Unverified50% confidenceFactExact time
In the Yoruba case study, three models achieve high accuracy under English prompts, but performance drops substantially when switched to Yoruba.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/18/2026
First Seen
Valid until: 2/16/2027
Sources
Related Entities
Related Claims
UnverifiedIn open-ended generation tasks, the study observed occasional accuracy drops, which were more pronounced in non-English responses.69% similarUnverifiedAfter rewriting tool definitions according to the three-element principle and similar guidelines, the model's invocation accuracy improved from less than half to over 80%.67% similarUnverified实验数据显示,经过5-10轮迭代训练后,模型在低频词汇和长尾知识上的覆盖率下降超过40%,而输出文本在表面流畅性上几乎没有可察觉的退化66% similarUnverifiedThat experiment reached the adaptation criterion specified in the paper (three consecutive sessions passing) at session 14, with both the bold rate and list rate metrics showing a downward trend across the first 36 sessions.66% similarUnverifiedThe study compared on four models, two in-domain corpora, and three held-out corpora, and the results show that only ASC simultaneously reduces the 'authored sentence' error rate and increases supported sentences.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1182960API
curl https://kongchang.com/api/v1/knowledge/claims/1182960MCP
get_claim(id=1182960)