SignalTech/AI
YANAMI: A Large-Scale Multimodal Dataset and Benchmark for Multiturn Stateful Interaction
High-priority alert based on source significance and keyword relevance.
Context
What happened: Researchers introduced YANAMI, a large-scale multimodal dialogue dataset and benchmark for stateful interaction, where models must track and realize evolving dialogue states across turns.
Key numbers: YANAMI-400K contains 400K dialogues, more than 2M images, and 2.6M dialogue turns.
Why it matters: Experiments show existing unified multimodal models still struggle to maintain, update, and visually realize evolving dialogue states, and the benchmark measures the gap between understanding states and realizing them in visual outputs.
What changed
Researchers introduced YANAMI, a large-scale multimodal dialogue dataset and benchmark for stateful interaction, where models must track and realize evolving dialogue states across turns.
Why it matters
Experiments show existing unified multimodal models still struggle to maintain, update, and visually realize evolving dialogue states, and the benchmark measures the gap between understanding states and realizing them in visual outputs.