Tag
Terence Tao and Mark Chen discuss how AI is changing mathematical research, from literature search to code generation, and the need to adapt workflows.
A user shares their firsthand experience and impressions from talking to Opus 4.8, an AI language model.
The paper introduces SeDT, a training-free inference-time method that improves LLM reliability in multi-turn conversations by annotating conversation history with cumulative relevance scores from three signals, achieving up to +37.7% performance gains on the Lost-in-Conversation benchmark.
This paper introduces Found in Conversation (FiC), a training framework using View-Asymmetric Self-Distillation to close the multi-turn performance gap in LLMs. The method teaches models to recover single-turn competence from underspecified multi-turn prompts, achieving 92-100% recovery across model families and sizes.
This paper introduces the eJSL Dialog dataset for emotion recognition in sign language conversations, addressing the lack of conversational context in existing datasets. Benchmarking shows a domain gap when applying generic multimodal models, highlighting the need for context-aware visual extractors for sign language.