codekansas / ai-art-hackathon

AI art hackathon code

ai_art_hackathon

AI art hackathon code

Getting Started

Set some environment variables:

export RUN_DIR=/path/to/run/dir  # This is where all your training logs and checkpoints will be written
export EVAL_RUN_DIR=/path/to/eval/run/dir  # This is where all your evaluation logs will be written

It's also a good idea to set these variables:

export DATA_DIR=/path/to/data/dir  # This is where your datasets are stored
export MODEL_DIR=/path/to/model/dir  # This is where your pretrained models are stored

Check out the documentation here.

Usage

Train a ResNet18 model on CIFAR10:

runml train configs/image_demo.yaml

Train an RL PPO model on BipedalWalker:

runml train configs/rl_demo.yaml

Launch a Slurm job (requires setting the SLURM_PARTITION environment variable):

runml launch configs/image_demo.yaml launcher.name=slurm launcher.num_nodes=1 launcher.gpus_per_node=1