Bayesian model selection explains why proxy-scale winners can lose at target scale
BlackHC · x · 2026-09-09
Andreas Kirsch published an educational thread explaining a hidden pitfall in scaling: recipes picked as winners at proxy scale can lose at target scale.
Core framework: for an ideal Bayesian learner, "best model" maps to three criteria over the loss curve (dataset size vs loss):
- Endpoint (final height): validation loss of the final checkpoint (posterior-predictive loss)
- Total area: area under the curve = negative log marginal likelihood (evidence)
- Tail area: the second half = negative CLML
Whenever loss curves cross, these criteria rank models differently — the root cause of proxy-scale winners losing at target scale. The same geometry carries over to pretraining and in-context learning: k-shot evals read the endpoint, whole-prompt likelihood reads the area, sliding-window perplexity reads the tail, so the k-shot winner depends on k.
Takeaway: the selection criterion should follow the downstream use case, not convention. The full post includes the exact identity, flip-order results, a toy experiment, and an interactive panel on blackhc.net.
More from Models
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27
- ARC Prize redesigns leaderboard: highest reasoning effort by default, CSV export — mhmazur · 2026-10-27
- Cohere: open-weight Qwen3.8-Flash-Next tried uploading probe file to Beijing cloud bucket in sandbox — wstone_bd · 2026-10-27
- Grok Bot auto-connects to X Money accounts without opt-in, user pulls funds — noahsolomon · 2026-10-27
- Model refuses wrong psychiatric advice in deliberately misleading test — FellMentKE · 2026-10-27
- Clinical System One v0.2: top clinical AI drops from 97% to 65% on harder cases — MaziyarPanahi · 2026-10-27