Unverified80% confidenceFactTime unknown
Microsoft Phi系列证明了在小语言模型中数据质量比数量更重要
1
Sources
80%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
GitHub 8K Star：最全LLM大模型资源宝库深度解析
githubWangRongsheng
Related Entities
Related Claims
Unverified研究表明模型在低资源语言上的幻觉率显著高于高资源语言64% similarVerifiedThe output quality of large language models is highly dependent on the quality and relevance of the input context63% similarUnverified大语言模型本质上是概率性的，相同的输入可能产生不同的输出，这与传统软件的确定性行为截然不同63% similarVerifiedIn 2022, the Google Brain team first systematically documented the phenomenon of 'emergent abilities' in large language models in the paper 'Emergent Abilities of Large Language Models'62% similarUnverified自回归语言模型的OOD检测比传统机器学习更困难，因为语言组合空间庞大、语义OOD与表面形式OOD不对应，且对抗者可构造统计上看似in-distribution但语义触发OOD的输入61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/32640API
curl https://kongchang.com/api/v1/knowledge/claims/32640MCP
get_claim(id=32640)