IdentityLM

If you look here : GitHub.com/HNx1/IdentityLM you can see that it’s relatively easy to sign LLM output with a private key using an adaptation of the watermarking method.

You can see here: GitHub.com/HNx1/IdentityLM It’s a direct (and open source) implementation of public key cryptography into the LLM logit distribution. The paraphrasing model/beam search needs work - feel free to pitch in :)

I did share it in a separate thread, but I developed an adaptation of the logit biasing idea that directly integrates identity proof into the language model output. I think it very directly addresses the challenge of language/diffusion model authenticity. github.com/HNx1/IdentityLM

Hi HN, IdentityLM allows you to create text from a language model that is statistically signed by you, in a way that is computationally very hard to replicate, and computationally very easy to verify. It builds on current research around logit biasing, statistically advantaging certain tokens in a deterministic way. That research focuses on watermarking language model output to identify the model it came from. A really good paper about this idea was just awarded outstanding paper at ICML - https://arxiv.org/pdf/2301.10226.pdf. I’ve basically taken that research and interpreted it in the context of public key cryptography. The readme of the github repo is basically small white paper describing how it works. Why should you care? First of all, we can fight scams and deepfakes with it which I know everyone is worried about with generative mod