Will we find tree search in a vanilla transformer trained on chess games by 2026?
7
62
Ṁ124Ṁ288
2026
22%
chance
1D
1W
1M
ALL
Inspired by https://adamkarvonen.github.io/machine_learning/2024/01/03/chess-world-models.html
I won't bet in this market.
Get Ṁ600 play money
Sort by:
https://arxiv.org/abs/2402.04494
TLDR:
... we train a 270M parameter transformer model [on chess]
Lichess blitz Elo of 2895 against humansWe also show that our model outperforms AlphaZero's policy and value networks (**without MCTS**) and GPT-3.5-turbo-instruct.
Related questions
Will AI be able to generate correct images of a chess game in 2024?
15% chance
Will chess be solved by 2040?
12% chance
Will superposition in transformers be mostly solved by 2026?
54% chance
Will an AI by OpenAI beat a super grandmaster playing chess by 2028?
53% chance
Will Transformer based architectures still be SOTA for language modelling by 2026?
66% chance
Will chess be solved by 2100?
27% chance
Will transformers still be the dominant DL architecture in 2026?
57% chance
Will a Language Model under 10B parameters play chess at Grandmaster level by 2050?
39% chance