CommitLLM: a cryptographic protocol for verifying LLM inference
Researchers from CCSD and LambdaClass present CommitLLM, a commit-and-audit protocol to verify that a provider actually ran the language model it claims to have run.
Read more →Tag: research
1 posts found