Cantonese-adapted language models show stronger predictive fit for human reading

A new study published on arXiv investigates whether language models specifically trained on Cantonese can better predict human reading patterns compared to models trained on Standard Chinese or general-purpose models. Researchers used eye-tracking data from Cantonese speakers and derived various linguistic measures from different language models. The findings suggest that models with more extensive Cantonese-specific training, like CantoneseLLM-7B, show stronger predictive alignment with human reading behavior, although the specific measures used can influence model rankings. AI

IMPACT This research could inform the development of more linguistically accurate AI models for under-resourced languages.

RANK_REASON The cluster contains an academic paper detailing a new evaluation of language models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• Cantonese
• CantoneseLLM-7B
• CKIP GPT-2 Tiny
• Hugging Face
• JED351
• Qwen2.5-7B
• Standard Chinese

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →