restructured job submission scripts

root_gnn_dgl/jobs/cpu.sh

@@ -1 +1 @@
-salloc --nodes=1 --ntasks=64 --cpus-per-task=1 --qos=interactive --time=04:00:00 --constraint=cpu --account=atlas
+salloc --nodes=1 --ntasks=64 --cpus-per-task=1 --qos=interactive --time=04:00:00 --constraint=cpu --account=trn007

root_gnn_dgl/jobs/interactive.sh

@@ -1 +1 @@
-salloc --nodes 1 --qos shared_interactive --time 04:00:00 --constraint gpu --account=atlas --gres=gpu:1 --image=joshuaho/dgl:1.0
+salloc --nodes 1 --qos shared_interactive --time 04:00:00 --constraint gpu --account=trn007 --gres=gpu:1

root_gnn_dgl/jobs/prep_data/prep_data.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+
+# Arguments
+config=$1   # YAML configuration file
+dataset=$2  # Dataset name
+CHUNKS=$3  # Number of chunks
+
+args=$4     # Additional arguments to pass to prep_data.py
+
+# Prepare the initial command for chunk 0
+command="scripts/prep_data.py --dataset $dataset --shuffle_mode --chunk 0 --config $config $args"
+echo "$command"
+python $command
+
+# Loop through all chunks
+for ((i=0; i<$CHUNKS; i++)); do
+  command="scripts/prep_data.py --dataset $dataset --shuffle_mode --chunk $i --config $config $args"
+  echo "$command"
+  python $command
+done

root_gnn_dgl/jobs/{process_data.py → prep_data/run_processing.py}

@@ -71,26 +71,18 @@ def process_yaml_file(config, base_directory, bash_script):
 
 def main():
     # Base directory containing the YAML files
-    base_directory = "/global/cfs/projectdirs/atlas/joshua/root_gnn/root_gnn_dgl/configs/model_configs/"
+    base_directory = os.getcwd() + "/"
 
     # List of YAML configuration files (relative to the base directory)
     configs = [
-        # "run_3_ttH/v05/sb_yukawa_cp_no_weights.yaml",
-        # "run_3_ttH/v05/sb_yukawa_cp_weights.yaml",
-        # "run_3_ttH/v06/sb_yukawa_no_weights.yaml",
-        # "run_3_ttH/v06/sb_yukawa_weights.yaml",
-        # "run_3_ttH/v06/tHjb_CP_no_weights.yaml",
-        # "run_3_ttH/v06/tHjb_CP_weights.yaml",
-        "run_3_ttH/v06/ttH_CP_no_weights.yaml",
-        "run_3_ttH/v06/ttH_CP_weights.yaml",
-        "run_3_ttH/v06/tWH_CP_no_weights.yaml",
-        "run_3_ttH/v06/tWH_CP_weights.yaml",
-        "run_3_ttH/v06/ttH_tHjb_tWH_CP_no_weights.yaml",
-        "run_3_ttH/v06/ttH_tHjb_tWH_CP_weights.yaml",
+        "configs/stats_100K/pretraining_multiclass.yaml",
+        "configs/stats_100K/ttH_CP_even_vs_odd.yaml",
+        "configs/stats_all/pretraining_multiclass.yaml",
+        "configs/stats_all/ttH_CP_even_vs_odd.yaml",
     ]
 
     # Path to the bash script to be called
-    bash_script = "/global/cfs/projectdirs/atlas/joshua/root_gnn/root_gnn_dgl/bash/prep_data.sh"  # Replace with the actual path to your bash script
+    bash_script = base_directory + "jobs/prep_data.sh"
 
     # Use ThreadPoolExecutor to process YAML files concurrently
     max_workers = os.cpu_count()  # Use all available CPU cores

root_gnn_dgl/jobs/salloc.sh

@@ -1 +1 @@
-salloc --nodes 4 --qos interactive --time 04:00:00 --constraint gpu --account=atlas --gres=gpu:4
+salloc --nodes 4 --qos interactive --time 04:00:00 --constraint gpu --account=trn007 --gres=gpu:4