Towards Understanding Extrapolation: a Causal Lens The paper addresses extrapolation, where models predict outcomes for target samples outside the training distribution. It introduces a latent-variable model to formalize when extrapolation is feasible and offers methods to
RESEARCH
-

Tarsier2: Advanced Video Understanding with Large Vision-Language Models
By
–
Tarsier2: Advancing Large Vision-Language Models from Detailed Description to Comprehensive Understanding Tarsier2 is a state-of-the-art large vision-language model (LVLM) that excels in video description and understanding, improved by scaling pre-training data,
-

Inference-Time-Compute Models: Evaluating Faithfulness in AI
By
–
Inference-Time-Compute: More Faithful? A Research Note This research evaluates the faithfulness of Inference-Time-Compute (ITC) models, comparing their ability to articulate cues influencing their answers with non-ITC models. Problem: Faithfulness in Chains of Thought (CoTs) is
-

Do Generative Video Models Learn Physical Principles
By
–
Do generative video models learn physical principles from watching videos? This paper explores whether generative video models learn physical principles by analyzing their performance on the Physics-IQ benchmark, which tests understanding of laws like fluid dynamics, optics, and
-

LLaVA-ST: Multimodal LLM for Spatial-Temporal Video Understanding
By
–
LLaVA-ST: A Multimodal Large Language Model for Fine-Grained Spatial-Temporal Understanding LLaVA-ST is a multimodal large language model designed for spatial-temporal understanding in videos, addressing challenges in coordinate alignment and feature compression. Problem:
-

Diffusion Model for Portrait Animation: Joint Depth and Appearance Learning
By
–
Joint Learning of Depth and Appearance for Portrait Image Animation This research presents a diffusion-based generative model for joint learning of depth and appearance in portrait image animation, enabling consistent visual and 3D outputs for various applications. Problem:
-

WebWalker: LLM Benchmarking Framework for Web Traversal
By
–
WebWalker: Benchmarking LLMs in Web Traversal The paper introduces WebWalkerQA, a benchmark for testing LLMs in web traversal, and WebWalker, a multi-agent framework for human-like navigation to extract layered information. Problem: RAG systems retrieve shallow content,
-

GPT-4o Advances Face Presentation Attack Detection Few-Shot Learning
By
–
Exploring ChatGPT for Face Presentation Attack Detection in Zero and Few-Shot in-Context Learning The paper investigates GPT-4o for Face Presentation Attack Detection (PAD), showing strong performance in few-shot in-context learning, outperforming specialized PAD models in
-

Weekly AI Breakthroughs: Detection, ChatGPT, and Web Navigation
By
–
From uncovering physical laws in videos to animating portraits and navigating the web, this week’s AI research has brought even more breakthroughs! – Exploring ChatGPT for Face Presentation Attack Detection in Zero and Few-Shot in-Context Learning
– WebWalker: Benchmarking -

MIT Introduces Boltz-1 Open-Source Biomolecular Structure Prediction
By
–
MIT researchers introduce Boltz-1, a fully open-source model for predicting biomolecular structures https://
bit.ly/3P4sTtZ
#AI #MachineLearning #DeepLearning #LLMs #DataScience
