I Made Some AI Creatures, This is how you can simulate them

Do you want to simulate life and see it evolve before your very own eyes? Then go and try out this totally not a virus Evolution Simulator (Vitanova) I made.
Developing and bug fixing code is a nightmare.
GitHub repo for the project github.com/4t-2/EvolutionSimu...
The Bibites Project, an inspiration for this / @thebibitesdigitallife
The Life Engine, an inspiration for this thelifeengine.net/
Discord Server - / discord

[@scasz]

Join the discord server if you want to try this out for yourself, if you have any bug reports or suggestions you can also say them there too. If you are the type of guy to do so then there is also the link to the github repo for this project so you can go and compile it yourself (though if you are on windows you will need to read the cmake file so you know where to put all the dlls and headers, I might fix that I might not) Discord - discord.gg/spGzjEwkHv GitHub - github.com/4t-2/EvolutionSimulator

[@EightLittleBears]

The hybrid approach sounds awesome! At that point it’s not just an evolution sim… it is evolution at the macro level (I.e. NEAT simulates generational change in brain physiology) and behaviour at the micro level (I.e. reinforcement simulates changes that come with age for individual creatures). I could see some absolutely fascinating behaviours coming from this (parenting, teamwork, etc.). Great work!

[@DefaultFlame]

The hybrid approach sounds pretty fascinating. PPO for short term improvement, NEAT for long term, or something like that.

[@revimfadli4666]

Could become a more modern version of Creatures' Hebbian RL + evolved reward system

[@Balacynka]

13:27 dude, this is such a life situation! I understand you so well

[@revimfadli4666]

Truly one of the evolution simulations ever (boosting the data science stuff)

[@kodirovsshik]

Yay a new video c: UPD: I felt all the pain you had to share, all of it is so relatable :c

[@revimfadli4666]

Really like the idea of blending ontogenic/lifelong learning via RL with generational learning via genetic algorithm. Tbh Creatures already did that by combining Hebbian reinforcement with evolved reward/punishment system(which comes in 2 flavors: on-policy hormones and off-policy instinct+dreaming). Asicerik also did a simpler version in his artificial neural network simulator life net(fixed food and poison reward and punishments, respectively). Maybe use that one for simplicity? Or, if you want uniqueness, Geoffrey Hinton's forward-forward algorithm or fast weights could be relatively easy options I haven't seen in any evosim. The latter might gel well with genetic algo(evolve the slow weights, learn the fast weights), and can be tweaked further with a "learning output" to modulate how much hidden node outer product memory is stored, and if it's negative or positive

[@0d1z]

finally a new video

[@themasteratstuff9357]

Can you please make a iOS or android version?

[@crabbydisk7658]

QT: exists

[@TitanLordofPizza]

hmmm nifty

[@revimfadli4666]

Indeed

[@FacePalmProduxtnsFPP]

...windows isn't hard to code for... not any more or less than anything else.. Sounds like you're very comfortable with Linux in comparison however.

[@Lawtro]

This is why I use Java instead of c++ 😂❤

[@Stanisaw1z34t]

First like

[@taurasandaras4699]

this is a comment

[@BeneDiv-lu4gx]

Subbing so I can come back at this channel at a later date when you figure out how to speak in a proper way. Not an insult, just your english isn't clear at all due to the accent.

[@user-dt8mf8nt2v]

his accent is fine. no need to tell him to "speak in a proper way", just turn subtitles on (that's kind of what they're there for).

[@BeneDiv-lu4gx]

@@user-dt8mf8nt2v imagine being so entitled to criticism that isnt headed toward you.

[@notreallykeven86]

his accent sounds fine to me

[@visheshl]

You spent thousands of hours on unimportant stuff when you could have just used some game engine like unity... And instead focused on what's important that is neural networks and evolution.