Add script and run.sh

README.rst

@@ -20,6 +20,15 @@ Setting up decentralizepy
 
     pip3 install --editable .\[dev\]
     
+----------------
+Running the code
+----------------
+
+* Choose and modify one of the config files in ``eval/{step,epoch}_configs``.
+* Modify the dataset paths and ``addresses_filepath`` in the config file.
+* In eval/run.sh, modify ``first_machine`` (used to calculate machine_id of all machines), ``original_config``, and other arguments as required.
+* Execute eval/run.sh on all the machines simultaneously. There is a synchronization barrier mechanism at the start so that all processes start training together.
+
 Node
 ----
 * The Manager. Optimizations at process level.

eval/epoch_configs/config_celeba.ini

@@ -17,9 +17,9 @@ lr = 0.001
 [TRAIN_PARAMS]
 training_package = decentralizepy.training.GradientAccumulator
 training_class = GradientAccumulator
-rounds = 5
+rounds = 1
 full_epochs = True
-batch_size = 512
+batch_size = 64
 shuffle = True
 loss_package = torch.nn
 loss_class = CrossEntropyLoss

eval/epoch_configs/config_celeba_100.ini

@@ -17,9 +17,9 @@ lr = 0.001
 [TRAIN_PARAMS]
 training_package = decentralizepy.training.Training
 training_class = Training
-rounds = 5
+rounds = 1
 full_epochs = True
-batch_size = 512
+batch_size = 64
 shuffle = True
 loss_package = torch.nn
 loss_class = CrossEntropyLoss

eval/epoch_configs/config_celeba_grow.ini

@@ -17,9 +17,9 @@ lr = 0.001
 [TRAIN_PARAMS]
 training_package = decentralizepy.training.GradientAccumulator
 training_class = GradientAccumulator
-rounds = 5
+rounds = 1
 full_epochs = True
-batch_size = 512
+batch_size = 64
 shuffle = True
 loss_package = torch.nn
 loss_class = CrossEntropyLoss

eval/epoch_configs/config_femnist.ini

@@ -16,9 +16,9 @@ lr = 0.001
 [TRAIN_PARAMS]
 training_package = decentralizepy.training.GradientAccumulator
 training_class = GradientAccumulator
-rounds = 5
+rounds = 1
 full_epochs = True
-batch_size = 1024
+batch_size = 128
 shuffle = True
 loss_package = torch.nn
 loss_class = CrossEntropyLoss

eval/epoch_configs/config_femnist_100.ini

@@ -16,9 +16,9 @@ lr = 0.001
 [TRAIN_PARAMS]
 training_package = decentralizepy.training.Training
 training_class = Training
-rounds = 5
+rounds = 1
 full_epochs = True
-batch_size = 1024
+batch_size = 128
 shuffle = True
 loss_package = torch.nn
 loss_class = CrossEntropyLoss

eval/epoch_configs/config_femnist_grow.ini

@@ -16,9 +16,9 @@ lr = 0.001
 [TRAIN_PARAMS]
 training_package = decentralizepy.training.GradientAccumulator
 training_class = GradientAccumulator
-rounds = 5
+rounds = 1
 full_epochs = True
-batch_size = 1024
+batch_size = 128
 shuffle = True
 loss_package = torch.nn
 loss_class = CrossEntropyLoss