[Distributed] Fix all arguments to the format of xx_xxx (#2005)

* update

* update

examples/pytorch/graphsage/experimental/README.md

@@ -50,7 +50,7 @@ python3 ~/dgl/tools/launch.py \
 --num_servers 1 \
 --part_config ogb-product/ogb-product.json \
 --ip_config ip_config.txt \
-"python3 train_dist.py --graph-name ogb-product --ip_config ip_config.txt --num-servers 1 --num-epochs 30 --batch-size 1000 --num-workers 4"
+"python3 train_dist.py --graph_name ogb-product --ip_config ip_config.txt --num_servers 1 --num_epochs 30 --batch_size 1000 --num_workers 4"
 ```
 
 To run unsupervised training:
@@ -62,7 +62,7 @@ python3 ~/dgl/tools/launch.py \
 --num_servers 1 \
 --part_config ogb-product/ogb-product.json \
 --ip_config ip_config.txt \
-"python3 ~/dgl/examples/pytorch/graphsage/experimental/train_dist_unsupervised.py --graph-name ogb-product --ip_config ip_config.txt --num-servers 1 --num-epochs 3 --batch-size 1000"
+"python3 ~/dgl/examples/pytorch/graphsage/experimental/train_dist_unsupervised.py --graph_name ogb-product --ip_config ip_config.txt --num_servers 1 --num_epochs 3 --batch_size 1000"
 ```
 
 ## Distributed code runs in the standalone mode
@@ -81,13 +81,13 @@ python3 partition_graph.py --dataset ogb-product --num_parts 1
 To run supervised training:
 
 ```bash
-python3 train_dist.py --graph-name ogb-product --ip_config ip_config.txt --num-epochs 3 --batch-size 1000 --part_config data/ogb-product.json --standalone
+python3 train_dist.py --graph_name ogb-product --ip_config ip_config.txt --num_epochs 3 --batch_size 1000 --part_config data/ogb-product.json --standalone
 ```
 
 To run unsupervised training:
 
 ```bash
-python3 train_dist_unsupervised.py --graph-name ogb-product --ip_config ip_config.txt --num-epochs 3 --batch-size 1000 --part_config data/ogb-product.json --standalone
+python3 train_dist_unsupervised.py --graph_name ogb-product --ip_config ip_config.txt --num_epochs 3 --batch_size 1000 --part_config data/ogb-product.json --standalone
 ```
 
 Note: please ensure that all environment variables shown above are unset if they were set for testing distributed training.

examples/pytorch/graphsage/experimental/train_dist.py

@@ -289,26 +289,26 @@ def main(args):
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='GCN')
     register_data_args(parser)
-    parser.add_argument('--graph-name', type=str, help='graph name')
+    parser.add_argument('--graph_name', type=str, help='graph name')
     parser.add_argument('--id', type=int, help='the partition id')
     parser.add_argument('--ip_config', type=str, help='The file for IP configuration')
     parser.add_argument('--part_config', type=str, help='The path to the partition config file')
-    parser.add_argument('--num-client', type=int, help='The number of clients')
-    parser.add_argument('--num-servers', type=int, default=1, help='The number of servers')
-    parser.add_argument('--n-classes', type=int, help='the number of classes')
+    parser.add_argument('--num_clients', type=int, help='The number of clients')
+    parser.add_argument('--num_servers', type=int, default=1, help='The number of servers')
+    parser.add_argument('--n_classes', type=int, help='the number of classes')
     parser.add_argument('--gpu', type=int, default=0,
         help="GPU device ID. Use -1 for CPU training")
-    parser.add_argument('--num-epochs', type=int, default=20)
-    parser.add_argument('--num-hidden', type=int, default=16)
-    parser.add_argument('--num-layers', type=int, default=2)
-    parser.add_argument('--fan-out', type=str, default='10,25')
-    parser.add_argument('--batch-size', type=int, default=1000)
-    parser.add_argument('--batch-size-eval', type=int, default=100000)
-    parser.add_argument('--log-every', type=int, default=20)
-    parser.add_argument('--eval-every', type=int, default=5)
+    parser.add_argument('--num_epochs', type=int, default=20)
+    parser.add_argument('--num_hidden', type=int, default=16)
+    parser.add_argument('--num_layers', type=int, default=2)
+    parser.add_argument('--fan_out', type=str, default='10,25')
+    parser.add_argument('--batch_size', type=int, default=1000)
+    parser.add_argument('--batch_size_eval', type=int, default=100000)
+    parser.add_argument('--log_every', type=int, default=20)
+    parser.add_argument('--eval_every', type=int, default=5)
     parser.add_argument('--lr', type=float, default=0.003)
     parser.add_argument('--dropout', type=float, default=0.5)
-    parser.add_argument('--num-workers', type=int, default=4,
+    parser.add_argument('--num_workers', type=int, default=4,
         help="Number of sampling processes. Use 0 for no extra process.")
     parser.add_argument('--local_rank', type=int, help='get rank of the process')
     parser.add_argument('--standalone', action='store_true', help='run in the standalone mode')

examples/pytorch/graphsage/experimental/train_dist_unsupervised.py

@@ -448,32 +448,32 @@ def main(args):
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='GCN')
     register_data_args(parser)
-    parser.add_argument('--graph-name', type=str, help='graph name')
+    parser.add_argument('--graph_name', type=str, help='graph name')
     parser.add_argument('--id', type=int, help='the partition id')
     parser.add_argument('--ip_config', type=str, help='The file for IP configuration')
     parser.add_argument('--part_config', type=str, help='The path to the partition config file')
-    parser.add_argument('--num-servers', type=int, default=1, help='Server count on each machine.')
-    parser.add_argument('--n-classes', type=int, help='the number of classes')
+    parser.add_argument('--num_servers', type=int, default=1, help='Server count on each machine.')
+    parser.add_argument('--n_classes', type=int, help='the number of classes')
     parser.add_argument('--gpu', type=int, default=0,
         help="GPU device ID. Use -1 for CPU training")
-    parser.add_argument('--num-epochs', type=int, default=20)
-    parser.add_argument('--num-hidden', type=int, default=16)
+    parser.add_argument('--num_epochs', type=int, default=20)
+    parser.add_argument('--num_hidden', type=int, default=16)
     parser.add_argument('--num-layers', type=int, default=2)
-    parser.add_argument('--fan-out', type=str, default='10,25')
-    parser.add_argument('--batch-size', type=int, default=1000)
-    parser.add_argument('--batch-size-eval', type=int, default=100000)
-    parser.add_argument('--log-every', type=int, default=20)
-    parser.add_argument('--eval-every', type=int, default=5)
+    parser.add_argument('--fan_out', type=str, default='10,25')
+    parser.add_argument('--batch_size', type=int, default=1000)
+    parser.add_argument('--batch_size_eval', type=int, default=100000)
+    parser.add_argument('--log_every', type=int, default=20)
+    parser.add_argument('--eval_every', type=int, default=5)
     parser.add_argument('--lr', type=float, default=0.003)
     parser.add_argument('--dropout', type=float, default=0.5)
-    parser.add_argument('--num-workers', type=int, default=0,
+    parser.add_argument('--num_workers', type=int, default=0,
         help="Number of sampling processes. Use 0 for no extra process.")
     parser.add_argument('--local_rank', type=int, help='get rank of the process')
     parser.add_argument('--standalone', action='store_true', help='run in the standalone mode')
-    parser.add_argument('--num-negs', type=int, default=1)
-    parser.add_argument('--neg-share', default=False, action='store_true',
+    parser.add_argument('--num_negs', type=int, default=1)
+    parser.add_argument('--neg_share', default=False, action='store_true',
         help="sharing neg nodes for positive nodes")
-    parser.add_argument('--remove-edge', default=False, action='store_true',
+    parser.add_argument('--remove_edge', default=False, action='store_true',
         help="whether to remove edges during sampling")
     args = parser.parse_args()
 

tools/launch.py

@@ -113,12 +113,12 @@ def main():
                         help='The number of trainer processes per machine')
     parser.add_argument('--num_samplers', type=int, default=0,
                         help='The number of sampler processes per trainer process')
+    parser.add_argument('--num_servers', type=int,
+                        help='The number of server processes per machine')
     parser.add_argument('--part_config', type=str,
                         help='The file (in workspace) of the partition config')
     parser.add_argument('--ip_config', type=str,
                         help='The file (in workspace) of IP configuration for server processes')
-    parser.add_argument('--num_servers', type=int,
-                        help='Server count on each machine.')
     parser.add_argument('--num_server_threads', type=int, default=1,
                         help='The number of OMP threads in the server process. \
                         It should be small if server processes and trainer processes run on \