method
active
method:spatial-understanding-taskSpatial Understanding Task
Training paradigm requiring prediction of upcoming sensory observations during spatial navigation across multiple environments sharing the same structure.
Neighborhood — ranked by edge-count
Frameworks (1)
framework
- TEM-Transformer (TEM-t)implementsThe transformer version directly analogous to TEM, introduced in this paper, offering dramatic performance improvements.
Concepts (1)
concept
- Training environments formalized as graphs where nodes have sensory observations and edges represent actions; used to test structural generalization.
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.
- The ability to reason about shapes, space, and topology, essential for ancient mathematical discoveries and observed in many animals.
- ToM task requiring integration of intent and tone to understand sarcasm; scored 0/1.
- The translation of semantic values into spatial coordinates and relations.
- what is the analogue of spatial positional encodings for higher order tasks such as language?question0.728Open question raised in Discussion about extending TEM-t principles beyond spatial navigation.
- The ability to generalize across tasks; lacking in latent methods.
- The paper identifies task difficulty as a key moderator: easy MMLU questions show performative CoT, hard GPQA-Diamond questions show genuine reasoning
- Novel task asking which of two sentences received a stronger injection, using matched-pairs design to control for positional bias