TypeSafe AI's RLCD post-training路线挑战 RLHF 正统,引发关注
FrankFelixAI · x · 2026-09-19
Commenting on TypeSafe AI's 'Jev' model, calebfoundry argues 2022's shift from token prediction to instruction following was pivotal, but scaling RLHF cost the field other possibilities. TypeSafe AI claims RLCD (Calibrated Decision) as an alternate post-training path that could chart a genuinely different route than the orthodox RLHF scaling approach — the training-paradigm debate matters more than the model's raw capability.
Related event: TypeSafe AI's Jev model and RLCD method spark debate(3 posts)→
More from Models
- Musk cites unsustainable Grok Bot costs as Hermes offers one-command agent migration — Teknium · 2026-11-04
- Hobbyist rewrites Qwen 0.6B weights with 39.8M tokens, no gradients, in 3h12m — Altruistic_Heat_9531 · 2026-11-04
- Same leak, reshared: ChatGPT's five family plan prices spotted again — heyshrutimishra · 2026-11-04
- KOL visits StepFun's Shanghai office, hints a powerful new model is coming — teortaxesTex · 2026-11-04
- Fable 5.5 delayed as Anthropic reportedly bets OpenAI's Astra 6.2 won't beat it — haider1 · 2026-11-04
- ByteDance open-sources Axion 1.0-35B agent foundation model with full reproducible recipe — pmttyji · 2026-11-04