Researcher: AlphaGenome distillation barely differs — variant correlations identical across models
jeffvierstra · x · 2026-09-19
Researcher jeffvierstra questions Google's AlphaGenome distillation, arguing that reporting 9 billion predictions isn't the goal — 'infinite GPUs/TPUs doesn't mean the predictions amount to much.' On closer inspection, signed Pearson correlations for variants are basically identical across all the models, suggesting the distillation mostly rescales noise; pre-distilled models aren't accessible to verify.
Related event: Benchmarks Question Distilled AlphaGenome's Performance(3 posts)→
More from Research
- Stanford's SALVE decodes hidden behaviors smuggled in distillation data — burny_tech · 2026-09-20
- New paper detects overfitting directly from weight matrices without any data — burny_tech · 2026-09-20
- Harvard Releases Full Lecture Notes for Its Dynamical Systems Course — burny_tech · 2026-09-20
- Databricks Co-founder: If Starting a PhD Today, I'd Work on Reward Hacking — burny_tech · 2026-09-20
- Gaussian Processes Explained: Bayesian Nonparametric Regression With Built-in Uncertainty — burny_tech · 2026-09-20
- ML paper explosion outpaces reviewer pool, peer review is getting 'vibey' — burny_tech · 2026-09-20