analiza aiqbench
This research presents a scalable system architecture that uses global calibration from high-fidelity LLMs to enable cost-effective prevalence measurement across thousands of concurrent A/B experiments. By amortizing labeling costs through continuous batch updates rather than per-experiment inference, the approach solves a critical infrastructure bottleneck for deploying reliable ML scoring at scale in production environments.
Calibrate Globally, Measure Everywhere: Scaling LLM-Based Prevalence Measurement Across A/B Experiments
arXiv:2602.16111v2 Announce Type: replace-cross Abstract: Online media platforms track the share of impressions associated with content attributes, or prevalence, to evaluate trade-offs and set guardrails in A/B experiments. LLM-based labeling provides a high-fidelity reference measurement, but is cost-prohibitive to run per experiment, per arm, per segment, and per day on a platform with hundreds of concurrent experiments. We describe a surrogate-based prevalence measurement system deployed in Pinterest's experimentation platform. The contribution is system-level rather than estimator-level: the system maintains a single global calibration of ML score buckets, continuously refreshed from a recurring LLM-labeled stream, and reuses the resulting bucket-level prevalences across every experiment via a per-experiment SQL metric and a delta-focused dashboard. Because the calibration is derived from the platform's daily-batch prevalence samples, it remains representative of production traffic as distributions drift, and in Pinterest's deployment it incurs zero incremental labeling cost. Teams without such infrastructure can instantiate the same pattern with a recurring calibration-labeling workflow whose cost is amortized across all downstream experiments rather than paid per experiment, arm, segment, and day. The system serves ~100 experiments and ~250 arms per day across six calibrated content categories, including a holdout program. Relative to per-experiment LLM labeling, which in practice yields a one-shot read per arm on a small subset of experiments, the surrogate provides daily per-arm prevalence on over 20$\times$ as many concurrent arms under the same labeling budget. Across roughly 300 production audits, the surrogate's 95\% confidence interval contains the LLM-based reference point estimate in 92\% of evaluations, and day-level delta aggregation recovers 2--5\% relative shifts that no single per-arm LLM measurement can detect.
Abstract:Online media platforms track the share of impressions associated with content attributes, or prevalence, to evaluate trade-offs and set guardrails in A/B experiments. LLM-based labeling provides a high-fidelity reference measurement, but is cost-prohibitive to run per experiment, per arm, per segment, and per day on a platform with hundreds of concurrent experiments. We describe a surrogate-based prevalence measurement system deployed in Pinterest's experimentation platform. The contribution is system-level rather than estimator-level: the system maintains a single global calibration of ML score buckets, continuously refreshed from a recurring LLM-labeled stream, and reuses the resulting bucket-level prevalences across every experiment via a per-experiment SQL metric and a delta-focused dashboard. Because the calibration is derived from the platform's daily-batch prevalence samples, it remains representative of production traffic as distributions drift, and in Pinterest's deployment it incurs zero incremental labeling cost. Teams without such infrastructure can instantiate the same pattern with a recurring calibration-labeling workflow whose cost is amortized across all downstream experiments rather than paid per experiment, arm, segment, and day. The system serves ~100 experiments and ~250 arms per day across six calibrated content categories, including a holdout program. Relative to per-experiment LLM labeling, which in practice yields a one-shot read per arm on a small subset of experiments, the surrogate provides daily per-arm prevalence on over 20$\times$ as many concurrent arms under the same labeling budget. Across roughly 300 production audits, the surrogate's 95\% confidence interval contains the LLM-based reference point estimate in 92\% of evaluations, and day-level delta aggregation recovers 2--5\% relative shifts that no single per-arm LLM measurement can detect.
Applications (stat.AP); Artificial Intelligence (cs.AI)
arXiv:2602.16111 [stat.AP]
(or arXiv:2602.16111v2 [stat.AP] for this version)
https://doi.org/10.48550/arXiv.2602.16111 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 355 cuvinte
Subiecte conexe
Does preservation make sense before we know how to revive?
LessWrong · 15 iun. 2026
Jev: System One models for Prod, not God — with Diogo Almeida, CEO, TypeSafe AI
Latent Space · 21 sept. 2026
34 Amazon Research Awards Build on Trainium recipients announced
Amazon Science · 5 aug. 2026
News
Quantum Computing Report · 25 iul. 2026