Skip to content

Research

Page 19 of 88

WorldOlympiad — Can Your World Model Survive a Triathlon?

research note

WorldOlympiad — Can Your World Model Survive a Triathlon?

·8 min read·Yuke Zhao, Wangbo Zhao, Weijie Wang et al.

WorldOlympiad introduces a comprehensive benchmark to evaluate video-based world models across three critical dimensions — physical faithfulness, geometric consistency, and interaction fidelity

researchvideo-world-modelslong-video-generationphysical-consistency3d-geometric-consistency

Read note → Source paper ↗

BrainSurgery — Reproducible and Reliable Declarative Weight Manipulations for Model Editing and Upcycling

research note

BrainSurgery — Reproducible and Reliable Declarative Weight Manipulations for Model Editing and Upcycling

·8 min read·Gianluca Barmina, Annemette Broch Pirchert, Andrea Blasi Núñez et al.

The paper addresses the growing complexity and fragility in modifying large neural network checkpoints, which increasingly hampers research workflows such as model merging, LoRA factorization, prun…

researchmodel-editingcheckpoint-manipulationdeclarative-ml-opsmodel-upcycling

Read note → Source paper ↗

Difference-Aware Retrieval Policies for Imitation Learning

research note

Difference-Aware Retrieval Policies for Imitation Learning

·8 min read·Quinn Pfeifer, Ethan Pronovost, Paarth Shah et al.

This paper addresses a fundamental limitation of behavior cloning (BC) in imitation learning — poor generalization and instability when encountering out-of-distribution (OOD) states due to compoundi…

researchimitation-learningbehavior-cloningretrieval-augmented-modelssemi-parametric-learning

Read note → Source paper ↗

Disentanglement with Holographic Reduced Representations

research note

Disentanglement with Holographic Reduced Representations

·8 min read·Jhonny J. Velasquez Olivera, Christo K. Thomas, Walid Saad

This paper addresses the problem of unsupervised disentangled representation learning by proposing a novel approach that leverages holographic reduced representations (HRRs), a type of vector symbo…

researchdisentangled-representation-learningvector-symbolic-architecturesholographic-reduced-representationsunsupervised-learning

Read note → Source paper ↗

Inflationary interpretation of the gravitational-wave signal in the European Pulsar Timing Array DR2 with constraints

research note

Inflationary interpretation of the gravitational-wave signal in the European Pulsar Timing Array DR2 with constraints

·9 min read·Philippe Turgeon, Chiara Caprini, Anton Chudaykin et al.

Analyzing the second EPTA data release's new timing measurements for 25 millisecond pulsars over 10.3 years, the authors identify two regimes in reheating temperature — a low Trh (<1 GeV) regime wit…

researchgravitational-wavespulsar-timing-arrayinflationary-cosmologyreheating

Read note → Source paper ↗

Articles are CC BY 4.0 — feel free to quote with attribution