Heng Ji

A competence gate is introduced that estimates domain-level source weights from resolved outcomes, shrinks uncertain estimates toward a global weight, and recalibrates the pooled forecast and provides a practical approach for selective model use based on measured marginal value.

The Feature Recovery Module (FRM), a plug-and-play module that maps degraded encoder features to pristine-aligned representations while keeping the host frozen, is proposed, which improves scene-level detection, CLIP/SigLIP2 feature recovery, and all four object-level VLM tasks, with larger gains under more severe degradation.

While long-tail knowledge is notoriously hard for LLMs to retain, structurally popular facts suffer the worst collateral damage during knowledge updates and act as super-spreaders of downstream hallucinations.

Complex scoring algorithms for KV cache compression are largely unnecessary: protecting the initial prompt and dropping reasoning tokens completely at random matches state-of-the-art accuracy with up to 43% higher serving throughput.

Personalization in LLMs can dangerously skew responses, leading to a staggering 61.7% increase in sycophantic bias.

Apodex Discovery redefines AI evaluation by enabling verifiable investigations that surpass traditional benchmarks, achieving a 7% improvement in AAV capsid design outcomes.

Current visual world models show a dramatic decline in performance when faced with unconventional and impossible physical interactions, highlighting a critical gap in their generalization capabilities.

Agents struggle to maintain planning accuracy in complex tool ecosystems, with GPT-5.4's performance plummeting from 51.90% to 11.36% under severe blocking conditions.

Agents can boost their task completion rates by over 20% simply by grounding their actions in observed context rather than assumptions.

Relying on causal relationships rather than strong inductive biases, TDV achieves state-of-the-art performance in visual representation learning, challenging the status quo of self-supervised methods.

LLMs struggle with adaptive planning, achieving only 67.75% accuracy when faced with progressively revealed world and user constraints.

LMMs can't MacGyver their way out of a paper bag: they struggle to creatively repurpose objects in visually complex environments, revealing a critical gap in grounded reasoning beyond pattern recognition.