Ex-TikTok MLE: JEV could revolutionize recommendation and search systems
hardimanjames · x · 2026-09-24
A former TikTok MLE argues JEV—which brings general-purpose semantic reasoning into a discriminative inference paradigm (arbitrary context in, calibrated structured decisions out)—could reshape modern recommendation and search, systems that need decisions both smart and fast. Example: JEV + Perfectly profiles AI researchers across 500 ECCV 2026 papers, matching one of Claude's models in performance.
More from Models
- DeepMind chief says Gemini 4 is in post-training and could arrive well before year-end — rohanpaul_ai · 2026-09-24
- DeepMind chief hints Gemini 4 lands before year-end, says Google is "certainly at the frontier" — koraykv · 2026-09-24
- GRPO is functional testing but for agents — marktenenholtz · 2026-09-24
- CLM-8B hits SOTA 81.6% on DeepSWE with light finetuning, up to 9x faster inference — anshulkundaje · 2026-09-24
- Tencent Open-Sources Hunyuan-A13B: an 80B-Parameter MoE That Activates Only 13B — Tencent-Hunyuan · 2026-09-24
- Luna 6 small model 'very good' and 4x cheaper than Grok, claims Bindu Reddy — bindureddy · 2026-09-24