OpenWeights DAO: A Retrospective
Our six-month pilot in community-governed model training just wrapped. Here is what worked, and what we would not do again.
Every model provider publishes a model card describing what a system is and isn't
supposed to do. But a model card is a promise, not a guarantee — there's no way for a
third party to check, after the fact, that a given output actually came from the model
described.
Blockchain gives us a narrow but useful tool here: a public, tamper-evident record that a
specific proof was verified at a specific time. Combined with zero-knowledge proofs of
inference, this lets us turn "trust us" into "check for yourself" — without requiring anyone
to hand over proprietary model weights.
Blockchain gives us a narrow but useful tool here: a public, tamper-evident record that a
specific proof was verified at a specific time. Combined with zero-knowledge proofs of
inference, this lets us turn "trust us" into "check for yourself" — without requiring anyone
to hand over proprietary model weights.
This is the thinking behind Ledgermind, and we think it generalizes well beyond our own
work. We'd like to see verifiable inference become a standard expectation for any model
making consequential decisions.
This is the thinking behind Ledgermind, and we think it generalizes well beyond our own
work. We'd like to see verifiable inference become a standard expectation for any model
making consequential decisions.
Written By
Rohit Ingulkar
Project Manager
Lorem Ipsum dolor sit ameut, dolor lorem ipsum.
Share This Post