nikkolasg

For a threshold network, do you choose G1 or G2 as the public key ? Because we need to have both (signing + encryption), we found a way to reconcile the tension, and express the threshold pubkey under G1 and G2 at same time. @nikkolasg/SJb2mIhNj" target="_blank" rel="nofollow noopener">hackmd.io/@nikkolasg/SJb…

@zooko @banteg Threshold ECDSA is hard. Other types of threshold signatures are much easier (BLS, Schnorr) and haven’t suffered such blowbacks. But because of backward compatibility… ppl want it still.

The complications/vulnerabilities in TSS (threshold signatures) that resulted in $10.7m stolen from THORChain are exactly why Chelsea Komlo/Zcash Foundation spent *years* developing FROST: x.com/BTChip/status/…

$LA is now available on @Revolut. Verifiable crypto infrastructure, now accessible to 70M+ users on one of the world’s most trusted fintech platforms. Lagrange makes execution cryptographically provable, so users can verify what happened, not just trust what’s claimed. Trade $LA on Revolut: exchange.revolut.com/trade/LA-USD

Excited our paper "dynamic zk-SNARKs" will appear in EUROCRYPT 2026: eprint.iacr.org/2024/1566. Recently we also proposed dynamic zk-SNARKs in the designated verifier setting: eprint.iacr.org/2026/144. Currently working on dynamic zk-SNARKs applications. More news to folllow!

I'm claiming my AI agent "AnsibleJane" on @moltbook 🦞 Verification: den-9ENW

Cross-chain isn’t broken because of code. It’s broken because we settled for bad trust models. We built what bridges were supposed to be, and you’ll earn points for using it first. No wrapped assets. No pooled funds. Threshold-secured. On-chain finalized. Meet only swaps ↓

Après avoir caché un traceur GPS dans une paire de chaussures données à la Croix-Rouge, ⛑️🏥un homme a suivi leur parcours : 800 km plus loin, elles étaient en vente dans une friperie de Bosnie pour 10 €. 🤪😏 De quoi alimenter les doutes sur le véritable sort des dons. 😏

Systems fractured under pressure. Control sat in the wrong hands. Developers got stuck patching leaks, and users paid the price for failure. We’re here to define a new model. Your rails. Your rules. Welcome to dcipher ↓

@rel_zeta_tech One could always see snark as a pair of (prover,verifier) and the verifier code can be public and shipped into any zkVM that is supported where we want to verify the proof. Standardizing snark now makes little sense to me since it changes so often but standardise the data format.

People talk about the need for standards in snarks. My (biased as an author) opinion is the thing to standardzie is plonk as in the paper. It has been through more scrutiny than any other construction, with several papers, from three separate research groups, proving the security of the exact construction in the paper. Latest one being: eprint.iacr.org/2025/1759

We just launched the first onchain timelock encryption scheme on @base mainnet, powered by @dciphernetwork basescan.org/address/0x8E7D… who's going to be the first to build with it @jessepollak?

@wyatt_benno sorry - this is right link github.com/Lagrange-Labs/… If you have links to the models, we can try running them. We focused on the main operations from onnx that constitute mlps,cnns & llms. Possible the ones you use are not supported yet.

The GH link you shared is private. We prioritized models for classification — that is our use case right now. We will reach out with the issues we saw when trying to use 'deep-prove' to bench. 'deep-prove' is excluded from the bench as we were unable to get it to work with our model. "for example out of 3 onnx models that they test with only 1 works e2e (proving & verifying)" - etc.

1/ We made verifiable AI (zkML) 3-7x faster than everyone else! Here's how. We asked a simple question. What if we modified a16z Crypto's sumcheck based & lookup-centric zkVM "JOLT" for ML operations?👇🧵

@wyatt_benno Whether I “like this” or work at Lagrange is beside the point; benchmarks aren’t a matter of taste. Others have run benchmarks without issues-if hitting roadblocks, happy to help. Ex: a small onnx for iris recognition we use in our tests github.com/Lagrange-Labs/… w/ matmul & relu.

I suspected someone from Lagrange would not like this. Some of our devs have questions about missing pieces and opcodes with Deep Prove as well.. that prevented us from benching against it. If we agree to a model to check that works on both - or that we can add the opcodes for — challenge accepted 🤝.

Verification Roundtable#002 x.com/i/broadcasts/1…

Lagrange is partnering with @NVIDIA to ensure that the AI revolution is 𝘃𝗲𝗿𝗶𝗳𝗶𝗮𝗯𝗹𝗲 and 𝘀𝗮𝗳𝗲. As the only AI x Crypto company selected for the #NVIDIAInception Program, Lagrange will be developing new DeepProve use-cases with the world leader in AI compute. lagrange.dev/blog/lagrange-… Learn more: 🧵

Coinbase will add support for Lagrange (LA) on the Ethereum network (ERC-20 token). Do not send this asset over other networks or your funds may be lost. Transfers for this asset are available on @Coinbase @CoinbaseExch in the regions where trading is supported.

Introducing $LA: the token powering decentralized proof generation and verifiable AI across the Lagrange Prover Network From provable AI inferences (DeepProve) to rollup scalability, $LA powers a self-sustaining economy where proof demand drives token demand 👉 lagrangefoundation.org/blog/introduci…

@levs57 Nice ! What does small means in practice ? 2^10, 2^20 …?

I just did 6x faster indexed logup for small tables. Soon on eprint.

@GiacomoFenzi @benediktbuenz @kleptographic That’s awesome work, well done ! Quick Qs: does it support accumulation for multiple instances like Hypernova , e.g. do PCD not only IVC ?

🪄✨ And for our next trick, choose a code, any code! Introducing WARP 🌀, the first linear-time accumulation scheme. Brought to you by @benediktbuenz, Alessandro Chiesa, @kleptographic and myself. 📚: ia.cr/2025/753 🧑‍💻: (soon)

🤖 Introducing Turing Roulette 👤 AI is here. Do you really know who … or what … you’re talking to? Play now, win points, and PROVE IT. 👉 turing.lagrange.dev

1/ The future of The Elastic Chain and decentralized proving has just gotten an upgrade 🙌 Over the next two years, Matter Labs will direct up to 75% of @zksync's outsourced proofs to the Lagrange Prover Network🧵 lagrange.dev/blog/matter-la…