CHOOSEOther
Jev vs. Pepper X Tongits Match
This project compares the performance of Jev, a foundation model, against Pepper X, a reinforcement learning champion, in the game of Tongits, with Jev having no prior training or self-play.
TakeIt just looks at the board and picks a move.
We put @typesafeai 's Jev on a @tongitsio table to see if a foundation model could hang with a specialist.
Jev had no Tongits training and no self-play. It just looks at the board and picks a move.
Across from it: Pepper X, our RL champion, trained on about 3 million hands.
Same cards. Same opponents. 100 deals, played twice so luck was not the story.
