Concentration bounds on response-based vector embeddings of black-box generative models
Aranyak Acharyya, Joshua Agterberg, Youngser Park, Carey E. Priebe · 2025
Read Concentration bounds on response-based vector embeddings of black-box generative models by Aranyak Acharyya, Joshua Agterberg, Youngser Park, Carey E. Priebe (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-08307.txt.
Description
Aranyak Acharyya, Joshua Agterberg, Youngser Park, and Carey Priebe prove closeness bounds for response based vectors of black box maker models. Vector each model as a vector, built from its answers to shared queries, lets analysts compare systems like language and image generators. The Data Kernel View Space method gives one such vector, but users need to know how many sample responses make it trustworthy.
Under stated regularity assumptions, the sample vector stays close to its population version with high probability. The bounds state exactly how many responses a desired accuracy requires. The algebraic tools also extend to classic multidimensional scaling vectors when dissimilarities are observed with noise.
Catalog record
| CREATOR | Aranyak Acharyya, Joshua Agterberg, Youngser Park, Carey E. Priebe |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-08307