[feat] Enabling ViT in OSS benchmarks (#322)

benchmarks/oss.py

@@ -21,7 +21,7 @@ from torch.nn.parallel import DistributedDataParallel as DDP
 from torch.utils.data import BatchSampler, DataLoader, Sampler
 from torch.utils.data.distributed import DistributedSampler
 from torchvision.datasets import MNIST
-from torchvision.transforms import ToTensor
+from torchvision.transforms import Compose, Resize, ToTensor
 
 from fairscale.nn.data_parallel import ShardedDataParallel as ShardedDDP
 from fairscale.optim import OSS
@@ -39,7 +39,11 @@ def dist_init(rank, world_size, backend):
 def get_problem(rank, world_size, batch_size, device, model_name: str):
     # Select the desired model on the fly
     logging.info(f"Using {model_name} for benchmarking")
-    model = getattr(importlib.import_module("torchvision.models"), model_name)(pretrained=False).to(device)
+
+    try:
+        model = getattr(importlib.import_module("torchvision.models"), model_name)(pretrained=False).to(device)
+    except AttributeError:
+        model = getattr(importlib.import_module("timm.models"), model_name)(pretrained=False).to(device)
 
     # Data setup, duplicate the grey channels to get pseudo color
     def collate(inputs: List[Any]):
@@ -48,7 +52,16 @@ def get_problem(rank, world_size, batch_size, device, model_name: str):
             "label": torch.tensor([i[1] for i in inputs]).to(device),
         }
 
-    dataset = MNIST(transform=ToTensor(), download=False, root=TEMPDIR)
+    # Transforms
+    transforms = []
+    if model_name.startswith("vit"):
+        # ViT models are fixed size. Add a ad-hoc transform to resize the pictures accordingly
+        pic_size = int(model_name.split("_")[-1])
+        transforms.append(Resize(pic_size))
+
+    transforms.append(ToTensor())
+
+    dataset = MNIST(transform=Compose(transforms), download=False, root=TEMPDIR)
     sampler: Sampler = DistributedSampler(dataset, num_replicas=world_size, rank=rank)
     batch_sampler = BatchSampler(sampler, batch_size, drop_last=True)
     dataloader = DataLoader(dataset=dataset, batch_sampler=batch_sampler, collate_fn=collate)
@@ -88,7 +101,7 @@ def train(
         torch.backends.cudnn.benchmark = False
 
     device = torch.device("cpu") if args.cpu else torch.device(rank)
-    model, dataloader, loss_fn = get_problem(rank, args.world_size, args.batch_size, device, args.torchvision_model)
+    model, dataloader, loss_fn = get_problem(rank, args.world_size, args.batch_size, device, args.model)
 
     # Shard the optimizer
     optimizer: Optional[torch.optim.Optimizer] = None
@@ -259,7 +272,7 @@ if __name__ == "__main__":
     parser.add_argument("--gloo", action="store_true", default=False)
     parser.add_argument("--profile", action="store_true", default=False)
     parser.add_argument("--cpu", action="store_true", default=False)
-    parser.add_argument("--torchvision_model", type=str, help="Any torchvision model name (str)", default="resnet101")
+    parser.add_argument("--model", type=str, help="Any torchvision or timm model name (str)", default="resnet101")
     parser.add_argument("--debug", action="store_true", default=False, help="Display additional debug information")
     parser.add_argument("--amp", action="store_true", default=False, help="Activate torch AMP")
 
@@ -293,8 +306,8 @@ if __name__ == "__main__":
     if args.optim_type == OptimType.vanilla or args.optim_type == OptimType.everyone:
         logging.info("\n*** Benchmark vanilla optimizer")
         mp.spawn(
-            train,
-            args=(args, BACKEND, OptimType.vanilla, False,),  # no regression check
+            train,  # type: ignore
+            args=(args, BACKEND, OptimType.vanilla, False),  # no regression check
             nprocs=args.world_size,
             join=True,
         )
@@ -302,13 +315,13 @@ if __name__ == "__main__":
     if args.optim_type == OptimType.oss_ddp or args.optim_type == OptimType.everyone:
         logging.info("\n*** Benchmark OSS with DDP")
         mp.spawn(
-            train, args=(args, BACKEND, OptimType.oss_ddp, args.check_regression), nprocs=args.world_size, join=True,
+            train, args=(args, BACKEND, OptimType.oss_ddp, args.check_regression), nprocs=args.world_size, join=True,  # type: ignore
         )
 
     if args.optim_type == OptimType.oss_sharded_ddp or args.optim_type == OptimType.everyone:
         logging.info("\n*** Benchmark OSS with ShardedDDP")
         mp.spawn(
-            train,
+            train,  # type: ignore
             args=(
                 args,
                 BACKEND,

requirements-test.txt

@@ -12,3 +12,4 @@ torch >= 1.5.1
 torchvision >= 0.6.0
 # NOTE(msb) not a dependency but needed by torch
 numpy == 1.17.4
+timm == 0.3.4