Drift Adaptation as Supervision Routing Under Heterogeneous Costs
-
Updated
Jul 28, 2026 - Python
Drift Adaptation as Supervision Routing Under Heterogeneous Costs
Cost-aware multi-agent LLM framework with quality gates. Simple questions: 1 call. Code tasks: writer+reviewer loop.
PyTorch implementation of REACT: longitudinal active feature acquisition with Gumbel-Sigmoid relaxation for cost-aware sequential feature selection with context. ACM BCB 2026.
High-Volume & Cost-Aware Merchant Payment Platform
Instruction-driven model lanes for bounded Codex subagent delegation.
Adaptive, cost-aware retry policies for external requests: a Beta posterior decides per call whether retrying is worth it, not a fixed count. Zero-dependency, node-free, TypeScript.
A fail-closed job-search agent that screens roles against hard location and travel constraints before drafting application materials.
Are multi-agent LLMs worth the cost? Empirical study on Thai sarcasm detection: one cheap model reading its own confidence ties systems costing 11x more. Paired bootstrap + McNemar, 13 findings, live demo.
Add a description, image, and links to the cost-aware topic page so that developers can more easily learn about it.
To associate your repository with the cost-aware topic, visit your repo's landing page and select "manage topics."