The framework scales gracefully: MAE degrades by less than 2% from 10K to 100K EdNet interactions with near-linear processing time¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? On sampled EdNet subsets of increasing size (10K, 50K, 100K interactions), MAE degrades by less than 2% and processing time scales nearly linearly with student count. → Yagna Manasa Boyapati 2026

Evidence¶

Yagna Manasa Boyapati 2026¶

Yagna Manasa Boyapati, Chong Yu, Tianyu Jiang, Justin Zhan. (2026). Privacy-Preserving Heterogeneous Multi-LLM Federated Inference for Cognitive Diagnosis. arXiv preprint. https://arxiv.org/abs/2609.02947 doi:10.48550/arxiv.2609.02947

q2 · i? · design · r2

Scalability analysis on EdNet (over 1.3 million student interactions, 188 knowledge topics), sampling three subsets of increasing size to allay concerns about small-scale evaluation of the primary benchmarks.

"MAE degrades by less than 2% from the 10K to 100K setting, and processing time scales nearly linearly with student count, confirming practical deployability at scale beyond our primary benchmarks."

Discussion¶

Related Claims¶

• Heterogeneous multi-LLM federated inference improves cognitive diagnosis MAE over single-LLM baselines across three educational benchmarks with minimal privacy cost — related