[plugin] torch ddp plugin supports sharded model checkpoint (#3775)

* [plugin] torch ddp plugin add save sharded model

* [test] fix torch ddp ckpt io test

* [test] fix torch ddp ckpt io test

* [test] fix low level zero plugin test

* [test] fix low level zero plugin test

* [test] add debug info

* [test] add debug info

* [test] add debug info

* [test] add debug info

* [test] add debug info

* [test] fix low level zero plugin test

* [test] fix low level zero plugin test

* [test] remove debug info

colossalai/booster/plugin/torch_ddp_plugin.py

@@ -1,4 +1,4 @@
-from typing import Callable, Iterator, List, Tuple, Union
+from typing import Callable, Iterator, List, Optional, Tuple, Union
 
 import torch.nn as nn
 from torch.nn.parallel import DistributedDataParallel as DDP
@@ -50,6 +50,16 @@ class TorchDDPCheckpointIO(GeneralCheckpointIO):
         if self.coordinator.is_master():
             super().save_lr_scheduler(lr_scheduler, checkpoint)
 
+    def save_sharded_model(self,
+                           model: nn.Module,
+                           checkpoint_path: str,
+                           gather_dtensor: bool = False,
+                           variant: Optional[str] = None,
+                           max_shard_size: int = 1024,
+                           use_safetensors: bool = False):
+        if self.coordinator.is_master():
+            super().save_sharded_model(model, checkpoint_path, gather_dtensor, variant, max_shard_size, use_safetensors)
+
 
 class TorchDDPModel(ModelWrapper):
 

colossalai/checkpoint_io/checkpoint_io_base.py

@@ -1,7 +1,6 @@
 from abc import ABC, abstractmethod
 from pathlib import Path
-from typing import Union
+from typing import Optional, Union
-from typing import Optional
 
 import torch
 import torch.nn as nn
@@ -84,7 +83,6 @@ class CheckpointIO(ABC):
         # containing no distributed tensors, dtensor -> full tensor conversion
         # should be done offline via our CLI
         # the existence of index file means it is a sharded checkpoint
-        ckpt_path = Path(checkpoint)
         index_file_exists, index_file_path = has_index_file(checkpoint)
 
         # return the origin model instead of the unwrapped model

colossalai/checkpoint_io/utils.py

@@ -1,10 +1,12 @@
 # coding=utf-8
+import re
 from pathlib import Path
+from typing import Iterator, List, Mapping, Optional, OrderedDict, Tuple
+
 import torch
 import torch.nn as nn
-from typing import List, Mapping, OrderedDict, Optional, Tuple, Iterator
+
 from colossalai.tensor.d_tensor.d_tensor import DTensor
-import re
 
 SAFE_WEIGHTS_NAME = "model.safetensors"
 WEIGHTS_NAME = "pytorch_model.bin"
@@ -15,6 +17,7 @@ WEIGHTS_INDEX_NAME = "pytorch_model.bin.index.json"
 # General helper functions
 # ======================================
 
+
 def calculate_tensor_size(tensor: torch.Tensor) -> float:
     """
     Calculate the size of a parameter in MB. Used to compute whether a group of params exceed the shard size.
@@ -28,6 +31,7 @@ def calculate_tensor_size(tensor: torch.Tensor) -> float:
     """
     return tensor.numel() * tensor.element_size() / 1024 / 1024
 
+
 def is_safetensors_available() -> bool:
     """
     Check whether safetensors is available.
@@ -78,7 +82,6 @@ def is_safetensor_checkpoint(checkpoint_file_path: str) -> bool:
 # Helper functions for saving shard file
 # ======================================
 def shard_checkpoint(state_dict: torch.Tensor, max_shard_size: int = 1024) -> Iterator[Tuple[OrderedDict, int]]:
- 
     """
     Splits a model state dictionary in sub-checkpoints so that the final size of each sub-checkpoint does not exceed a
     given size.
@@ -107,26 +110,30 @@ def shard_checkpoint(state_dict: torch.Tensor, max_shard_size: int = 1024) -> It
     yield current_block, current_block_size
 
 
-def load_shard_state_dict(checkpoint_file: Path, use_safetensors: bool =False):
+def load_shard_state_dict(checkpoint_file: Path, use_safetensors: bool = False):
     """
     load shard state dict into model
     """
     if use_safetensors and not checkpoint_file.suffix == ".safetensors":
         raise Exception("load the model using `safetensors`, but no file endwith .safetensors")
     if use_safetensors:
-        from safetensors.torch import safe_open
         from safetensors.torch import load_file as safe_load_file
+        from safetensors.torch import safe_open
         with safe_open(checkpoint_file, framework="pt") as f:
             metadata = f.metadata()
         if metadata["format"] != "pt":
             raise NotImplementedError(
-                f"Conversion from a {metadata['format']} safetensors archive to PyTorch is not implemented yet."
+                f"Conversion from a {metadata['format']} safetensors archive to PyTorch is not implemented yet.")
-            )
         return safe_load_file(checkpoint_file)
     else:
         return torch.load(checkpoint_file)
 
-def load_state_dict_into_model(model: nn.Module, state_dict: torch.Tensor, missing_keys: List, strict: bool = False, load_sub_module: bool = True):
+
+def load_state_dict_into_model(model: nn.Module,
+                               state_dict: torch.Tensor,
+                               missing_keys: List,
+                               strict: bool = False,
+                               load_sub_module: bool = True):
     r"""Copies parameters and buffers from :attr:`state_dict` into
     this module and its descendants.
 
@@ -166,11 +173,12 @@ def load_state_dict_into_model(model: nn.Module, state_dict: torch.Tensor, missi
 
     if strict:
         if len(unexpected_keys) > 0:
-            error_msgs = 'Unexpected key(s) in state_dict: {}. '.format(
+            error_msgs = 'Unexpected key(s) in state_dict: {}. '.format(', '.join(
-                        ', '.join('"{}"'.format(k) for k in unexpected_keys))
+                '"{}"'.format(k) for k in unexpected_keys))
             raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
                 model.__class__.__name__, "\n\t".join(error_msgs)))
 
+
 # ======================================
 # Helper functions for saving state dict
 # ======================================
@@ -350,6 +358,8 @@ def has_index_file(checkpoint_path: str) -> Tuple[bool, Optional[Path]]:
             return True, index_files[0]
         else:
             return False, None
+    else:
+        raise RuntimeError(f'Invalid checkpoint path {checkpoint_path}. Expected a file or a directory.')
 
 
 def load_state_dict(checkpoint_file_path: Path):
@@ -382,7 +392,6 @@ def load_state_dict(checkpoint_file_path: Path):
         return torch.load(checkpoint_file_path)
 
 
-
 def add_variant(weights_name: str, variant: Optional[str] = None) -> str:
     if variant is not None and len(variant) > 0:
         splits = weights_name.split(".")
@@ -392,7 +401,7 @@ def add_variant(weights_name: str, variant: Optional[str] = None) -> str:
     return weights_name
 
 
-def get_base_filenames(variant: str=None, use_safetensors: bool=False):
+def get_base_filenames(variant: str = None, use_safetensors: bool = False):
     """
     generate base weight filenames
     """
@@ -404,6 +413,7 @@ def get_base_filenames(variant: str=None, use_safetensors: bool=False):
 
     return weights_name, save_index_file
 
+
 def get_shard_filename(weights_name: str, idx: int):
     """
     get shard file name

tests/test_booster/test_plugin/test_low_level_zero_plugin.py

@@ -11,9 +11,9 @@ from colossalai.testing import parameterize, rerun_if_address_is_in_use, spawn
 from tests.kit.model_zoo import model_zoo
 
 # These models are not compatible with AMP
-_AMP_ERR_MODELS = ['timm_convit', 'dlrm', 'deepfm_interactionarch', 'deepfm_simpledeepfmnn`']
+_AMP_ERR_MODELS = ['timm_convit', 'dlrm', 'deepfm_interactionarch', 'deepfm_simpledeepfmnn']
 # These models have no parameters
-_LOW_LEVEL_ZERO_ERR_MODELS = ['dlrm_interactionarch']
+_LOW_LEVEL_ZERO_ERR_MODELS = ['dlrm_interactionarch', 'deepfm_overarch', 'deepfm_sparsearch', 'dlrm_sparsearch']
 # These models will get stuck
 _STUCK_MODELS = [
     'diffusers_vq_model', 'transformers_albert', 'transformers_albert_for_pretraining', 'transformers_bert',
@@ -67,6 +67,7 @@ def check_low_level_zero_plugin(stage: int, early_stop: bool = True):
             skipped_models.append(name)
             continue
         err = run_fn(stage, model_fn, data_gen_fn, output_transform_fn)
+
         torch.cuda.empty_cache()
 
         if err is None:
@@ -91,7 +92,7 @@ def run_dist(rank, world_size, port, early_stop: bool = True):
 
 @rerun_if_address_is_in_use()
 def test_low_level_zero_plugin(early_stop: bool = True):
-    spawn(run_dist, 2, early_stop=early_stop)
+    spawn(run_dist, 4, early_stop=early_stop)
 
 
 if __name__ == '__main__':

tests/test_checkpoint_io/test_torch_ddp_checkpoint_io.py

@@ -1,6 +1,7 @@
 import tempfile
 
 import torch
+import torch.distributed as dist
 from torch.nn.parallel import DistributedDataParallel as DDP
 from torch.optim import SGD
 from torchvision.models import resnet18
@@ -8,12 +9,12 @@ from torchvision.models import resnet18
 import colossalai
 from colossalai.booster import Booster
 from colossalai.booster.plugin import TorchDDPPlugin
-from colossalai.booster.plugin.torch_ddp_plugin import TorchDDPCheckpointIO
 from colossalai.interface import OptimizerWrapper
-from colossalai.testing import check_state_dict_equal, rerun_if_address_is_in_use, spawn
+from colossalai.testing import check_state_dict_equal, parameterize, rerun_if_address_is_in_use, spawn
 
 
-def check_torch_ddp_checkpointIO():
+@parameterize('shard', [True, False])
+def check_torch_ddp_checkpointIO(shard: bool):
     plugin = TorchDDPPlugin()
     booster = Booster(plugin=plugin)
     model = resnet18()
@@ -34,24 +35,39 @@ def check_torch_ddp_checkpointIO():
     optimizer.step()
     scheduler.step()
 
-    optimizer_ckpt_tempfile = tempfile.NamedTemporaryFile()
+    with tempfile.TemporaryDirectory() as tempdir:
-    lr_scheduler_ckpt_tempfile = tempfile.NamedTemporaryFile()
+        obj = [tempdir]
-    ckpt_io = TorchDDPCheckpointIO()
+        dist.broadcast_object_list(obj, src=0)
-    ckpt_io.save_optimizer(optimizer, optimizer_ckpt_tempfile.name)
+        tempdir = obj[0]    # use the same directory on all ranks
-    ckpt_io.save_lr_scheduler(scheduler, lr_scheduler_ckpt_tempfile.name)
+
+        model_ckpt_path = f"{tempdir}/model"
+        optimizer_ckpt_path = f"{tempdir}/optimizer"
+        lr_scheduler_ckpt_path = f"{tempdir}/lr_scheduler"
+        booster.save_model(model, model_ckpt_path, shard=shard)
+        if not shard:
+            # TODO(ver217): optimizer checkpointing is not supported for sharded checkpoint
+            booster.save_optimizer(optimizer, optimizer_ckpt_path)
+            booster.save_lr_scheduler(scheduler, lr_scheduler_ckpt_path)
+        dist.barrier()
 
         new_model = resnet18()
         new_optimizer = SGD((new_model.parameters()), lr=0.001)
         new_scheduler = torch.optim.lr_scheduler.StepLR(new_optimizer, step_size=1, gamma=0.1)
-    _, new_optimizer, _, _, new_scheduler = booster.boost(new_model, new_optimizer, lr_scheduler=new_scheduler)
+        new_model, new_optimizer, _, _, new_scheduler = booster.boost(new_model,
+                                                                      new_optimizer,
+                                                                      lr_scheduler=new_scheduler)
 
-    if ckpt_io.coordinator.is_master():
+        booster.load_model(new_model, model_ckpt_path)
-        ckpt_io.load_optimizer(new_optimizer, optimizer_ckpt_tempfile.name)
+        check_state_dict_equal(model.state_dict(), new_model.state_dict(), False)
+
+        if not shard:
+            booster.load_optimizer(new_optimizer, optimizer_ckpt_path)
             check_state_dict_equal(optimizer.state_dict(), new_optimizer.state_dict(), False)
-
+            booster.load_lr_scheduler(new_scheduler, lr_scheduler_ckpt_path)
-        ckpt_io.load_lr_scheduler(new_scheduler, lr_scheduler_ckpt_tempfile.name)
             check_state_dict_equal(scheduler.state_dict(), new_scheduler.state_dict(), False)
 
+        dist.barrier()
+
 
 def run_dist(rank, world_size, port):
     colossalai.launch(config=(dict()), rank=rank, world_size=world_size, port=port, host='localhost')