#!/bin/bash # Run training with distributed setup set -e # Install accelerate if not present pip install -q accelerate # Launch with accelerate accelerate launch train-on-this-server.sh --config training/configs/llama2-7b-lora.yaml