Yoav Goldberg: the 'post-training adds no knowledge' dogma is clearly no longer true
yoavgo · x · 2026-10-03
AI researcher Yoav Goldberg argues the popular dogma that "post-training shapes behavior rather than adding new knowledge" is clearly no longer true: knowledge in software engineering, math, and computer use is evidently added during post-training, likely alongside other kinds. He asks whether this phenomenon has been systematically documented, quantified, or ontologized anywhere.
More from Models
- OpenAI's remote chat feature now far less stable, users report — TheZachMueller · 2026-10-03
- GPT learns to pass tests, not to engineer: the RL reward-mismatch behind ugly code — xiaohu · 2026-10-03
- Local LLM setup: Strata lets a 7900XTX + 64GB RAM run Qwen Flash at 60 tok/s at 250K context — soyalemujica · 2026-10-03
- User reports OpenAI API account with $2,000+ spend wiped clean, Codex credits zeroed — RecursivelyYours · 2026-10-03
- No slowdown: Gemini, Bel, Fable and SSI models all launching this month, says account — iruletheworldmo · 2026-10-03
- ChatGPT Plus users report weekly usage cap cut to ~half, from $80-100 to $50 API-equivalent — pensuke89 · 2026-10-03