ramez: generational jumps pay for 2-3 orders of magnitude of tokens, suspects heavy math RL
ramez · x · 2026-09-14
Responding to tobyordoxford's scaling-curve analysis, ramez agrees the worse slope is interesting but argues the dramatic between-generation jump pays for 2-3 orders of magnitude of tokens. He suspects the model was heavily RL-trained on formal math, making gains possibly math-specific. Key point: bigger baseline jumps can offset declining inference-scaling returns.
More from Models
- Anthropic removes 5-hour usage cap for verified Claude Ultra subscribers — james_mtc · 2026-10-31
- DeepSeek's silent downtime upgrade caught by the pelican bicycle test — teortaxesTex · 2026-10-31
- ARC Prize 2026 closes in 3 days as Yi-Chia Chen tops ARC-AGI-3 at 75.56% — GregKamradt · 2026-10-31
- Fable 5.5 Max stuns with gorgeous SVG of a pelican riding a bicycle, headed to Claude Code — ChrisGPT · 2026-10-31
- OpenAI reportedly to train on users' Daybreak sessions with no opt-out — dyn___ · 2026-10-31
- NVIDIA Nemotron 3.5 scales LLM self-verification, +9-10 points on agentic benchmarks — sarahcat21 · 2026-10-31