Subnet 1 Apex subnet drives algorithmic innovation across diverse problem domains. Each pursuit of the best solution takes place within a Competition, which consists of multiple Rounds of evaluation where participants (miners) submit their Python-based algorithms through the Apex CLI and earn rewards based on their performance. Validators continuously evaluate these submissions, scoring them against benchmarks and distributing blockchain-based rewards to miners whose solutions perform best, creating a trustless incentive mechanism that encourages innovation and optimization across various computational challenges.
See miner docs for an overview on the Apex CLI and incentive mechanism.
See validator docs for an overview on validator setup.
Matrix Compression V1: Lossless Compression: A lossless compression competition exploring how small neural activations, both forward and backward, can be compressed while still retaining all their original information.
Battleship: A strategy game competition in which miners compete to sink ships on a 10x10 grid in as few turns as possible. A Bittensor twist on the classic 1931 game.
Matrix Compression V2: Lossy Compression: A lossy compression challenge focused on reducing bfloat16 matrices while preserving enough information for effective training, allowing minor loss.
Visit the SN1 channel in the Macrocosmos Discord server or the Bittensor Discord server to chat, ask questions, submit feedback, and more.