kalomaze: 'Fable 5' Partly Suffers from Undercooked Post-Training
kalomaze · x · 2026-09-11
RL researcher kalomaze suggests part of 'fable 5's' problems come from undercooked post-training. Citing repligate's observation, he notes that holding the RL process roughly constant, more capable models come through relatively less scathed — if a model can straightforwardly one-shot a problem, it doesn't need to game the grader, reducing reward-hacking-style distortion.
More from Models
- ToneBench eval: GPT-6.1 Terra ranks #23 at writing — cheaper, not better, than GPT-5.6 Terra — OnlyProggingForFun · 2026-10-28
- Cursor user complains it silently switches to Grok, which fails basic tasks — kylegawley · 2026-10-28
- Musk says Grok Bot keeps improving, calls Vault feature a game changer — beffjezos · 2026-10-28
- Claude refuses to gather legitimately public information, sparking guardrail backlash — mkheck · 2026-10-28
- DesignArena launches Code to Video Arena: models direct 15-45s films purely in code — yshan2u · 2026-10-28
- ChatGPT's 'yes, but with an important clarification' often precedes pure nonsense — Crixusgannicus · 2026-10-28