About
Research & Development
Keeping Diffusion Models on Track: Introducing Temporal Alignment Guidance
TLDR: Temporal Alignment Guidance (TAG) is a novel method designed to improve the quality and fidelity of samples generated by diffusion models. It addresses...
Research & Development
HoMer: A Unified Approach to Enhance Click-Through Rate Prediction in Recommender Systems
TLDR: HoMer is a novel Transformer-based model that improves click-through rate (CTR) prediction by addressing three key challenges: feature, context, and architecture heterogeneities. It...
Research & Development
Shaping AI Behavior Through Information Geometry
TLDR: ENIGMA is a new method for training large language models (LLMs) that improves their reasoning, alignment with principles, and robustness by treating these...
Research & Development
The Inner Workings of AI Emotion: Discovering and Modulating Circuits in Large Language Models
TLDR: Researchers have systematically identified and validated "emotion circuits" within Large Language Models (LLMs) that are responsible for generating emotional text. By constructing a...
Research & Development
Closing the Modality Gap for Audio-Language AI
TLDR: Diffusion-Link is a novel diffusion-based module that effectively bridges the gap between audio and text embeddings. By generatively mapping audio embeddings to align...
Research & Development
EVP: Enhancing Dynamic Graph Learning with Event-Aware Prompting
TLDR: The research paper introduces EVP, an Event-aware Prompt learning framework for dynamic graphs. EVP addresses the limitation of existing methods by explicitly leveraging...