framework
active
framework:in-context-learning-of-representations-park-et-al-2025In-Context Learning of Representations (Park et al. 2025)
Reports phase-like breakpoints and geometry changes as context scales; UCCT provides measurable predictor
Neighborhood — ranked by edge-count
Claims (1)
claim
- Positioning claim distinguishing UCCT's contribution from Park et al. 2024/2025
Artifacts (1)
artifact
- Main paper presenting UCCT and semantic anchoring framework.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Graph structure inferred purely in-context.
- The thesis that transformers develop a self-model via ICL, not only from training data; base models bootstrap self-referential reasoning.
- Test-time adaptation from prompt or retrieved context with no parameter updates.
- A follow-up paper extending the framework and induction head concept to larger more realistic models
- Extension of the thesis to deployed LLM inference via in-context learning
- Future direction.
- Framing that ICL reweights latent concepts by posterior.