Added ResNet20

composer/models/__init__.py

@@ -19,6 +19,8 @@
 from composer.models.resnet9_cifar10 import CIFARResNet9Hparams as CIFARResNet9Hparams
 from composer.models.resnet18 import ResNet18 as ResNet18
 from composer.models.resnet18 import ResNet18Hparams as ResNet18Hparams
+from composer.models.resnet20_cifar10 import CIFAR10_ResNet20 as CIFAR10_ResNet20
+from composer.models.resnet20_cifar10 import CIFARResNet20Hparams as CIFARResNet20Hparams
 from composer.models.resnet50 import ResNet50 as ResNet50
 from composer.models.resnet50 import ResNet50Hparams as ResNet50Hparams
 from composer.models.resnet56_cifar10 import CIFAR10_ResNet56 as CIFAR10_ResNet56

composer/models/resnet20_cifar10/__init__.py

@@ -0,0 +1,12 @@
+# Copyright 2021 MosaicML. All Rights Reserved.
+
+from composer.models.resnet20_cifar10.model import CIFAR10_ResNet20 as CIFAR10_ResNet20
+from composer.models.resnet20_cifar10.resnet20_cifar10_hparams import CIFARResNet20Hparams as CIFARResNet20Hparams
+
+_task = 'Image Classification'
+_dataset = 'CIFAR10'
+_name = 'ResNet20'
+_quality = 'tbd'
+_metric = 'Top-1 Accuracy'
+_ttt = 'tbd'
+_hparams = 'resnet20_cifar10.yaml'

composer/models/resnet20_cifar10/model.py

@@ -0,0 +1,36 @@
+# Copyright 2021 MosaicML. All Rights Reserved.
+
+from typing import List, Optional
+
+from composer.models.base import MosaicClassifier
+from composer.models.model_hparams import Initializer
+from composer.models.resnets import CIFAR_ResNet
+
+
+class CIFAR10_ResNet20(MosaicClassifier):
+    """A ResNet-20 model extending :class:`MosaicClassifier`.
+
+    See this `paper <https://arxiv.org/abs/1512.03385>`_ for details
+    on the residual network architecture.
+
+    Args:
+        num_classes (int): The number of classes for the model. Default = 10.
+        initializers (List[Initializer], optional): Initializers
+            for the model. ``None`` for no initialization.
+            (default: ``None``)
+    """
+
+    def __init__(
+        self,
+        num_classes: int = 10,
+        initializers: Optional[List[Initializer]] = None,
+    ) -> None:
+        if initializers is None:
+            initializers = []
+
+        model = CIFAR_ResNet.get_model_from_name(
+            "cifar_resnet_20",
+            initializers,
+            num_classes,
+        )
+        super().__init__(module=model)

composer/models/resnet20_cifar10/resnet20_cifar10_hparams.py

@@ -0,0 +1,13 @@
+# Copyright 2021 MosaicML. All Rights Reserved.
+
+from dataclasses import asdict, dataclass
+
+from composer.models.model_hparams import ModelHparams
+
+
+@dataclass
+class CIFARResNet20Hparams(ModelHparams):
+
+    def initialize_object(self):
+        from composer.models import CIFAR10_ResNet20
+        return CIFAR10_ResNet20(**asdict(self))

composer/models/resnet56_cifar10/__init__.py

@@ -8,5 +8,5 @@
 _name = 'ResNet56'
 _quality = '93.1'
 _metric = 'Top-1 Accuracy'
-_ttt = '15m'
+_ttt = 'tbd'
 _hparams = 'resnet56_cifar10.yaml'

composer/models/resnet56_cifar10/model.py

@@ -14,7 +14,7 @@ class CIFAR10_ResNet56(MosaicClassifier):
     on the residual network architecture.
 
     Args:
-        num_classes (int): The number of classes for the model.
+        num_classes (int): The number of classes for the model. Default = 10.
         initializers (List[Initializer], optional): Initializers
             for the model. ``None`` for no initialization.
             (default: ``None``)

composer/models/resnets.py

@@ -303,7 +303,7 @@ def forward(self, x):
     @staticmethod
     def is_valid_model_name(model_name):
         return (model_name.startswith('cifar_resnet_') and 4 >= len(model_name.split('_')) >= 3 and
-                model_name.split('_')[2].isdigit() and int(model_name.split('_')[2]) in [56])
+                model_name.split('_')[2].isdigit() and int(model_name.split('_')[2]) in [20,56])
 
     @staticmethod
     def get_model_from_name(model_name, initializers: List[Initializer], outputs=10):
@@ -327,6 +327,7 @@ def get_model_from_name(model_name, initializers: List[Initializer], outputs=10)
 
         model_arch = {
             56: [(width, num_blocks), (2 * width, num_blocks), (4 * width, num_blocks)],
+            20: [(width, num_blocks), (2 * width, num_blocks), (4 * width, num_blocks)],
         }
 
         return CIFAR_ResNet(model_arch[depth], initializers, outputs)

composer/trainer/trainer_hparams.py

@@ -24,6 +24,7 @@
 from composer.models import (BERTForClassificationHparams, BERTHparams, CIFARResNet9Hparams, CIFARResNetHparams,
                              DeepLabV3Hparams, EfficientNetB0Hparams, GPT2Hparams, MnistClassifierHparams, ModelHparams,
                              ResNet18Hparams, ResNet50Hparams, ResNet101Hparams, UnetHparams)
+from composer.models.resnet20_cifar10.resnet20_cifar10_hparams import CIFARResNet20Hparams
 from composer.optim import (AdamHparams, AdamWHparams, DecoupledAdamWHparams, DecoupledSGDWHparams, OptimizerHparams,
                             RAdamHparams, RMSPropHparams, SchedulerHparams, SGDHparams, scheduler)
 from composer.profiler import ProfilerHparams
@@ -63,6 +64,7 @@
     "deeplabv3": DeepLabV3Hparams,
     "efficientnetb0": EfficientNetB0Hparams,
     "resnet56_cifar10": CIFARResNetHparams,
+    "resnet20_cifar10": CIFARResNet20Hparams,
     "resnet9_cifar10": CIFARResNet9Hparams,
     "resnet101": ResNet101Hparams,
     "resnet50": ResNet50Hparams,

composer/yamls/models/resnet20_cifar10.yaml

@@ -0,0 +1,54 @@
+train_dataset:
+  cifar10:
+    datadir: /datasets/CIFAR10
+    is_train: true
+    download: false
+    shuffle: true
+    drop_last: true
+val_dataset:
+  cifar10:
+    datadir: /datasets/CIFAR10
+    is_train: false
+    download: false
+    shuffle: false
+    drop_last: false
+optimizer:
+  decoupled_sgdw:
+    lr: 1.2
+    momentum: 0.9
+    weight_decay: 2.0e-3
+schedulers:
+  - warmup:
+      warmup_iters: "5ep"
+      warmup_method: linear
+      warmup_factor: 0
+      verbose: false
+      interval: step
+  - multistep:
+      milestones:
+        - "80ep"
+        - "120ep"
+      gamma: 0.1
+      interval: epoch
+model:
+  resnet20_cifar10:
+    initializers:
+      - kaiming_normal
+      - bn_uniform
+loggers:
+  - tqdm: {}
+max_duration: 160ep
+train_batch_size: 1024
+eval_batch_size: 1000
+seed: 17
+validate_every_n_epochs: 1
+grad_accum: 1
+device:
+  gpu: {}
+dataloader:
+  pin_memory: true
+  timeout: 0
+  prefetch_factor: 2
+  persistent_workers: true
+  num_workers: 8
+precision: amp

composer/yamls/models/resnet56_cifar10.yaml

@@ -35,7 +35,6 @@ model:
     initializers:
       - kaiming_normal
       - bn_uniform
-    num_classes: 10
 loggers:
   - tqdm: {}
 max_duration: 160ep

composer/yamls/models/resnet56_cifar10_synthetic.yaml

@@ -35,7 +35,6 @@ model:
     initializers:
       - kaiming_normal
       - bn_uniform
-    num_classes: 10
 loggers:
   - tqdm: {}
 max_duration: 160ep