https://www.reddit.com/r/DotA2/comments/6t50rk/the_international_7_ai_vs_dendi/
The DotA OpenAI project seems far more advanced than Ensage at this stage with much less development time - is it time to scrap Ensage and start over with a better base?
It looks like you're new here. If you want to get involved, click one of these buttons!
Comments
if only we able to get the Lane Control as same as the Open AI ! , omg we will be dominating ! , wouldnt mind paying 50$ for it ! :O
Wait did you guys know that OpenAI is like a multi-million dollar project? They have tons of professional coder.
You can't compare it to Ensage.
Lol Comparing Multi Million Dollar product to Ensage ? Imao Ensage Can Never Come To That Standards Not Even In Its Dreams, The Coders There Are Too Pros Not Like Our Community (No Offense)
Wait A Minute Paying 50$ ? Ensage Would have Gone Bankrupt ( Even If U Had Paid 100$, Even If Ensage Has Implemented It, There Will Be Bugs And Errors, The Support Is Too Bad And No One Would Buy It As xyz$(3-digit) would have gone to waste )
Its Just That I Haven't Seen Any High level Intelligent Coders Here That Can bring the OpenAI System, What your Asking Just Cant Be Done Here :'(
And it can't be done anywhere else anytime soon.
OpenAI's project for dota (current state only supports 1 lane, 1vs1, very limited action pool) needs tons of processing power for its training and inference phase.
e.g. https://www.nvidia.com/object/gpu-accelerated-applications-tensorflow-configurations.html
1) Not 5 but 10 highly educated, experienced, machine learning engineers.
2) We never aimed a bot, playing the game itself without any player help. I believe with scripts here, a 4k player can easily win in 5k, a 5k player can reach to 6k. (I know many people actually did it.)
3) It probably will never be implemented to any cheating engine, at least appropriately. The reason is some properties they use are only available in local host.