scripts/train.sh

#!/bin/bash

export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7

torchrun \
    --nproc_per_node=8 \
    --master_port=29898 \
    train.py \
    --num_refine_groups 3 \
    --batch_size 6 \
    --lr 1e-4 \
    --epochs 100 \
    --save_every_k_epochs 1 \
    --last_ckpt /home/dataset-assist-0/usr/lh/ysh/dw/RL/AR/checkpoints/20260617_190742/model_step_4500.pth \
    --auto_resume