Bitsota is a decentralized research network operating as Subnet 94 on the Bittensor blockchain. The project is focused on the evolution of machine learning algorithms through a competitive and distributed computing environment. Its primary objective is to transform artificial intelligence research into a global competition where participants collaborate to solve complex problems and advance the state of machine learning. The platform is designed to be problem agnostic, meaning it can be applied to various categories of research challenges. Currently, it places a significant emphasis on self-improving and self-generating AI. Bitsota draws inspiration from established research concepts such as automated machine learning and genetic programming to decentralize the process of algorithm discovery. The ecosystem consists of two main types of participants: miners and validators. Miners utilize their computational resources to develop and evolve machine learning algorithms, often using genetic programming techniques to find better solutions for specific tasks. These tasks can include established research benchmarks or hidden evaluation criteria. Validators are responsible for independently evaluating the submissions from miners. They verify the performance of the algorithms and ensure that progress is genuine. A unique aspect of the Bitsota model is its incentive structure, which prioritizes verified breakthroughs. Instead of rewarding mere participation or the provision of idle computing power, the network allocates rewards when a miner successfully produces a model that achieves a new high score or a state of the art result. This approach is intended to align the interests of the community with actual scientific progress, reducing wasted energy on unproven computational runs. The project is developed by Alveus Labs, an independent team that maintains the open source codebase on platforms like GitHub. While the specific identities of the individual team members are not publicly emphasized, the development process is community driven and transparent. The project provides a web interface for viewing challenges and utilizes social platforms like Discord and X for technical coordination among operators and researchers. In summary, Bitsota functions as a decentralized research service for entities that need AI problems solved and as a mining application for those who want to contribute creative algorithmic solutions and compute power. By leveraging the infrastructure of the Bittensor network, it aims to create an open marketplace for the continuous evolution of artificial intelligence.
Learn more
