Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,657 papers · 148 categories

Trend · papers per month

305989118 · May 202619922001200920172026
48 results for Test-Time Alignment

QAlign improves language model alignment with less compute, outperforming existing methods.

problem Improving language model performance with limited test-time computation.
method QAlign: sampling from optimal aligned distribution using Markov chain Monte Carlo.
result Consistent improvements over existing methods on various benchmarks.

The abstract explores connections between reinforcement learning, scaling, and diffusion.

problem Aligning reinforcement learning with human feedback and scaling techniques.
method Clarifying connections between reinforcement learning, scaling, and diffusion.
result Introducing a resampling approach for alignment and reward-directed diffusion models.

The study evaluates how well local explanations align with model predictions.

problem Capturing the faithfulness of local explanations to model predictions.
method Introducing consistency and sufficiency as properties, and developing quantitative measures and estimators.
result Quantitative measures of consistency and sufficiency depend on test-time data distribution.

New study reveals task alignment is key to ICL performance.

problem Understanding how task alignment affects generalization in in-context learning.
method Derived an exact expression for ICL generalization error in high dimensions under task covariance mismatch.
result Identified train-test task alignment as a key determinant of generalization in ICL.

We consider neural network training, in applications in which there are many possible classes, but at test-time, the task is a binary classification task of determining whether the given example belongs to a specific class, where the class of interest can be different each time the classifier is applied. For instance, …

2017-05-29abs ↗pdf ↗

Test-time training adapts a pretrained model to each prompt via parameter updates, improving accuracy under pretraining-to-test distribution shifts.

problem Improving accuracy of pretrained models under distribution shifts.
method Explaining TTT behavior through a decision-theoretic lens.
result TTT reduces prediction error when updates are spectrally matched to the prompt's signal-to-noise ratio and aligned with query-relevant eigen-directions.

Framework controls uncertainty in LLMs without labels or probabilities.

problem Managing uncertainty in black-box LLMs without token-level probability or true labels.
method Integrates generative models, UCP, and conformal alignment to control uncertainty.
result Achieves close-to-nominal coverage and tighter thresholds than split UCP.

We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ensemble of models from…

2019-10-21abs ↗pdf ↗

Sequence-to-sequence models with soft attention have been successfully applied to a wide variety of problems, but their decoding process incurs a quadratic time and space cost and is inapplicable to real-time sequence transduction. To address these issues, we propose Monotonic Chunkwise Attention (MoChA), which adaptiv…

2017-12-14abs ↗pdf ↗

This paper improves test-time adaptation for distribution shifts using confidence maximization and input transformation.

problem Improving deep networks' performance on data shifted from the training distribution.
method Proposes a novel loss function combining confidence maximization and batch-wise entropy maximization with an input transformation module.
result Significantly improves robustness of pretrained networks to corruptions on benchmarks like ImageNet-C.

The study examines how extra compute during testing affects the performance of large language models.

problem Understanding the conditions under which test-time scaling improves model performance.
method An in-context weight prediction task for linear regression was used to train transformers. The performance was analyzed under varying levels of test-time compute.
result Training transformers on diverse, relevant, and hard tasks leads to the best performance for test-time scaling.

This work explores test-time scaling strategies for LLMs, improving sample efficiency and expressiveness.

problem Understanding the sample efficiency and expressiveness of test-time scaling strategies for LLMs.
method Established separation and expressiveness results for self-consistency, best-of-nn, and self-correction strategies.
result Self-correction enables Transformers to simulate online learning over multiple tasks without prior knowledge.

Implicit models can match or exceed explicit models with more test-time compute.

problem Understanding the expressive power and scaling of implicit models.
method Nonparametric analysis of expressive power, mathematical characterization of implicit operators, and test-time scaling experiments.
result Implicit models can progressively express more complex mappings through iteration, matching a richer function class with test-time compute.

New method optimizes language model performance for test-time strategies.

problem Mismatch between training objectives and test-time deployment of large language models.
method Tail-Extrapolated estimators to approximate best-of-N performance from limited training rollouts.
result Improved performance of best-of-N deployment across various models and datasets.

The study examines how verifier imperfections impact test-time scaling techniques.

problem Understanding how verifier imperfections affect test-time scaling methods.
method Proves the instance-level accuracy of Best-of-N and Rejection Sampling methods using the geometry of the verifier's ROC curve.
result RS outperforms BoN for fixed compute, but both converge to the same accuracy in the infinite-compute limit.

VCoTTA uses variational Bayesian methods to adapt models under continuous domain shifts.

problem Error accumulation in continual test-time adaptation.
method VCoTTA employs variational Bayesian techniques to update a Bayesian Neural Network (BNN) during testing, combining priors from source and teacher models.
result VCoTTA effectively mitigates error accumulation in CTTA, as shown by experimental results on three datasets.

CPATTA uses conformal prediction for efficient test-time adaptation.

problem Low data selection efficiency in existing ATTA methods.
method Conformal Prediction, online weight-update algorithm, domain-shift detector, staged update scheme.
result CPATTA consistently outperforms state-of-the-art methods by 5% in accuracy.

New method designs joint initial noises for diffusion models to improve diversity and alignment.

problem Independent initial noises limit diversity in generated images.
method Coupling of initial noises, maintaining Gaussian distribution while allowing dependence.
result Repulsive Gaussian coupling improves diversity without increasing sampling cost.

Machine learning models are often used at test-time subject to constraints and trade-offs not present at training-time. For example, a computer vision model operating on an embedded device may need to perform real-time inference, or a translation model operating on a cell phone may wish to bound its average compute tim…

2017-02-24abs ↗pdf ↗

Enhances image quality to improve test-time adaptation accuracy.

problem Reducing accuracy loss due to distribution shift in deep networks.
method Integrates image enhancement with TTA methods to reduce prediction uncertainty.
result TECA method increases accuracy of TTA methods without hyperparameters.

A new approach for test-time adaptation detects and reacts to distribution shifts.

problem Improving test-time accuracy under distribution shifts.
method Online self-training with a detection tool based on entropy values and betting martingales.
result The classifier's entropy values match those of the source domain, building invariance to distribution shifts.

DAD++ improves test-time adversarial defense without access to training data.

problem Improving adversarial robustness of pre-trained models without access to training data.
method Test-time Data-free Adversarial Defense (DAD++) with detection and correction schemes.
result DAD++ significantly improves adversarial robustness with minimal clean accuracy drop.

New scaling laws optimize model size, training, and inference for better performance.

problem Trade-off between model size and inference cost in modern LLMs.
method Train-to-Test (T2T^2) scaling laws that jointly optimize model size, training tokens, and inference samples.
result Optimal pretraining decisions shift into overtraining regime, leading to stronger performance.

SAGE enhances reinforcement learning by injecting hints to prevent model stagnation.

problem Sparse rewards cause large language models to stall under relative policy optimization.
method SAGE injects privileged hints during training to increase within-group outcome diversity.
result SAGE consistently outperforms GRPO on 6 benchmarks with LLMs, achieving significant improvements.

Adapts CNN for robust medical image segmentation across different scanners and protocols.

problem Performance degradation of CNNs in medical image segmentation due to mismatch between training and test images.
method Designs a segmentation CNN as a concatenation of a shallow normalization CNN and a deep CNN. At test time, adapts the normalization sub-network for each test image using a denoising autoencoder.
result Consistently improves performance on multi-center MRI datasets of brain, heart, and prostate.

Unified framework for sequence models using test-time regression.

problem Designing efficient sequence models with associative memory.
method Formalizing associative recall as regression over input tokens, deriving various sequence models.
result Clarifies the effectiveness of query-key normalization in softmax attention and offers new generalizations.

Method adapts frozen models for few-shot tasks without training.

problem Deployment constraints limit model updates, necessitating new adaptation methods.
method Exponential tilting of latent distribution for inference.
result Method outperforms parameter-update methods across benchmarks.

TTLSA adapts models to label shifts across domains with nuisance factors.

problem Adapting models to changes in label distributions with nuisance factors.
method TTLSA uses EM on unlabeled samples to adapt a trained model to new label distributions.
result TTLSA improves model performance over invariance methods and baseline methods.

Duel-Evolve uses LLM self-preferences for test-time optimization of discrete outputs.

problem Optimizing LLM outputs at test time with limited or unreliable scalar rewards.
method Duel-Evolve uses pairwise comparisons from the LLM to guide optimization, aggregating them via a Bayesian Bradley-Terry model.
result Achieves significant improvement over existing methods in accuracy.

Adaptive compute allocation improves model performance by prioritizing harder queries.

problem Inefficiency in allocating test-time compute uniformly across all queries.
method Formulated as a bandit learning problem, proposed adaptive algorithms that estimate query difficulty and allocate compute accordingly.
result Achieved up to 15.29% relative performance improvement on various benchmarks.

PG-EVIKAL refines molecular property predictions using neighbor fusion and evidential neural networks.

problem Improving molecular property predictions using test-time neighbor fusion.
method Adapting evidential neural networks to refine predictions by re-ranking structurally similar neighbors.
result PG-EVIKAL reduces RMSE on 14 out of 16 molecular datasets, improving calibration and sequential refinement.

This paper tackles over-certainty in test-time adaptation models, proposing a solution to improve calibration.

problem Over-certainty in predictions caused by domain shifts, leading to misplaced trust.
method Introduces a certainty regularizer that dynamically adjusts pseudo-label confidence based on backbone entropy and logit norm.
result Achieves state-of-the-art performance in terms of Expected Calibration Error and Negative Log Likelihood, while maintaining accuracy.

Paper proposes an efficient method for calibrating spatio-temporal forecasts.

problem Real-world spatio-temporal forecasting challenges like signal anomalies and distributional shifts.
method Learning with Calibration (ST-TTC) for real-time bias correction.
result ST-TTC improves spatio-temporal forecasting accuracy with reduced computational cost.

Best-of-\infty improves LLM performance by efficiently allocating inference-time computation.

problem Achieving optimal performance in test-time LLM ensembling with infinite budget.
method Adaptive generation scheme and weighted ensembles of LLMs, formulated as mixed-integer linear program.
result Optimal ensemble weighting improves performance over individual models.