trainer.slurm

#!/bin/bash

#SBATCH --job-name=transformer-trainer

#SBATCH --ntasks=2

#SBATCH --nodes=2

#SBATCH --gpus-per-task=8

#SBATCH --cpus-per-task=96


nodes=( $( scontrol show hostnames $SLURM_JOB_NODELIST ) )
nodes_array=($nodes)
head_node=${nodes_array[0]}
head_node_ip=$(srun --nodes=1 --ntasks=1 -w "$head_node" hostname --ip-address)

echo Node IP: $head_node_ip
export LOGLEVEL=INFO
# Enable for A100
export FI_PROVIDER="efa"
# Enable this on A100 gpus to avoid NCCL error
export NCCL_P2P_DISABLE=1
export NCCL_IB_DISABLE=1

# debugging flags (optional)
export NCCL_DEBUG=WARN
export NCCL_DEBUG_SUBSYS=WARN
export PYTHONFAULTHANDLER=1

export LD_LIBRARY_PATH=/opt/amazon/efa/lib:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/local/lib/:$LD_LIBRARY_PATH
export CUDA_LAUNCH_BLOCKING=0

# on your cluster you might need these:
# set the network interface
export NCCL_SOCKET_IFNAME="eth0,en,eth,em,bond"
dcgmi profile --pause
srun torchrun --nnodes 2 --nproc_per_node 8 --rdzv_id 101 --rdzv_backend c10d --rdzv_endpoint "$head_node_ip:29500" ./main_training.py
dcgmi profile --resume