LLaMA-Factory-Mirror/examples/lora_multi_gpu/multi_node.sh

7 lines
241 B
Bash

#!/bin/bash
# also launch it on slave machine using slave_config.yaml
CUDA_VISIBLE_DEVICES=0,1,2,3 accelerate launch \
--config_file examples/accelerate/master_config.yaml \
src/train.py examples/lora_multi_gpu/llama3_lora_sft.yaml