How you can help

[152334H]

Thank you for checking out this page!

This project is very new and will probably change substantially really quickly, but right now, the most pressing problems I can think of are,

1. Run training experiments! I have no idea what the optimal parameters are yet, and it'll be easier to figure out with more experimental data.
2. Build better training interfaces. We could use a colab notebook, but anything that improves the current admittedly janky state of training could help.
3. See the rest of the issues page. Problems that I would really like help from the public with are tagged with good first issue and help wanted, but any issue can be tackled.

[rikabi89]

I have no programming/coding knowledge but like you guys I am trying to find out what is the "optimal" training settings and will share anything note worthy. Is it worth setting up a discord?

[8tm]

I can help with basics - not with AI but with functional issues like saving space, fixing colab template, create additional options.
Discord can be a good place to start to discuss some ideas.

[rikabi89]

https://discord.gg/gzTxHe3auT

I setup a channel on my discord. We can use this unless the owner wishes to setup an official channel. In the meantime everyone is welcome to share ideas.

[TEJASAMA-TECH]

Hello, I have finetuned the ARmodel for 1000 iterations. Now, how can I make an inference to that checkpoint?