a2c-QbertNoFrameskip-v4 / lambda_labs /starpilot_hard_benchmark.sh
sgoodfriend's picture
A2C playing QbertNoFrameskip-v4 from https://github.com/sgoodfriend/rl-algo-impls/tree/0760ef7d52b17f30219a27c18ba52c8895025ae3
c41aee4
raw
history blame
392 Bytes
source benchmarks/train_loop.sh
# export WANDB_PROJECT_NAME="rl-algo-impls"
BENCHMARK_MAX_PROCS="${BENCHMARK_MAX_PROCS:-1}"
ALGOS=(
"ppo"
)
ENVS=(
"procgen-starpilot-hard"
"procgen-starpilot-hard-2xIMPALA"
"procgen-starpilot-hard-2xIMPALA-fat"
"procgen-starpilot-hard-4xIMPALA"
)
train_loop "${ALGOS[*]}" "${ENVS[*]}" | xargs -I CMD -P $BENCHMARK_MAX_PROCS bash -c CMD