Add ddp config and script.

examples/torch/MNIST/config_ddp.yaml

@@ -0,0 +1,72 @@
+trainer:
+  accelerator: "gpu"
+  devices: 8
+  strategy: "ddp"
+  max_epochs: 10
+  callbacks:
+    - _target_: lightning.pytorch.callbacks.ModelCheckpoint
+      filename: "{epoch:04d}"
+      save_last: True
+      save_top_k: -1
+
+task:
+  _target_: torchvision_task.Resnet50Task
+  model:
+    _target_: torchvision.models.resnet50
+    weights:
+      _target_: hydra.utils.get_object
+      path: torchvision.models.ResNet50_Weights.DEFAULT
+  optimizer_config:
+    _target_: aiaccel.torch.lightning.OptimizerConfig
+    optimizer_generator:
+      _partial_: True
+      _target_: torch.optim.AdamW
+      lr: 1.e-4
+  num_classes: 10
+
+datamodule:
+  _target_: aiaccel.torch.lightning.datamodules.single_datamodule.SingleDataModule
+  train_dataset_fn:
+    _partial_: true
+    _target_: torchvision.datasets.MNIST
+    root: "./dataset"
+    train: True
+    download: True
+    transform:
+      _target_: torchvision.transforms.Compose
+      transforms:
+        - _target_: torchvision.transforms.Resize
+          size:
+            - 256
+            - 256
+        - _target_: torchvision.transforms.Grayscale
+          num_output_channels: 3
+        - _target_: torchvision.transforms.ToTensor
+        - _target_: torchvision.transforms.Normalize
+          mean:
+            - 0.5
+          std:
+            - 0.5
+  val_dataset_fn:
+    _partial_: true
+    _target_: torchvision.datasets.MNIST
+    root: "./dataset"
+    train: False
+    download: True
+    transform:
+      _target_: torchvision.transforms.Compose
+      transforms:
+        - _target_: torchvision.transforms.Resize
+          size:
+            - 256
+            - 256
+        - _target_: torchvision.transforms.Grayscale
+          num_output_channels: 3
+        - _target_: torchvision.transforms.ToTensor
+        - _target_: torchvision.transforms.Normalize
+          mean:
+            - 0.5
+          std:
+            - 0.5
+  batch_size: 128
+  wrap_scatter_dataset: False

examples/torch/MNIST/train_ddp.sh

@@ -0,0 +1,18 @@
+#! /bin/bash
+
+#PBS -q rt_HF
+#PBS -l select=1
+#PBS -l walltime=1:00:00
+#PBS -P grpname
+#PBS -j oe
+
+cd ${PBS_O_WORKDIR}
+
+source /etc/profile.d/modules.sh
+module load cuda/12.6/12.6.1
+
+source path_to_aiaccel_env/bin/activate
+
+wd=path_to_working_directory
+
+singularity exec --nv path_to_python.sif python -m aiaccel.torch.apps.train $wd/config_ddp.yaml --working_directory $wd