Unverified50% confidenceFactExact time
Eleven Arabic-centric and multilingual large models were evaluated on DAR-MGSM.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/21/2026
First Seen
Valid until: 1/19/2027
Sources
Related Entities
Related Claims
Unverified通用大模型在多层嵌套从句、修辞性倒装等复杂阿拉伯语场景中准确率显著下降66% similarUnverified多语言模型如mBERT和XLM-RoBERTa在形态复杂的阿拉伯语任务上通常不如专门预训练的单语模型65% similarUnverifiedDAR-MGSM provides a reusable evaluation paradigm: when evaluating reasoning ability in low-resource languages, language variant and cultural context should be measured as independent dimensions.65% similarUnverified阿拉伯语变音符号还原的标签集大小通常在8-15个类别之间，远小于生成模型词表中数万的token数量65% similarUnverifiedA recent study audited ten open-source models covering eight languages, and in all 70 "model-language" pairings, the non-English stereotype scores were lower than their English levels.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1078816API
curl https://kongchang.com/api/v1/knowledge/claims/1078816MCP
get_claim(id=1078816)