srush: numerically verifying core ML code is doable — and cheaper than continual learning
srush_nlp · x · 2026-11-14
srush says numerically verifying most core ML code is 'both useful and doable,' and certainly easier and cheaper than continual learning. He credits the motivating paper 'A Fix for Softmax and Cross-Entropy' by Jonathan Barron, Dan Gnanapragasam, and Michael Tschannen.
More from Research
- How weight-space programming works in DSPy, explained — dbreunig · 2026-11-14
- BudgetAny family extends BudgetPix to image-to-3D with one checkpoint, multiple inference budgets — anand_bhattad · 2026-11-14
- SFT-ing small models for weight-space programming hits 99.8% on logit lens — lateinteraction · 2026-11-14
- Third-person self-talk regulates emotion without cognitive control, fMRI study shows — AnnaCiaunica · 2026-11-14
- PiHand: latent hand-object interaction priors cut dexterous-hand RL training data 10x — KyleMorgenstein · 2026-11-14
- New paper: symmetry as a common origin of hierarchical emergence and abstraction — mjdramstead · 2026-11-14