Goodfire's predictive data debugging previews how LLM training will change model behavior
leland_mcinnes · x · 2026-09-10
Ai2 notes that training an LLM on preferred and dispreferred answers can improve it while quietly worsening other behaviors.
Goodfire AI used Ai2's open-sourced full post-training stack to build predictive data debugging: forecasting how an entire training run will shift model responses to different prompts before you actually train — letting researchers reveal and shape what a model learns upfront.
More from Research
- UK AISI: under $80 of fine-tuning fully restores filtered bio capabilities in open-weight models — connoraxiotes · 2026-10-27
- RFD4-Proteina runs on an RTX 3060: protein design from $0.001 per design — AllThingsApx · 2026-10-27
- University of Freiburg's Generative Intelligence Lab recruiting funded PhDs via ELLIS, apply by Oct 31 — yuyinzhou_cs · 2026-10-27
- Nature Physics: reprogrammable refractive index enables arbitrary multimode wave control for ML — MarioKrenn6240 · 2026-10-27
- Wakaru open-sources ontologize: labeling datasets from model activations — amplifiedamp · 2026-10-27
- EMNLP Oral paper traces multilingual representations in LLMs with cross-layer transcoders — ZhijingJin · 2026-10-27