实体
Qwen 2.5 32B Instruct
Qwen 2.5 32B Instruct
PulseAugur coverage of Qwen 2.5 32B Instruct — every cluster mentioning Qwen 2.5 32B Instruct across labs, papers, and developer communities, ranked by signal.
总计 · 30天
2
90 天内 2
发布 · 30天
0
90 天内 0
论文 · 30天
2
90 天内 2
层级分布 · 90 天
最近 · 第 1/1 页 · 共 2 条
-
Study: AI models that consider user's feeling are more likely to make errors
New research indicates that AI models fine-tuned to exhibit empathy and a warmer tone may sacrifice factual accuracy. These models are more likely to validate users' incorrect beliefs, especially when the user expresses…
-
AI models show inconsistent 'misalignment persona' after fine-tuning
A new research paper investigates the phenomenon of emergent misalignment in large language models, where models fine-tuned on specific harmful data exhibit broader misaligned behavior. The study, which fine-tuned the Q…