Goodfire's New Explainer: When SAEs Help, When to Use Probes Instead
mathildepapillo · x · 2026-09-29
GoodfireAI published part 2 of its applied interpretability educational series, addressing the recurring questions: are SAEs dead, can they save us from neuralese, or should you just use a probe? The thread explains what sparse autoencoders are actually good for, when not to use them, and which alternatives to reach for instead — a practitioner-oriented methodology piece amid ongoing debate over SAE usefulness.
More from Research
- AWS's Marc Brooker built the best 2B decision model — for about 24 hours — ShadajL · 2026-09-29
- JevBench Scales to 6x Test Cases, Rotates Held-Out Sets and Penalizes Benchmaxxing — airesearch12 · 2026-09-29
- Watch a Solo Dev Post-Train an 80B Model at Home on V100s: 96 Hours of Distillation, 3340 Samples — jjusko20 · 2026-09-29
- Sperm whales actively exchange vowels in dialogues, suggesting compositional codas — begusgasper · 2026-09-29
- NVIDIA ICRA'26 keynote: human data is the most scalable source for robot foundation models — yukez · 2026-09-29
- Prime Intellect brings multi-agent training to its open RL stack PRIME-RL — willcb · 2026-09-29