claim
active
claim:temporal-feature-analysis-recovers-temporal-straightening-in-an-unsupervised-mannerTemporal Feature Analysis recovers temporal straightening in an unsupervised manner.
Interpretation of smooth predictive trajectories.
Source paper
extracted_from(2025) · Lubana, Ekdeep Singh · Rager, Can · Hindupur, Sai Sumedh R. · Costa, Valerie +12
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (4)
finding
- Table 4.
- Fig. 26.
- Fig. 5.
- Fig. 10.
Concepts (1)
concept
- Linearization of representational trajectories observed in brains and LMs.
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.
- Proposed objective decomposing activations into predictable and novel components.
- §7.2 assertion.
- Can latent codes from SAEs or Temporal Feature Analyzers capture the in-context graph phenomenon?question0.748§6.3 question.
- Motivates nested SAE.
- §2, discussion of precision.
- Limitation of single-prompt selection.
- Cited as evidence that hardware change can be faster than software change for injury recovery
- Cited as enabling precise behavioral control through SAE features, extending the same methodological line