scripts/single_node/eval_4gpu_ckp_300.sh

#!/usr/bin/env bash

# set -euo pipefail

CURRENT_SERVER_IP="169.254.3.1"

if ip -o -4 addr show 2>/dev/null | awk '{split($4, address, "/"); print address[1]}' | grep -qx "$CURRENT_SERVER_IP"; then
  PROJECT_ROOT=$(CDPATH= cd -- "$(dirname -- "$0")/../.." && pwd)
  export PYTHONPATH=$PROJECT_ROOT:${PYTHONPATH:-}
  export OMNIGEN_CODE_ROOT=/home/wenting/zr/gen_code
  export PYTORCH_CUDA_ALLOC_CONF=${PYTORCH_CUDA_ALLOC_CONF:-expandable_segments:True}
else
  export PYTHONPATH=/NAS_REMOTE/vicky/wt/codes/cxr_syn:/NAS_REMOTE/vicky/wt/codes/cxr_syn/flow_grpo:${PYTHONPATH:-}
  export OMNIGEN_CODE_ROOT=/NAS_REMOTE/vicky/wt/codes/cxr_syn/gen_code
fi

export PYTHONPATH=$OMNIGEN_CODE_ROOT:${PYTHONPATH:-}
export HF_HOME="/NAS_REMOTE/vicky/wt/huggingface/models"
export HF_HUB_CACHE="/tmp/flow_grpo_hf_cache/hub"

export CUDA_VISIBLE_DEVICES=${CUDA_VISIBLE_DEVICES:-0,1,2,3}

CONFIG_ENTRY=${CONFIG_ENTRY:-"config/grpo.py:general_radiomics_omnigen_4gpu_kl_eval"}
EVAL_LORA_PATH=${EVAL_LORA_PATH:-"logs/radiomics/omnigen-lora-r32-a64-4gpu-bs32-evalbs24-kl-beta0p005-from-cxr-lora30000/checkpoints/checkpoint-300/lora"}

python3 -m accelerate.commands.launch \
  --config_file scripts/accelerate_configs/multi_gpu.yaml \
  --num_processes=${NUM_PROCESSES:-4} \
  --main_process_port ${MAIN_PROCESS_PORT:-29501} \
  scripts/eval_omnigen.py \
  --config ${CONFIG_ENTRY} \
  --eval_lora_path ${EVAL_LORA_PATH} \
  --resume_dir outputs/eval_eval_2026.05.06_01.24.15 \
  "$@"