Tsinghua releases SRHarness symbolic regression framework and MechBench mechanism benchmark
Overview
Tsinghua University's Data Science and Intelligence Lab releases SRHarness, an open-source runtime for agentic symbolic regression, and MechBench, a benchmark testing whether AI can recover the mechanisms behind a formula rather than only the formula itself.
Both are available on GitHub, installable from PyPI, and have an online demo.
The papers report that SRHarness reaches a 93.69% symbolic recovery rate on LSR-Transform, compared with 62.16% for SR-Scientist. On MechBench's core set, Codex with GPT-5.6-sol recovers mechanisms in 13.75% of cases, against 35% for laws.
Written by AI from the articles below · updated Oct 14, 12:38 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- PandailyNewsTsinghua releases SRHarness and MechBench to test whether AI agents explain formulas
AITsinghua University's Data Science and Intelligence Lab released SRHarness, an open-source runtime for agentic symbolic regression, and MechBench, a benchmark for recovering the mechanisms behind formulas. The papers report SRHarness reaching a 93.69% symbolic recovery rate on LSR-Transform versus 62.16% for SR-Scientist, while Codex with GPT-5.6-sol recovers mechanisms in only 13.75% of core-set cases against 35% for laws.
Heat trend
Not enough continuous observations to show a trend yet.