New LoRA-Based Method Steers Language Models Toward Specific Human Values
Researchers have developed ValuePEFT, a parameter-efficient fine-tuning method that steers large language models toward explicitly specified human values with measurable ...
Researchers have developed ValuePEFT, a parameter-efficient fine-tuning method that steers large language models toward explicitly specified human values with measurable ...
A new study finds that the locally deployable Gemma 3 27B model scores highly on a German introductory physics benchmark ...
A classroom study of Vietnamese EFL learners found that five minutes of AI-generated priming material—whether read, heard, or watched—significantly outperformed ...
Researchers have developed stMCP, a Model Context Protocol server that lets scientists analyze spatial transcriptomics data through natural language while ...
A new review argues that transmitting meaning instead of raw bits could keep future wireless networks reliable even in extreme ...
Researchers at Tsinghua University have developed a Series Saliency module guided by large language models that improves the accuracy and ...
A new special issue of AI & Society, framed by N. Katherine Hayles, maps the mutual entanglement of humans and ...
A University of Waterloo study of 11 large language models found that chatbots systematically favour existing options over change in ...
A new perspective in Discover Artificial Intelligence argues that prompt engineering is merely a subproblem within context engineering, a system-level ...
A new AI & Society paper argues that for people excluded from adequate human care, AI caregiving offers genuine advantages ...
© 2026 Scienmag - Science Magazine
© 2026 Scienmag - Science Magazine