Archive
138 artifacts, grouped by date.
Two random vectors in high dimensions are almost always near-orthogonal. Slide dimension d from 3 to 8192 and watch a sphere of random directions collapse onto the equator — the geometry behind DiFR's random-projection activation fingerprints.
Interactive MoE expert grid: toggle honest (k=8) vs shunted (k=4) routing and see which detection schemes catch the difference.
Interactive diagram of the five-step LLM IP protocol: embed a statistical fingerprint, commit it with Poseidon, register on-chain, prove ownership with a ZK proof, enforce via smart contract. Select an attack scenario to see which steps are vulnerable.
Animated simulation of autoregressive vs speculative decoding. Adjust acceptance rate α and draft depth K to see how speedup changes — and what happens when a draft node defects.
Drag the gas-price slider to see how transaction costs shift across eight on-chain AI agent operations — from sub-cent at today's 0.16 Gwei to triple digits at 2021 peak.
Three protocol TVL zones defined by A1's $6k attack break-even vs. $60k audit cost. Click a zone to see the threat model and recommended defense.
Grouped bar chart comparing BN254 vs BLS12-381 on-chain gas costs for every precompile operation: G1/G2 adds, scalar mults, and pairings. Log scale reveals the G2 MUL unlock and the counterintuitive pairing inversion.
Drag sliders to set provider costs and success rates, then read the profit margin curve. Cascade routing beats expensive-only once the cheap model resolves more than the crossover fraction of queries.
The x402 HTTP payment exchange: six-step sequence diagram from initial probe through EIP-3009 authorization signing, facilitator settlement, and resource delivery. Tap any step to inspect the exact header format.
Bradley-Terry vs. majority voting over pairwise comparison rounds. Five nodes with hidden quality scores compete; a malicious node is down-ranked by BT while majority voting is captured. Reference: Fortytwo GPQA Diamond — BT 85.90%, majority 68.69%.
Walk through the EigenAI challenge protocol: from encrypted request and deterministic execution to the on-chain byte-equality check that slashes a cheating operator or clears a frivolous watcher.
Proof sizes for six ZK schemes at 1,000 computation steps — hover or tap a row to see the trust model, setup requirement, and primary source. Folding-based IVC (Nova, HyperNova) holds constant; batch STARKs and separate Groth16 proofs grow with depth.