This is a containerized machine learning (and other methods) trainer with swarm capabilities. It has primitives for GPU acceleration. Sort of resembles a miner?
Version 0.0.5B introduces collaboration between bots visa-vi shared context file, individual bot context file is also retained. The highest quality responses are stored, the others are discarded. This is not a fully new version as it seems to overall behave poorly beside 0.0.5, but with the new features it will eventually be worth it as we improve the script some more. It’s a known issue that the bot will retain old data for too long. For this I have introduced a decay factor but this has problems of its own, try turning it down below ~0.95 if it’s remembering bad data for too long.
V0.0.6 RELEASED!
Sorry I took so long, I’ve surely been busy! I realize that this was time-sensitive, but maybe it’s still use to someone, such as like a person attempting to not have all their crap usurped by hidden copyright/NDA agreements? If so, you will need to borrow heavily from the dates located in the .zip to understand the timeline, keeping in mind that progress is not always forward. Isn’t that true in much of life? Thank you, and God bless each and every one of you.
