Engram borrows RecSys sparse hashing, may unlock 100T-parameter online-learning models
bingxu_ · x · 2026-09-11
bingxu notes that Engram essentially borrows sparse hashing from recommendation systems — nostalgia for RecSys veterans — and argues this could soon enable 100T-parameter models with online-learning features, adding he's bullish on SSDs.
More from Research
- COLM 2026 names three AdvML Rising Star Award winners, talks set for workshop — mohitban47 · 2026-09-11
- ClawBench Tests Agents on 144 Real Websites: Best Model Succeeds Only 33% of the Time — jiqizhixin · 2026-09-11
- SREGym from UIUC benchmarks SRE agents on real outages: GPT-5.6 Sol leads at 81% E2E — tianyin_xu · 2026-09-11
- Viral Claude materials-design joke sparks serious bet on parallelized alloy discovery — CatAstro_Piyush · 2026-09-11
- FetterFly: the spoof model that acts human when watched and fly when private — cephaloform · 2026-09-11
- Wiz launches Cyber Arena: 300+ real offensive-security challenges benchmark AI hacking ability — evilsocket · 2026-09-11