Required libraries and installation instructions.

Step 1: Download the Prerequisites

chmod +x download_docker_and_data.sh
sbatch download_docker_and_data.sh

Step 2: Install other libraries

pip install requirements.txt

This is in case for lack of some libraries. If it is not working, put this 'install' code in main.py cause it is running in the containner.

wandb login
pip install requirements.txt
python3 train.py \ ...
                    ...

WandB

Just like the NNCV instruction.

1. Open the .env file using a text editor:

    nano .env

   If you are using MobaXTerm, you can also just open it using the file explorer windown on the right side.

2. Update the following variables:

   • WANDB_API_KEY: Your Weights & Biases API key (for logging experiments).
   • WANDB_DIR: Path to the directory where the logs will be stored.

3. Save and exit the file.

If not use the HPC, run the code to log in WandB:

wandb login

then paste your API key (can be found in WandB personal profile)

Steps to run the code

• If you are using the HPC to run the codes, also like the steps mentioned in the NNCV instruction:

Submit the job with the following command:

chmod +x jobscript_slurm.sh
sbatch jobscript_slurm.sh

• If run it on the local computer, run the code in the terminal:

python3 train.py \
   --data-dir ./data/cityscapes \
   --batch-size 64 \
   --epochs 100 \
   --lr 0.001 \
   --num-workers 10 \
   --seed 42 \
   --experiment-id "FinalModel_Fan" \

Codalab username and TU/e email address

• Username: GoFun5
• Tu/e email: f.wu@student.tue.nl