CodedLang: 7,744 Chinese reviews benchmark shows strong LLMs fail at coded language
windx0303 · x · 2026-10-24
ACL 2026 paper introduces CodedLang, a dataset of 7,744 real-world Chinese Google Maps reviews with span-level annotations (900 coded reviews) and a seven-class taxonomy of encoding strategies (phonetic, orthographic, cross-lingual). Benchmarks on detection, classification, and rating prediction show even strong models struggle to decode coded language; phonetic analysis included. Code and dataset are public.
More from Research
- Ropedia A0: code-only robot policy hits 70.24% on RoboDojo, 75% first-try sim2real — liuziwei7 · 2026-10-24
- Holi-World turns egocentric video into 3D/4D data: 4B model beats zero-shot 27B — liuziwei7 · 2026-10-24
- DecisionFormer open-sourced: fine-tune calibrated decision models on any LLM backbone — kalyan_kpl · 2026-10-24
- RLCR trains LLM agents as organizations: emergent hierarchies and division of labour — algo_diver · 2026-10-24
- Simplified Matching Distillation: few-step diffusion without adversarial training, validated up to 33B — ArashVahdat · 2026-10-24
- tszzl predicts stunning biology breakthroughs within years, cancer included — tszzl · 2026-10-24