4 papers · 1 filter
Same Answer, Different Future: Matched Predictions Do Not Imply Matched Belief Updating in Language Models
Weihan Li, Junhao Wu, Xinlei Chen +1
Language models can produce similar probabilistic predictions after observing outcomes, reading summaries, receiving instructions, or undergoing activation interventions. Whether t…
Erased, Rerouted, or Rescaled? Post-Training and the Causal Quotient of a Language Model's Belief State
Weihan Li, Tianshi Zheng, Junhao Wu +1
What happens to information a pretrained model already encodes when post-training no longer rewards using it? The common language of representation compression conflates three fate…
ULTRADISCOVERY: Abductive Exploration in an Interconnected, Epistemically Open Universe
Weihan Li, Tianshi Zheng, Yangqiu Song +2
Scientific discovery often begins when scattered clues call for a new way of describing the world. Such abductive exploration can require constructing the representation in which a…
Same Outcome, Different Readout: What Does a Steerable Valence Direction in LLMs Represent?
Weihan Li, Xinlei Chen, Yuhan Song +2
Claims about what an internal direction in an LLM represents need evidential constraints beyond an observer's prior beliefs about the system. Decodability and successful activation…