southeast-asia

Introduces OSCD, a post-training algorithm to improve native multilingual chain-of-thought reasoning in low-resource Southeast Asian languages, achieving up to 3.2x improvements on math benchmarks.

Satellite images reveal that at least 25 suspected scam compounds have appeared in Myanmar's Myawaddy region since early 2024, indicating that last year's crackdowns did little to halt the expanding criminal operations, which involve human trafficking and global online scams.

The 5th WeChat Mini Program Global Innovation Challenge Southeast Asia Regional Competition launched in Singapore, featuring the first AI Agent themed competition, attracting over 70 primary and secondary school students, showcasing the trend of accessible AI development tools.

Introduces SEATauBench, the first agent-focused evaluation framework for Southeast Asian languages, adapting τ²-Bench to Mandarin, Vietnamese, Thai, Indonesian, and Filipino, and reveals a significant capability gap when moving from English to localized settings.

Introduces SEA-NLI, a culturally grounded NLI benchmark covering eight Southeast Asian countries, revealing low performance of LLMs on culturally specific knowledge, especially in languages and science/technology. Shows that culture-aware prompting helps but chain-of-thought offers limited gains.

SEA-Embedding presents a fully open and reproducible text embedding pipeline for Southeast Asian languages, trained solely on public data, achieving state-of-the-art results on the SEA-BED benchmark.

An article exploring the history and cultural significance of fish sauce, a fundamental condiment in Southeast Asian cuisine made from fermented fish and salt, tracing its origins from ancient times to modern-day usage across Vietnam, Thailand, and other regions.

This paper introduces Anthropogenic Regional Adaptation, a paradigm for optimizing vision-language models to specific regional contexts while maintaining global generalization. The authors propose GG-EZ, an adaptation method using regional data filtering and model merging, demonstrating 5-15% improvements in cultural relevance for Southeast Asia across three VL architectures.

Grab uses OpenAI's GPT-4o vision fine-tuning to improve GrabMaps, achieving significant accuracy gains in speed limit sign localization (13%), lane counting (20%), and reducing manual mapping efforts across Southeast Asia's complex road networks.