# STaR-DRO: Stateful Tsallis Reweighting for Group-Robust Structured Prediction

Source: [STaR-DRO: Stateful Tsallis Reweighting for Group-Robust Structured Prediction](<https://www.sota2.com/research/paper/69ddca077d0172140ec79963>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-16T05:00:14.254Z

Authors: Samah Fodeh, Ganesh Puthiaraju, Elyas Irankhah, Afshan Khan, Sreeraj Ramachandran, Linhai Ma, Srivani Talakokkul, Sarah Schellhorn

Published: 2026-04-09T21:58:42.000Z

## Abstract

Structured prediction with large language models requires outputs that are label-accurate, ontology-constrained, structurally valid, and evidence-grounded under label imbalance and heterogeneous group difficulty. We present a unified framework for ontology-constrained generation. First, we introduce a modular prompt-engineering architecture combining XML-style structure, expert disambiguation rules, chain-of-thought reasoning, metadata-aware decision logic, schema contracts, and a self-validation gate. It targets recurrent in-context failures, including format drift, label ambiguity, evidence hallucination, and metadata-conditioned confusion. Second, we propose STaR-DRO, combining Tsallis mirror ascent, sparse entmax-style primal mapback, EMA-smoothed group-loss tracking, rescaled ascent signals, and bounded excess-only multipliers. Unlike conventional DRO, which relies on dense Shannon-entropy exponentiated-gradient updates, can introduce high-variance stochastic reweighting, assigns positive adversarial mass to groups that are not persistently hard, and incurs costs through simplex competition, STaR-DRO upweights only persistently hard groups without suppressing easier ones. We evaluate the framework on EPPC Miner, a clinically grounded high-stakes structured-prediction task requiring hierarchical label prediction and evidence-span extraction from patient-provider secure messages. Across 1B-70B Llama models, prompt engineering improves zero-shot extraction, yielding an average label F1 gain of +14.46 and a Span F1 gain of +17.40. Building on supervised fine-tuning, STaR-DRO further improves accuracy and robustness, increasing average label F1 by +1.08 and +2.20 while reducing mean groupwise validation cross-entropy by 21.3% and 14.8% relative to SFT and standard DRO, respectively. These results advance reliable automated communication mining for patient-centered clinical care analysis.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Structured Prediction on EPPC Miner (test)](<https://www.sota2.com/research/sota/structured-prediction-on-eppc-miner-test.md>): Structured Prediction; EPPC Miner (test); Code Precision: 79.08.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.09737>)

[PDF](<https://arxiv.org/pdf/2604.09737.pdf>)
