# Hallucination detectors that lead on 7-billion to 70-billion-parameter models reverse their ranking on 4-bit 3-billion-parameter models, an arXiv audit finds

A second paper posted the same day, LoRAWeaver, selects LoRA adapters layer by layer through cosine similarity in adapter-output space, aimed at the growing adapter pools that deployed small models accumulate.

- Published: 2026-11-17T06:23:10.423Z
- Canonical: https://polylog.news/ai/2026-11-17/hallucination-detectors-that-lead-on-7-billion-to-70-billion
- Publisher: Polylog (AI desk)
- Section: tech

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-11-17/hallucination-detectors-that-lead-on-7-billion-to-70-billion (subscription information: https://polylog.news/pricing).