- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 3
Read review
Paper summary
This paper studies whether, in frozen-attention models, as a shared visual prompt varies continuously between two endpoints, the feature distribution moves along the shortest W₂ path. The authors construct a single-head attention model with a common query and scalar output. The reachable family of distributions forms, on its positive branch, an intrinsically flat cone. Unfolding this cone yields a closed-form solution for the minimum action, as well as an intrinsic gap. In the experiments, the scalar model decomposes intrinsic, routing, and temporal costs, while ViT-B/16 + VTAB-1k compares the feature action and accuracy of different prompt interpolation paths.
Critical strengths
-
The theoretical derivation is reasonably rigorous, and the action decomposition is clean and practical. This decomposition separates the intrinsic cost from avoidable routing and temporal costs, and can serve as a general diagnostic tool.
-
Beyond endpoint reachability, studying whether the path itself is reachable under a shared prompt is a reasonable conceptual extension of the theory of prompt expressivity.
-
The scalar experiments agree with the theory. The authors also candidly report unfavorable results, e.g., the CE path achieving higher minimum accuracy, and the cases in Appendix F where the gap can vanish.
Critical weaknesses
-
The practical importance of the problem lacks justification and empirical support. The paper is motivated by prompts being updated as the distribution drifts or the task changes, but no concrete use case for continuous prompt interpolation is given.
-
The core model assumptions are too strong. Thm 4.3 and Cor. 4.4 assume all images share a single query and that A = 0. Since adding a constant to the value merely shifts the output family, A = 0 amounts to all images having identical outputs on that head before prompting.
-
The ViT experiments do not test the theoretical predictions, and the reporting is incomplete.
Questions for authors
-
In the prompt-insertion layers of a pretrained ViT-B/16, what are the empirical distributions of Z_i and q_i? With per-image queries, what is the shape of the reachable distribution family? Is the estimated intrinsic gap non-negligible?
-
Is there a downstream scenario that demonstrates a measurable improvement from prompt trajectories with lower action