This project provides a platform for users who have interesting data but no ML experience to submit "requests for training". These requests include some reward in ETH. ML engineers can train models against this data and upload them. Depending on the accuracy of their model, they will earn some proportion of the total reward when the request's end date is reached. The deployed platform is currently expecting a local Ganache setup, the contract has not been deployed to any public network as of yet.
How It's Made
The project is implemented using React for the frontend, with the training data being uploaded to IPFS. Information about the proposal itself (name, description) is also in IPFS, while important info is stored in an Ethereum smart contract. We provide a CLI tool which can use Powergate to interact with Filecoin and IPFS to upload the data in a persistent way. We use Pinata for pinning some of the smaller data (such as the request and solution metadata). We use Fleek to host the frontend content.