run.slurm

#!/bin/bash
#SBATCH --job-name=meshgpt_training_job
#SBATCH --output=print_output_%j.log      # Output file (%j for job ID)
#SBATCH --account=module-mlp
#SBATCH --error=print_error_%j.log        # Error file
#SBATCH --time=7-00:00:00                   # Max runtime (HH:MM:SS)
#SBATCH --partition=PGR-Standard                # Partition (use appropriate partition)
#SBATCH --ntasks-per-node=1                      # Number of tasks (adjust as needed)
#SBATCH --gres=gpu:8
#SBATCH --cpus-per-task=2

# Print job details
echo "SLURM Job ID: $SLURM_JOB_ID"
echo "Node List: $SLURM_NODELIST"
echo "Tasks per Node: $SLURM_NTASKS_PER_NODE"
echo "Total Tasks: $SLURM_NTASKS"
echo "CPUs per Task: $SLURM_CPUS_PER_TASK"
echo "GPUs per Task: $SLURM_GPUS_ON_NODE"


# Run the Python script
#srun python meshgpt-pytorch/train_autoencoder.py
source ~/venv3.12/bin/activate
# srun accelerate launch --multi-gpu train_auto.py
# srun accelerate launch --multi-gpu train_shapenet.py
# srun accelerate launch --multi-gpu continue_auto.py
# srun accelerate launch --multi-gpu inspect_auto.py --quant lfq --codeSize 4096
# srun accelerate launch --multi-gpu train_shapenet.py --quant lfq --codeSize 8192 --data shapenet
srun accelerate launch --multi-gpu generate_mesh.py --quant lfq --codeSize 4096 --data shapenet