4arXiv cs.CL (Computation and Language)·
Ensemble Diversity Optimization (EDO) improves calibration on subjective NLP tasks
Researchers introduce Ensemble Diversity Optimization (EDO), a differentiable framework that jointly optimizes ensemble weights, effective cardinality, and calibration for subjective NLP tasks where annotator disagreement is systematic. EDO uses Gumbel-Softmax relaxation and a signed diversity regularizer to prevent ensemble collapse and navigate the utility-calibration trade-off. Experiments on four subjective text-classification benchmarks show 40-78% reductions in cross-entropy versus baselines while maintaining competitive F1 and better alignment with annotator distributions.
Related guides (2)
Related events (8)
Calibrated Mixture-of-Experts under distribution shift: adversarial reweighting approach
A new arXiv preprint analyzes how mixture-of-experts (MoE) models maintain calibration under distribution shift, examining the interaction between routing mechanisms and expert-level calibration. The authors prove that expert calibration is sufficient for overall model calibration in hard-routed MoE but insufficient for soft-routed variants. To address the soft-routing gap, they propose an adversarial reweighting method that penalizes calibration errors of the routed aggregate under distribution shift, demonstrating improved accuracy-calibration tradeoffs across model classes and tasks.
AGDO: Attention-guided denoising and optimization framework improves diffusion language model reasoning
Researchers propose AGDO, a framework that replaces random masking in diffusion large language models (dLLMs) with attention-guided denoising order and token weighting during fine-tuning and reinforcement learning. The work is motivated by an empirical finding that tokens with stronger attention to unmasked context are more stable and critical for reasoning. Experiments on math and coding benchmarks show AGDO outperforms existing post-training methods for dLLMs, advancing the case for attention-aware training in parallel-decoding language models.
Ensembling Tabular Foundation Models: A Diversity Ceiling and a Calibration Trap
This paper benchmarks six ensemble strategies across six tabular foundation models (TFMs) on 153 OpenML classification tasks, finding that ensembling provides minimal gains over the best single TFM. The best ensemble strategy (two-level cascade stacking) achieves only +0.18% accuracy improvement at 253× the compute cost. A key finding is that logistic-regression meta-learner stacking improves accuracy while severely degrading calibration (log-loss), because sharpening class boundaries destroys probability estimates. The authors recommend greedy ensemble selection as the practical default.
GoBOED: Goal-Driven Bayesian Optimal Experimental Design for Decision-Focused Robustness
GoBOED is a new framework for Bayesian optimal experimental design (BOED) that replaces information-gain maximization with direct optimization for a specified downstream decision objective. It combines an amortized variational posterior surrogate with a differentiable convex decision layer to enable gradient-based, decision-focused design optimization. The authors prove that GoBOED gradients are insensitive to parameter directions irrelevant to the decision goal, formally justifying why goal-driven design achieves equivalent decision quality over a wider range of experimental designs. Empirical results across source localization, epidemic management, and pharmacokinetic control show improved alignment with decision objectives compared to goal-agnostic BOED.
GGRO: Gradient-Guided Reward Optimization for inference-time LLM alignment
Researchers introduce Gradient-Guided Reward Optimization (GGRO), an inference-time alignment method that uses gradient signals from a reward model to inject 'nudging tokens' at high-uncertainty decoding steps, rather than relying on sampling-intensive re-ranking approaches like Best-of-N. The method monitors token-level entropy to detect distribution drift and steers generation trajectories directly, claiming improved robustness to reward hacking with minimal computational overhead. Experiments show gains across safety, helpfulness, and reasoning benchmarks compared to standard inference-time alignment baselines.
Optimal deterministic multicalibration achieved, resolving open problem on randomization necessity
A new arXiv preprint resolves an open problem in multicalibration theory by constructing a minimax-optimal multicalibration algorithm that outputs a deterministic predictor, achieving the same O(ε⁻³) sample complexity previously only attainable by randomized predictors. The result extends to outcome indistinguishability, deterministic omnipredictors, and panpredictors with optimal sample complexity, resolving multiple open problems from recent works. Multicalibration is a fairness and reliability property requiring calibration to hold across reweighted subgroups, making this relevant to trustworthy ML research.
DIVE: Dynamic In-context Vector Distillation with Decisive-Token Supervision for Long-form Medical Report Generation
DIVE is a frozen-backbone distillation framework that addresses a fundamental limitation in token-level in-context vector distillation: uniform cross-entropy supervision treats all output tokens equally, but long-form outputs like medical reports are dominated by low-information template tokens while diagnostically critical tokens receive insufficient gradient signal. The method introduces decisive-token supervision (upweighting pathology-related tokens and EOS events) and state-conditioned dynamic steering (hidden-state-dependent adapters replacing fixed residuals) to correct supervision imbalance and autoregressive drift. Evaluated on MIMIC-CXR and CheXpert Plus with two medical VLM backbones, DIVE achieves best BLEU-4, ROUGE-L, and RadGraph F1 across all dataset-backbone combinations while remaining competitive on CheXbert F1.
Reward uncertainty as a principled mechanism for diverse RL behaviour
A new arXiv preprint proposes replacing the scalar reward in RL with a distribution over reward functions, applying a non-linear objective over sets of actions to induce calibrated behavioural diversity without sacrificing expected reward. The authors derive a principled gradient estimator in the contextual bandit setting and prove the formulation generalizes vanilla policy gradient and action-set approaches. The work is motivated by applications like language model fine-tuning where diversity is desirable but entropy regularization and diversity bonuses introduce fragile trade-offs. Empirical results support the framework as a theoretically grounded alternative to heuristic diversity methods.