#!/bin/bash

#SBATCH --job-name=m_16_2k          # 作业名称
#SBATCH --account=PAS2490		    # Project ID
#SBATCH --nodes=1                   # 节点数
#SBATCH --ntasks-per-node=1         # 每个节点的任务数
#SBATCH --cpus-per-task=16           # 每个任务使用的 CPU 核心数
#SBATCH --gpus-per-node=1           # GPU per node
#SBATCH --mem=200G                   # 内存限制
#SBATCH --time=80:00:00             # 作业运行时间限制

#SBATCH --mail-type=BEGIN,END,FAIL
#SBATCH --mail-user=uceckz0@ucl.ac.uk





convert_to_seconds() {
    date -d "${1//_/ }" +%s
}
start_time=$(date +%Y-%m-%d_%H:%M:%S)
echo "Train start time: $start_time"


source $HOME/miniconda3/etc/profile.d/conda.sh
conda activate mace_env

m -rf ./checkpoints
rm -rf ./logs
rm -rf ./results
rm -f valid_indices_123.txt

mace_run_train \
    --name="MACE_model" \
    --train_file="../trainset.xyz" \
    --num_workers=16 \
    --valid_fraction=0.25 \
    --test_file="../testset.xyz" \
    --config_type_weights='{"Default":1.0}' \
    --E0s='{1:-13.663181292231226, 6:-1029.2809654211628, 7:-1484.1187695035828, 8:-2042.0330099956639, 30:-500.123456789}' \
    --lr=0.0001 \
    --model="MACE" \
    --hidden_irreps='128x0e + 128x1o' \
    --r_max=5.0 \
    --batch_size=10 \
    --max_num_epochs=500 \
    --swa \
    --ema \
    --ema_decay=0.99 \
    --amsgrad \
    --restart_latest \
    --device=cuda \





end_time=$(date +%Y-%m-%d_%H:%M:%S)
echo "Train end time: $end_time"

start_seconds=$(convert_to_seconds "$start_time")
end_seconds=$(convert_to_seconds "$end_time")
time_difference=$((end_seconds - start_seconds))

echo "Training duration: $time_difference seconds"

duration_h=$((time_difference / 3600))
duration_m=$(((time_difference % 3600) / 60))
duration_s=$((time_difference % 60))

echo "Training duration: ${duration_h}h ${duration_m}m ${duration_s}s"