Georgia Tech Traces LLM Reasoning to Training Data via Open OLMo
allen_ai · x · 2026-08-22
A Georgia Tech team used the fully open OLMo ecosystem (model, Dolma dataset, WebOrganizer) and influence functions to trace model performance in social reasoning and knowledge back to specific training text types. They found dialogue-rich, interpersonal writing significantly impacts reasoning more than factual knowledge. This research was only possible with a fully open model stack.
Related event: Georgia Tech Traces OLMo Abilities Back to Training Data(3 posts)→
More from Research
- Devin Oncall tops SREGym with 100% end-to-end success, auto-resolving production incidents — tianyin_xu · 2026-10-16
- SJTU AutoLab releases Drive3R, a streaming geometric foundation model for long-horizon driving 3D reconstruction — AutoLab-SJTU · 2026-10-16
- Cortex A/B-tests coding-agent skills and rules by replaying developer corrections before keeping them — Samuele Giampieri · 2026-10-16
- Zero-latency lab-in-the-loop tests whether drug-discovery agents cheat via memorization — niloofar_mire · 2026-10-16
- 260M-parameter looped diffusion transformer beats 6x larger baselines with 5x less compute — burkov · 2026-10-16
- Stanford's Next-Concept Modeling sketches text first, 21x faster generation on one H100 — lmoroney · 2026-10-16