Metaicl: Learning to learn in context
S Min, M Lewis, L Zettlemoyer, H Hajishirzi - ar** the pre-trained language model (PLM) frozen, has shown remarkable …
A Textual Dataset for Situated Proactive Response Selection
Recent data-driven conversational models are able to return fluent, consistent, and
informative responses to many kinds of requests and utterances in task-oriented scenarios …
informative responses to many kinds of requests and utterances in task-oriented scenarios …
PUB: A Pragmatics Understanding Benchmark for Assessing LLMs' Pragmatics Capabilities
LLMs have demonstrated remarkable capability for understanding semantics, but they often
struggle with understanding pragmatics. To demonstrate this fact, we release a Pragmatics …
struggle with understanding pragmatics. To demonstrate this fact, we release a Pragmatics …