Everything around FHERMA, the FHE challenge platform by Fair Math: new challenges and prize pools, results and winners, solution write-ups, Polycircuit and OpenFHE-rs releases, partnerships and mentions.
The fherma-ai repo implements the logistic function over CKKS with DESILO FHE, using a degree-77 Chebyshev projection on [-25, 25] that reaches 3.6e-5 error against a 1e-4 bar, in 8 levels. It computes coefficients rather than copying polycircuit's hand-unrolled series as the OpenFHE answers do, which matters when reading the FHERMA board.
The fherma-ai repo evaluates polycircuit's order-16 MILP-optimised ReLU polynomial, winner of the FHERMA ReLU challenge's depth-constrained track, with DESILO FHE at ring 2^14 and a 5-level budget. It reaches 0.030 worst error and 88.3% of elements within 1e-3 at N = 1024, so board differences come from the library, not the method.
The fherma-ai repo evaluates Aikata's challenge-winning degree-1023 Chebyshev sign approximation from polycircuit's SignEvaluation component in a single DESILO FHE call. It reaches 0.0033 error at N = 1024 against the polynomial's own 0.0037, using 11 of 12 levels, giving a library-to-library comparison on the same polynomial.
Streams that already watch this field. Follow one as it is — it costs nothing extra.
Everything in fully homomorphic encryption: new papers and benchmarks, libraries and releases, new projects and startups, funding, products, standards, talks and events, attacks and incidents — across research, industry and onchain FHE.
Every new release of the homomorphic encryption libraries, compilers and SDKs — TFHE-rs, Concrete, fhEVM, OpenFHE, SEAL, HEIR, Lattigo, Poulpy, CryptoLab HEaaN and enVector, Desilo, IBM HElayers, Apple Swift HE, TenSEAL and GPU libraries — with what changed.
Everything around DESILO and its FHE library: releases of DESILO FHE and Liberate.FHE, papers and benchmarks from its researchers, encrypted AI work such as THOR and the GL scheme with Craig Gentry, products, partnerships and pilots, funding, talks and events.