New preprint: grammaticality is an organizing principle of NLM representations
najoungkim · x · 2026-09-18
Researchers lforlasagna and Najoung Kim released a new preprint asking whether grammaticality is a major organizing principle of neural language model (NLM) representations.
Key finding: many NLMs exhibit abstract representational separation for grammaticality. The authors argue this work addresses ongoing debates about confounds in measuring models' grammatical knowledge, offering new evidence for the linguistic competence of NLMs.
More from Research
- CAIS ships HLE-Rolling, a continuously updated fork of Humanity's Last Exam — devindkim · 2026-09-18
- RLE-Bench shows robots can grow their own bodies — but coding agents lack physical reasoning — daibond_alpha · 2026-09-18
- ENCODE GRAMMAR models switch to interactive HTML READMEs on the ENCODE portal — anshulkundaje · 2026-09-18
- World Modeling for Physics Workshop at Aspen Center, Feb 2027 — Abstracts Due Oct 9 — randall_balestr · 2026-09-18
- Cell Focus: Biology Needs World Models That Predict Interventions, Not Just Descriptions — rishabh16_ · 2026-09-18
- RLCD: RL Fine-Tuning Pushes LLMs Toward Calibrated Confidence in Decisions — prdeepakbabu · 2026-09-18