BetaZero

a diffusion climb generator for system boards

Last verified:

Visit BetaZero

What is BetaZero?

BetaZero is a free web application powered by a diffusion machine learning model (ClimbDDPM) that generates climbing board problems for home wall owners. It was created by Evan McCormick to solve the variety problem he experienced with his own homewall, where he wanted the joy of discovering random climbs he wouldn't have come up with himself.

The tool uses a diffusion model trained on the BoardLib dataset to automatically generate climbing routes for system boards. It currently supports three major training boards: the Kilter Board (16' with 18ft=11ft climbs), the Tension Board 2 (TB2), and the Decoy Board (12' with 8+++12 configuration). Users can generate climbs, edit them, and share them with others.

BetaZero is designed for home climbing wall owners, particularly those with system boards who want fresh training problems without relying on friends to couch-set routes or trying international boards through other apps. It's especially useful for climbers in areas where friends with different walls can't physically test routes on their specific board.

The application is completely free to use and represents version 3 of the creator's homewall climb generator, evolving from an LSTM model trained on his own wall to a diffusion model trained on the broader BoardLib dataset. The creator has a functional API for uploading and editing walls, which is how the current boards were added to the database.

Future development plans include adding an account/security system, allowing users to upload and set climbs for their own custom walls, and training an improved diffusion model based on lessons learned from the initial training run.

BetaZero pricing

Pricing model: Freemium

BetaZero is completely free to use. There is no paid tier, subscription, or premium features. The application is offered as a free public resource for generating board climbs using machine learning. The creator built it as a hobby project to solve his own home wall variety problem and shares it freely with the climbing community.

BetaZero pros

  • Completely free to use with no paid tier
  • Powered by advanced diffusion model (ClimbDDPM) for high-quality climb generation
  • Supports three major system boards: Kilter, TB2, and Decoy
  • Generates climbs you wouldn't come up with yourself for variety
  • Allows users to edit generated climbs
  • Enables sharing climbs with other climbers
  • Open-source project on GitHub
  • Trained on comprehensive BoardLib dataset
  • No account required to start using
  • Solves the couch-setting problem for remote friends
  • Works for home wall owners specifically
  • Created by an active climber who understood the problem firsthand
  • API already exists for uploading custom walls
  • Quick generation of multiple climb options
  • Addresses variety gap in home climbing experience

BetaZero cons

  • No account/security system yet implemented
  • Cannot yet upload custom walls through the UI (only via API)
  • No permissions system to control who can edit/see walls
  • Limited to only 3 board types (Kilter, TB2, Decoy)
  • No scalability improvements implemented yet
  • No improved diffusion model trained yet
  • Cannot set climbs for custom/home-built walls through UI
  • No beta videos or community content yet
  • No progress tracking features
  • Single developer project with limited resources

Frequently asked questions about BetaZero

What is BetaZero?

BetaZero is a free web application that uses machine learning (specifically a diffusion model called ClimbDDPM) to automatically generate climbing board problems for home wall owners. It was created by Evan McCormick and is currently hosted at betazero.live.

What climbing boards does BetaZero support?

BetaZero currently supports three major system boards: the Kilter Board (16' with 18ft=11ft climbs), the Tension Board 2 (TB2), and the Decoy Board (12' with 8+++12 configuration). These were added to the database using the creator's functional API for uploading walls.

How does BetaZero generate climbs?

BetaZero uses a diffusion model called ClimbDDPM that was trained on the BoardLib dataset. This is version 3 of the creator's climb generator, evolving from V1 (ClimbLSTM trained on his own wall) and V2 (ClimbDDPM diffusion model).

Is BetaZero free to use?

Yes, BetaZero is completely free. It is described as a free web application and public resource for generating board climbs using machine learning. There are no paid tiers or subscription fees.

Can I edit the climbs generated by BetaZero?

Yes, BetaZero allows users to generate, edit, and share board climbs. The application has editing functionality built in so you can modify generated problems to better suit your needs.

Can I upload my own custom wall to BetaZero?

Not yet through the user interface. The creator has a functional API for uploading and editing walls (which is how the current boards were added), but there is no UI for users to upload custom walls yet. This is planned for future development along with an account/security system.

Why was BetaZero created?

BetaZero was created because the developer built a home wall but noticed something was missing: variety. He wanted the joy of opening an app and scrolling to random climbs he'd never come up with himself. Friends' couch-set routes didn't work because they lived in Colorado while he was in upstate NY and couldn't test on his specific wall.

Can I share climbs with others?

Yes, BetaZero allows users to share the climbs they generate. The application is built with sharing functionality so climbers can share problems with friends or the broader climbing community.

What are the future plans for BetaZero?

Next steps include integration of an account/security system, allowing users to upload/edit/set climbs for their own custom walls, improving the scalability of the app, and training an improved diffusion model based on what was learned from the prior training run.

Is BetaZero open source?

Yes, BetaZero is an open-source project. The underlying code is available on GitHub under the repository ml-homewall-climb-generator by EvanMcCormick37, where the creator documents that BetaZero is live and free to use.

Use cases

Browse all AI tools on NeedAnAI