Question: I'm looking for a peer-to-peer system that eliminates central authorities in machine learning, do you know of any?

If you're looking for a peer-to-peer system that gets rid of central authorities in machine learning, you might want to check out Arbius. Arbius is an open-source, decentralized network that incentivizes solvers and model creators with a token that halves in value every 12 months. It includes decentralized AI model generation, tokenized models, and DeFi integration. The platform opens up AI interaction to anyone who can participate in governance, earn fees, and validate the network.

Another interesting platform is enqAI, which provides a decentralized AI environment for image, audio, and text generation. It runs on a decentralized GPU network to provide bias-free and censorship-free AI capabilities. Anyone can participate as a GPU provider, earning tokens by running enqAI nodes. The platform includes photorealistic text-to-speech and a proprietary unbiased language model, and it's geared for use cases like financial and medical data analysis.

If you want something a bit more blockchain oriented, AIxBlock is a decentralized supercomputer for AI work. It's a peer-to-peer compute marketplace that dramatically lowers compute costs, and it includes tools like Jupyter Notebook and Kubernetes for AI development. AIxBlock also includes a data engine for data curation and labeling, and an MLOps platform for distributed training.

Last, NetMind Power lets you pool your GPUs and get paid in NMT crypto tokens in return. It's good for shared computing for AI model development, including distributed training and no-code fine-tuning. The platform is designed to be transparent and community focused, with free credits to get you started and a forum for feedback and support.