claim
active
claim:the-workspace-may-be-verbalizable-because-the-model-s-output-space-and-input-space-are-verbalThe workspace may be verbalizable because the model's output space and input space are verbal.
Speculative explanation.
Source paper
extracted_fromAuthors
Neighborhood — ranked by edge-count
Papers (1)
paper
Hypotheses (1)
hypothesis
- Language models able to generate images could develop a visual component of their workspace.associated_withPrediction from verbal-output account.
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.
- Central thesis.
- Dissociation.
- Explicit hedge.
- Directed modulation.
- Gurnee et al.'s privileged set of verbalisable representations behaving as a within-pass global workspace.
- The model attends to the J-space on the user prompt when determining its introspective answer, so concepts there are reportable.hypothesis0.783Motivated by elephant observation.
- Directed modulation prediction.