New term 'Motion blur' describes LLM trajectory retention
nbaschez · x · 2026-08-15
Defines 'Motion blur' as the observed tendency of LLMs to include too much from their previous trajectory in their outputs, potentially leading to irrelevant disclaimers or context in the current response.
More from Research
- Qwen3.8-27B Local Test: Fact-Extraction F1 Ties Qwen3.6, Decode Speed Drops 16% — KitchenAmoeba4438 · 2026-08-15
- Information Abundance Paradox: long-context training can hurt short-task performance — DanielKhashabi · 2026-08-15
- Python 3.14 supports attaching pdb to running processes without instrumentation — StasBekman · 2026-08-15
- AI-assisted breakthrough on 70-year-old Grothendieck constant problem — Dr_Singularity · 2026-08-15
- Anthropic Research: Patterns and Problems in Multiagent Systems — rseroter · 2026-08-15
- Low loss doesn't equal high success rate: robotics needs better metrics — mathildepapillo · 2026-08-15