multi machines training

Hi!
When I run the following code, the program is stuck after printing the cfg both on node0 and node 1.
Can you help me and provide a multi-machine multi-GPU training script for reference?
#### Node 0
```
python3 /opt/tiger/SlowFast/tools/run_net.py
--cfg configs/Kinetics/LSS_SLOW_8x8_R50.yaml \
--shard_id 0 \
--num_shards 2 \
--init_method tcp://10.136.71.98:34567 \
DATA.PATH_TO_DATA_DIR /data/kinetics400large \
DATA.PATH_PREFIX /data/kinetics400large \
NUM_GPUS 2 \
TRAIN.BATCH_SIZE 16
```
#### Node 1
```
python3 /opt/tiger/SlowFast/tools/run_net.py
--cfg configs/Kinetics/LSS_SLOW_8x8_R50.yaml \
--shard_id 1 \
--num_shards 2 \
--init_method tcp://10.136.71.98:34567 \
DATA.PATH_TO_DATA_DIR /data/kinetics400large \
DATA.PATH_PREFIX /data/kinetics400large \
NUM_GPUS 2 \
TRAIN.BATCH_SIZE 16
```