Follow

Please help, also boost

Imagine your computer/server/device had federated learning*, what would you like it to do?

*technique that trains an algorithm across multiple decentralized devices holding local data samples, without exchanging their data samples (only your device sees your data)

from what "private" datasets would you like your thing to learn and what would you like the algorithm to improve?

Show thread

@u I want something as fucked up as Tay but for the fediverse.

@crash_override the project must convince the grad school people that it will "contribute to the field of engineering" so I can't make a meme thing 😭

@u I think that this neural network might have written a cry for help: invidio.us/watch?v=BjngNWP9C5s

You can basically coerce the AI to write or respond in a certain way given the training data-set was deliberately chosen for that effect.

Still funny though.

@u artbreeder.com but federated (in any way, even just processing power so we dont have to rely on aws)

Sign in to participate in the conversation
pyyython.org

The pythonic instance!