# Apple — Toward a Theory of Evaluability

- Company: Apple (apple.com)
- Announced: 2026-11-10
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://machinelearning.apple.com/research/theory-evaluability
- Record: https://forck.live/items/36826-toward-a-theory-of-evaluability
- Subject: Machine Learning Research

Apple researchers have formalized a notion of evaluability for machine learning models that does not assume the test set was held out from training, addressing a gap that is increasingly hard to justify for large foundation models. The work asks whether black-box models can be evaluated when test samples may have been seen during training, and studies the problem under different levels of access to the model. In the setting where the evaluator only sees the model's output on the test set, the authors fully characterize the distributions where accurate evaluation is possible and give tight upper and lower bounds on the sample complexity of evaluation. Their results show that black-box evaluation in this setup is feasible if and only if the distribution is close to being small support. The paper then considers evaluation algorithms that can query the model on additional inputs, drawing connections to self-correctors studied in program testing. Using these, the authors show that a small amount of query access can increase the power of the evaluator for natural distributions and concept classes.

## Evidence

Verbatim from https://machinelearning.apple.com/research/theory-evaluability:

> Machine learning models, particularly large foundation models, are increasingly evaluated along multiple axes that are relevant for their usage. In addition to accuracy measures appropriate for the specific task of interest, measures capturing safety, calibration and fairness properties are often of interest. The classical approach to these problems assumes that the test set has been held out from the training process, an assumption that is increasingly harder to justify for large foundation models. In this work, we formalize a notion of evaluability without this assumption, and ask if it is feasible to evaluate black-box models in settings where our test samples may have been used during training. We study this problem under different settings, capturing different levels of access to the model. We first study the natural setting, where the evaluator only looks at the output of the model on the test set. We fully characterize the class of distributions where accurate evaluation is possible, and show tight upper and lower bounds on the sample complexity of evaluation.

---

Record: https://forck.live/items/36826-toward-a-theory-of-evaluability
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md
