DDPM-2param / scripts /shell /run_ddim_investigation_2param.sh
collins909's picture
Upload 2-parameter conditional DDPM (HI emulation, CAMELS LH params_2, epoch 200) with full training/eval/posterior toolchain
f513198 verified
#!/bin/bash
#SBATCH --account=<your-slurm-account>
#SBATCH --partition=l40s
#SBATCH --nodes=1
#SBATCH --ntasks=4
#SBATCH --gres=gpu:l40s:1
#SBATCH --time=08:00:00
#SBATCH --job-name=ddim2param
#SBATCH --mail-user=<your-email> # replace before submitting
#SBATCH --output=slurm-ddim-invest-%j.out
#SBATCH --error=slurm-ddim-invest-%j.err
set -euo pipefail
ROOT="<DDPM_ROOT>/Models/2param_DDPM_HI_Emulation"
cd "$ROOT"
module load python/miniconda3-py3.12-usr
CHECKPOINT="${CHECKPOINT:-<DDPM_ROOT>/Models/notebook_model_weights/2param_epoch200/checkpoint_epoch_200.pt}"
OUTPUT_DIR="${OUTPUT_DIR:-<DDPM_ROOT>/Models/ddim_investigation_2param_out}"
STEPS="${STEPS:-1500 1000 500 250 100 50}"
NUM_SAMPLES="${NUM_SAMPLES:-100}"
IMAGE_SIZE="${IMAGE_SIZE:-64}"
DEVICE="${DEVICE:-cuda}"
RUN_ID="${RUN_ID:-post-fix}"
CAMELS_DATA="${CAMELS_DATA:-<DDPM_ROOT>/data/LH_data/params_2/train_LH.npy}"
PK_COMPARE_SAMPLES="${PK_COMPARE_SAMPLES:-100}"
if [[ ! -f "${CHECKPOINT}" ]]; then
echo "Checkpoint file not found: ${CHECKPOINT}" >&2
exit 1
fi
echo "==============================================="
echo "Job ID: ${SLURM_JOB_ID:-local}"
echo "Node: ${SLURM_NODELIST:-$(hostname)}"
echo "GPU: ${CUDA_VISIBLE_DEVICES:-default}"
echo "Started: $(date)"
echo "Checkpoint: ${CHECKPOINT}"
echo "Output dir: ${OUTPUT_DIR}"
echo "Steps: ${STEPS}"
echo "Num samples: ${NUM_SAMPLES}"
echo "CAMELS data: ${CAMELS_DATA}"
echo "PK compare samples: ${PK_COMPARE_SAMPLES}"
echo "==============================================="
python ddim_investigation_2param.py \
--checkpoint "${CHECKPOINT}" \
--output_dir "${OUTPUT_DIR}" \
--steps ${STEPS} \
--num_samples "${NUM_SAMPLES}" \
--image_size "${IMAGE_SIZE}" \
--device "${DEVICE}" \
--camels_data "${CAMELS_DATA}" \
--pk_compare_samples "${PK_COMPARE_SAMPLES}" \
--run_id "${RUN_ID}" \
"$@"
echo "==============================================="
echo "Finished: $(date)"
echo "Artifacts: ${OUTPUT_DIR}"
echo "==============================================="