Tsinghua's SRHarness and MechBench push AI from curve-fitting to mechanism discovery
jiqizhixin · x · 2026-10-22
Tsinghua's Data Science and Intelligence Lab presents SRHarness, a runtime framework for agentic symbolic regression with reusable scientific actions, persistent scientific state and trajectory lifecycle management, and MechBench, a benchmark testing whether AI can reveal the physical mechanism behind discovered formulas rather than just fit curves (ArXiv 2609.35501 / 2609.35515).
More from Research
- MIT's Juan Antolin-Diaz to present Attention-Based Vector Autoregressions at AMLEDS — TaniaBabina · 2026-10-22
- UW–Madison Computational Imaging Group launches site, recruiting PhDs for Fall 2027 — prof_kamilov · 2026-10-22
- Information Geometry a la carte: build your own dual geometry in three steps — FrnkNlsn · 2026-10-22
- Bayesian take: LLM benchmark noise means all models are probably tied — remilouf · 2026-10-22
- Carbon-X as a sub-agent lifts Kimi K3's DNA intelligence index by 11.1 points — LoubnaBenAllal1 · 2026-10-22
- RL alone stalls at chance: Carbon-X shows midtraining unlocks gains RL can't — LoubnaBenAllal1 · 2026-10-22