PPO Agent playing procgen-bossfight-easy

This is a trained model of a PPO agent playing procgen-bossfight-easy using the /sgoodfriend/rl-algo-impls repo.

All models trained at this commit can be found at https://api.wandb.ai/links/sgoodfriend/f3w1hwyb.

Training Results

This model was trained from 3 trainings of PPO agents using different initial seeds. These agents were trained by checking out 21ee1ab. The best and last models were kept from each training. This submission has loaded the best models from each training, reevaluates them, and selects the best model from these latest evaluations (mean - std).

algo	env	seed	reward_mean	reward_std	eval_episodes	best	wandb_url
ppo	bossfight	1	8.03125	6.37125	64		wandb
ppo	bossfight	2	9.90625	5.36982	64	*	wandb
ppo	bossfight	3	8.98438	5.94635	64		wandb

Prerequisites: Weights & Biases (WandB)

Training and benchmarking assumes you have a Weights & Biases project to upload runs to. By default training goes to a rl-algo-impls project while benchmarks go to rl-algo-impls-benchmarks. During training and benchmarking runs, videos of the best models and the model weights are uploaded to WandB.

Before doing anything below, you'll need to create a wandb account and run wandb login.

Usage

/sgoodfriend/rl-algo-impls: https://github.com/sgoodfriend/rl-algo-impls

Note: While the model state dictionary and hyperaparameters are saved, the latest implementation could be sufficiently different to not be able to reproduce similar results. You might need to checkout the commit the agent was trained on: 21ee1ab.

# Downloads the model, sets hyperparameters, and runs agent for 3 episodes
python enjoy.py --wandb-run-path=sgoodfriend/rl-algo-impls-benchmarks/goavynh9

Setup hasn't been completely worked out yet, so you might be best served by using Google Colab starting from the colab_enjoy.ipynb notebook.

Training

If you want the highest chance to reproduce these results, you'll want to checkout the commit the agent was trained on: 21ee1ab. While training is deterministic, different hardware will give different results.

python train.py --algo ppo --env procgen-bossfight-easy --seed 2

Setup hasn't been completely worked out yet, so you might be best served by using Google Colab starting from the colab_train.ipynb notebook.

Benchmarking (with Lambda Labs instance)

This and other models from https://api.wandb.ai/links/sgoodfriend/f3w1hwyb were generated by running a script on a Lambda Labs instance. In a Lambda Labs instance terminal:

git clone git@github.com:sgoodfriend/rl-algo-impls.git
cd rl-algo-impls
bash ./lambda_labs/setup.sh
wandb login
bash ./lambda_labs/benchmark.sh

Alternative: Google Colab Pro+

As an alternative, colab_benchmark.ipynb, can be used. However, this requires a Google Colab Pro+ subscription and running across 4 separate instances because otherwise running all jobs will exceed the 24-hour limit.

Hyperparameters

This isn't exactly the format of hyperparams in hyperparams/ppo.yml, but instead the Wandb Run Config. However, it's very close and has some additional data:

algo: ppo
algo_hyperparams:
  batch_size: 2048
  clip_range: 0.2
  clip_range_vf: 0.2
  ent_coef: 0.01
  gae_lambda: 0.95
  gamma: 0.999
  learning_rate: 0.0005
  n_epochs: 3
  n_steps: 256
  vf_coef: 0.5
env: procgen-bossfight-easy
env_hyperparams:
  is_procgen: true
  make_kwargs:
    distribution_mode: easy
  n_envs: 64
  normalize: true
env_id: bossfight
eval_params:
  deterministic: false
  ignore_first_episode: true
n_timesteps: 25000000
policy_hyperparams:
  activation_fn: relu
  cnn_feature_dim: 256
  cnn_layers_init_orthogonal: false
  cnn_style: impala
  init_layers_orthogonal: true
seed: 2
use_deterministic_algorithms: true
wandb_entity: null
wandb_project_name: rl-algo-impls-benchmarks
wandb_tags:
- benchmark_21ee1ab
- host_138-2-238-100

sgoodfriend
/

ppo-procgen-bossfight-easy