 +702,8 @@ class Engine:
                     self.model, "gpt"
             ) and self.model.__class__.__name__ == 'GPTForPretraining':
                 exact_ckpts = self.model.gpt.checkpoints
+            else:
+                exact_ckpts = config["checkpoints"]
         else:
             exact_ckpts = config["checkpoints"]
 

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

@@ -1997,6 +1997,28 @@ class DistributedStrategy(object):
         else:
             print("WARNING: auto-search should have value of bool type")
 
+    @property
+    def split_data(self):
+        """
+        Indicating whether we split the data. If True, we split the data.
+        Default Value: True
+        Examples:
+          .. code-block:: python
+            import paddle
+            paddle.enable_static()
+            import paddle.distributed.fleet as fleet
+            strategy = fleet.DistributedStrategy()
+            strategy.split_data = True
+        """
+        return self.strategy.split_data
+
+    @split_data.setter
+    def split_data(self, flag):
+        if isinstance(flag, bool):
+            self.strategy.split_data = flag
+        else:
+            print("WARNING: split_data should have value of bool type")
+
     @property
     def qat(self):
         """

python/paddle/fluid/tests/unittests/auto_parallel/CMakeLists.txt

@@ -72,4 +72,9 @@ if(WITH_DISTRIBUTE AND WITH_GPU)
   py_test_modules(test_lr_grad_clip MODULES test_lr_grad_clip)
   py_test_modules(test_quantization MODULES test_quantization)
   py_test_modules(test_dist_matmul MODULES test_dist_matmul)
+
+  py_test_modules(test_iterable_dataset MODULES test_iterable_dataset ENVS
+                  ${dist_ENVS})
+  set_tests_properties(test_iterable_dataset
+                       PROPERTIES LABELS "RUN_TYPE=EXCLUSIVE" TIMEOUT 80)
 endif()

python/paddle/fluid/tests/unittests/auto_parallel/iterable_dataset.py

+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import unittest
+import time
+import tempfile
+import copy
+import os
+import numpy as np
+import subprocess
+import paddle
+import paddle.nn as nn
+import paddle.fluid as fluid
+import paddle.static as static
+import paddle.nn.functional as F
+import paddle.utils as utils
+from paddle.fluid import layers
+from paddle.io import Dataset, IterableDataset, DataLoader
+from paddle.static import InputSpec
+from paddle.distributed import fleet
+import paddle.distributed.auto_parallel as auto
+from paddle.distributed.auto_parallel.engine import Engine
+from paddle.optimizer.lr import CosineAnnealingDecay
+from paddle.fluid.dataloader.collate import default_collate_fn
+
+paddle.enable_static()
+global_process_mesh = auto.ProcessMesh(mesh=[0, 1])
+PP_MESH_0 = auto.ProcessMesh([0])
+PP_MESH_1 = auto.ProcessMesh([1])
+batch_size = 2
+batch_num = 10
+hidden_size = 1024
+sequence_len = 512
+image_size = hidden_size
+class_num = 10
+
+paddle.seed(44)
+
+
+class MyDataset(IterableDataset):
+
+    def __init__(self, num_samples):
+        super(MyDataset, self).__init__()
+        self.num_samples = num_samples
+
+    def __iter__(self):
+        for i in range(self.num_samples):
+            input = np.random.uniform(size=image_size).astype("float32")
+            label = np.random.randint(0, class_num - 1, dtype="int64")
+            yield input, label
+
+
+class MyDataset1(Dataset):
+
+    def __init__(self, num_samples):
+        super(MyDataset1, self).__init__()
+        self.num_samples = num_samples
+        self.data = []
+        for i in range(self.num_samples):
+            input1 = np.random.uniform(size=image_size).astype("float32")
+            label1 = np.array(np.random.randint(0, class_num - 1,
+                                                dtype="int64"))
+            input2 = np.random.uniform(size=image_size).astype("float32")
+            label2 = np.array(np.random.randint(0, class_num - 1,
+                                                dtype="int64"))
+            input = np.stack((input1, input2))
+            label = np.stack((label1, label2))
+            self.data.append((input, label))
+
+    def __getitem__(self, idx):
+        return self.data[idx]
+
+    def __len__(self):
+        return len(self.data)
+
+
+class MLPLayer(nn.Layer):
+
+    def __init__(self,
+                 hidden_size=1024,
+                 intermediate_size=4 * 1024,
+                 dropout_ratio=0.1,
+                 initializer_range=0.02):
+        super(MLPLayer, self).__init__()
+        d_model = hidden_size
+        dim_feedforward = intermediate_size
+        weight_attr = paddle.ParamAttr(
+            initializer=nn.initializer.Normal(mean=0.0, std=initializer_range))
+        bias_attr = None
+
+        self.linear0 = nn.Linear(d_model,
+                                 dim_feedforward,
+                                 weight_attr,
+                                 bias_attr=bias_attr)
+        self.linear1 = nn.Linear(dim_feedforward,
+                                 d_model,
+                                 weight_attr,
+                                 bias_attr=bias_attr)
+        self.linear2 = nn.Linear(d_model, 1, weight_attr, bias_attr=bias_attr)
+        self.norm = nn.LayerNorm(d_model, epsilon=1e-5)
+        self.dropout = nn.Dropout(dropout_ratio, mode="upscale_in_train")
+
+    def forward(self, input):
+        out = auto.shard_op(self.norm, dist_attr={"process_mesh":
+                                                  PP_MESH_0})(input)
+        out = self.linear0(out)
+        out = F.gelu(out, approximate=True)
+        out = auto.shard_op(self.linear1, dist_attr={"process_mesh":
+                                                     PP_MESH_1})(out)
+        out = self.dropout(out)
+        out = self.linear2(out)
+        self.out = out
+        return out
+
+
+def train(fetch):
+    mlp = MLPLayer(hidden_size=hidden_size,
+                   intermediate_size=4 * hidden_size,
+                   dropout_ratio=0.1,
+                   initializer_range=0.02)
+    loss = paddle.nn.CrossEntropyLoss()
+    optimizer = paddle.optimizer.Adam(learning_rate=0.00001,
+                                      beta1=0.9,
+                                      beta2=0.999,
+                                      epsilon=1e-08,
+                                      grad_clip=None)
+
+    inputs_spec = InputSpec([batch_size, hidden_size], 'float32', 'x')
+    labels_spec = InputSpec([batch_size], 'int64', 'label')
+
+    dist_strategy = fleet.DistributedStrategy()
+    dist_strategy.semi_auto = True
+    dist_strategy.split_data = True
+    fleet.init(is_collective=True, strategy=dist_strategy)
+
+    # init engine
+    engine = Engine(mlp,
+                    inputs_spec=inputs_spec,
+                    labels_spec=labels_spec,
+                    strategy=dist_strategy)
+    engine.prepare(optimizer, loss, metrics=paddle.metric.Accuracy())
+
+    # fetch
+    if fetch:
+        fetches = {'out': mlp.out}
+    else:
+        fetches = None
+
+    # train
+    train_dataset = MyDataset(batch_num * batch_size)
+    train_dataset1 = MyDataset1(batch_num)
+    engine.fit(train_dataset,
+               epochs=2,
+               batch_size=batch_size,
+               steps_per_epoch=batch_num,
+               fetches=fetches)
+
+    engine.fit(train_dataset1,
+               epochs=2,
+               batch_size=None,
+               steps_per_epoch=batch_num,
+               fetches=fetches)
+
+    # eval
+    eval_dataset = MyDataset(batch_size)
+    engine.evaluate(eval_dataset, batch_size, fetches=fetches)
+
+    # predict
+    test_dataset = MyDataset(batch_size)
+    engine.predict(test_dataset, batch_size, fetches=fetches)
+
+    # save
+    temp_dir = tempfile.TemporaryDirectory()
+    model_filename = os.path.join(temp_dir.name, 'mlp_inf')
+    engine.save(model_filename, training=False, mode='predict')
+    temp_dir.cleanup()
+
+
+if __name__ == "__main__":
+    train(fetch=True)

python/paddle/fluid/tests/unittests/auto_parallel/test_iterable_dataset.py

+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import tempfile
+import unittest
+import os
+import sys
+import shutil
+import subprocess
+from paddle.distributed.fleet.launch_utils import run_with_coverage
+
+
+class TestEngineAPI(unittest.TestCase):
+
+    def test_engine_api(self):
+        file_dir = os.path.dirname(os.path.abspath(__file__))
+        launch_model_path = os.path.join(file_dir, "iterable_dataset.py")
+
+        if os.environ.get("WITH_COVERAGE", "OFF") == "ON":
+            coverage_args = ["-m", "coverage", "run", "--branch", "-p"]
+        else:
+            coverage_args = []
+
+        tmp_dir = tempfile.TemporaryDirectory()
+        cmd = [sys.executable, "-u"] + coverage_args + [
+            "-m", "paddle.distributed.launch", "--devices", "0,1", "--log_dir",
+            tmp_dir.name, launch_model_path
+        ]
+
+        process = subprocess.Popen(cmd)
+        process.wait()
+        self.assertEqual(process.returncode, 0)
+
+        tmp_dir.cleanup()
+
+
+if __name__ == "__main__":
+    unittest.main()