#!/bin/bash

# Manually specify values used in the config
main_task="preprocess"
data_task_type="embeddings"
timestamp=$(date "+%Y-%m-%d_%H-%M-%S")

run_dir="$HOME/DPACMAN/logs/${main_task}/${data_task_type}/runs/${timestamp}"
mkdir -p "$run_dir"

nohup python -s -u -m scripts.preprocess \
  hydra.run.dir="${run_dir}" \
  data_task="${data_task_type}/dna" \
  data_task.chrom_model="caduceus" \
  data_task.debug="true" \
  > "${run_dir}/run.log" 2>&1 &

echo $! > "${run_dir}/pid.txt"