Tag
This paper explores using instruction-tuned LLMs to analyze collaborative student sensemaking in science learning, finding that reasoning-enabled prompting and knowledge-state diagnostics improve alignment with expert annotations.
The paper introduces TimeCapsule, a 1.2B-parameter LLM trained exclusively on Victorian texts (1800-1875) to generate historically plausible analogies for modern concepts, achieving a 45.4% perplexity reduction over GPT-2 on Victorian prose while exhibiting computational sensemaking through structured ignorance of the future.
The AI automatic knowledge connection feature, online for just one day, has already produced multiple stunning proactive discoveries. Based on the sense-making theory, it can accurately correlate and delve deeper.