Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add PPO + Transformer-XL #459

Open
wants to merge 21 commits into
base: master
Choose a base branch
from
Open

Add PPO + Transformer-XL #459

wants to merge 21 commits into from

Commits on Apr 15, 2024

  1. initial commit of ppo trxl

    @MarcoMeter
    MarcoMeter committed Apr 15, 2024
    Configuration menu
    Copy the full SHA
    990238e View commit details
    Browse the repository at this point in the history

  2. removed video capture

    @MarcoMeter
    MarcoMeter committed Apr 15, 2024
    Configuration menu
    Copy the full SHA
    513d2f8 View commit details
    Browse the repository at this point in the history

  3. Switched from grayscale to RGB

    @MarcoMeter
    MarcoMeter committed Apr 15, 2024
    Configuration menu
    Copy the full SHA
    cd596c8 View commit details
    Browse the repository at this point in the history

  4. RGB obs reconstruction

    @MarcoMeter
    MarcoMeter committed Apr 15, 2024
    Configuration menu
    Copy the full SHA
    cab8e07 View commit details
    Browse the repository at this point in the history

  5. print reconstruction loss instead of total loss

    @MarcoMeter
    MarcoMeter committed Apr 15, 2024
    Configuration menu
    Copy the full SHA
    2a012eb View commit details
    Browse the repository at this point in the history

  6. Ensure that transformer memory length is not larger than max episode … 

    …steps
    @MarcoMeter
    MarcoMeter committed Apr 15, 2024
    Configuration menu
    Copy the full SHA
    2efe26a View commit details
    Browse the repository at this point in the history

Commits on Apr 18, 2024

  1. fixed enjoy.py in the case of Searing Spotlights

    @MarcoMeter
    MarcoMeter committed Apr 18, 2024
    Configuration menu
    Copy the full SHA
    8737485 View commit details
    Browse the repository at this point in the history

Commits on Apr 19, 2024

  1. added video capture support again after updating memory gym

    @MarcoMeter
    MarcoMeter committed Apr 19, 2024
    Configuration menu
    Copy the full SHA
    5ef2e07 View commit details
    Browse the repository at this point in the history

Commits on Apr 21, 2024

  1. default hyperparameters

    @MarcoMeter
    MarcoMeter committed Apr 21, 2024
    Configuration menu
    Copy the full SHA
    4f88baa View commit details
    Browse the repository at this point in the history

Commits on Apr 22, 2024

  1. remove unnecessary padding from TrXL memory, if applicable

    @MarcoMeter
    MarcoMeter committed Apr 22, 2024
    Configuration menu
    Copy the full SHA
    9974f20 View commit details
    Browse the repository at this point in the history

  2. print SPS

    @MarcoMeter
    MarcoMeter committed Apr 22, 2024
    Configuration menu
    Copy the full SHA
    7761993 View commit details
    Browse the repository at this point in the history

  3. updated pyproject.toml because of memory-gym 1.0.2

    @MarcoMeter
    MarcoMeter committed Apr 22, 2024
    Configuration menu
    Copy the full SHA
    5374027 View commit details
    Browse the repository at this point in the history

  4. fixed comment

    @MarcoMeter
    MarcoMeter committed Apr 22, 2024
    Configuration menu
    Copy the full SHA
    fe85e75 View commit details
    Browse the repository at this point in the history

Commits on Apr 24, 2024

  1. refactored code + added a comment

    @Horrible22232
    Horrible22232 committed Apr 24, 2024
    Configuration menu
    Copy the full SHA
    eb36a32 View commit details
    Browse the repository at this point in the history

Commits on Apr 25, 2024

  1. added annealing entropy coefficient. learning rate anneals also from … 

    …initial to final value.
    @MarcoMeter
    MarcoMeter committed Apr 25, 2024
    Configuration menu
    Copy the full SHA
    2fe62d4 View commit details
    Browse the repository at this point in the history

Commits on May 27, 2024

  1. aligned monitoring of losses and further metrics

    @MarcoMeter
    MarcoMeter committed May 27, 2024
    Configuration menu
    Copy the full SHA
    86e59ef View commit details
    Browse the repository at this point in the history

  2. slight adjustments due to pre-commit, pre-commit still fails due to u… 

    …nused imports, however these imports are necessary for the used environments to be registered
    @MarcoMeter
    MarcoMeter committed May 27, 2024
    Configuration menu
    Copy the full SHA
    62fea9d View commit details
    Browse the repository at this point in the history

Commits on Jun 18, 2024

  1. heads share parameters again for reproduction purposes, set default h… 

    …yperparameters for succesfull MMG training
    @MarcoMeter
    MarcoMeter committed Jun 18, 2024
    Configuration menu
    Copy the full SHA
    3297da4 View commit details
    Browse the repository at this point in the history

Commits on Jun 25, 2024

  1. fixed entropy for multi-discrete action spaces

    @MarcoMeter
    MarcoMeter committed Jun 25, 2024
    Configuration menu
    Copy the full SHA
    8ff1edf View commit details
    Browse the repository at this point in the history

  2. advantage normalization is off per default, added hidden layer after … 

    …TrXL
    @MarcoMeter
    MarcoMeter committed Jun 25, 2024
    Configuration menu
    Copy the full SHA
    b9f8131 View commit details
    Browse the repository at this point in the history

Commits on Jul 19, 2024

  1. set max_episode_steps for endless environments

    @MarcoMeter
    MarcoMeter committed Jul 19, 2024
    Configuration menu
    Copy the full SHA
    931bc81 View commit details
    Browse the repository at this point in the history