The paper establishes the first theoretical framework for analyzing the learnability of Test-Time Adaptation (TTA) under non-stationary data streams by introducing Recovery Complexity, which quantifie…

TIMEGATE introduces a resource-aware policy layer that manages continual ML adaptation by dynamically budgeting time and evaluation resources, achieving significant compute and energy savings without…

This paper evaluates the reliability of using item response theory (IRT) models for AI benchmarking, comparing four estimation tools under various simulation conditions.

SkillAdaptor is a novel, training-free framework that enables stable, step-level adaptation of external skills for LLM agents by precisely attributing failures to specific skills.

This paper introduces Test-Time Harness Evolution (TTHE), a method for optimizing LLM agent harnesses during evaluation using execution traces, without requiring gold labels or training.

The paper proposes Test-Time Training for Supervised Causal Learning (TTT-SCL), a novel framework that dynamically generates training data aligned with specific test instances to significantly improve…

The paper introduces AGENTCL, a rigorous evaluation framework that uses controlled task streams to accurately measure an agent's ability to accumulate and reuse knowledge across multiple tasks, thereb…

The paper introduces a comprehensive benchmark to test if physics foundation models learn generalizable dynamics, finding that their performance is highly conditional and not universally general.

The paper introduces 5WBENCH, a new benchmark for causal unlearning, and proposes MAAT, a novel three-phase framework that achieves high forgetting and high retention specifically on complex 'Why'-typ…

The paper introduces a sample-wise targeted adversarial attack that successfully misclassifies only specific, triggered inputs during test-time adaptation while maintaining the overall label distribut…

The paper introduces BenchTrace, a novel benchmark designed to rigorously evaluate the self-evolution and reflection capabilities of LLM agents, revealing that current models struggle with accurate fa…

This paper introduces Test-Time-Training Robot Policies (RoboTTT), a robot model and training recipe that scales visuomotor context to 8K timesteps, enabling new capabilities like one-shot imitation a…

The paper introduces and evaluates five parameter alignment strategies that significantly mitigate catastrophic forgetting when continually pretraining multilingual expert language models across multi…

ADWIN introduces an adaptive window framework for on-policy distillation (OPD) that efficiently manages the supervision horizon by training on short, teacher-anchored prefixes while using delayed full…

TRACE proposes a novel method to mitigate catastrophic forgetting in continual LLM fine-tuning by identifying and isolating a small, task-specific subset of essential parameters for each task.

This paper introduces QuasiMoTTo, a method for generating correlated but exact samples in parallel to improve sample efficiency in scaling inference compute and reinforcement learning.

The paper introduces a novel, training-free method to automatically generate fine-grained evaluation rubrics for LLM-as-a-Judge, and further proposes an iterative fine-tuning strategy that significant…

This paper proposes a context-aware AI framework for generating telecom test scripts that adapt to fine-grained changes in the system using a delta engine and a knowledge graph.

The paper introduces a novel, transferable learned attack (LT-MIA) that detects a universal 'signature of memorization' in language models, achieving high accuracy across diverse model architectures (…