GitHub

Discrete Diffusion Forcing (D2F): dLLMs Can Do Faster-Than-AR Inference

Real-time generation demo: our D2F model (left) uses parallel block decoding, while the AR baseline (right) generates tokens sequentially. This visualizes the source of D2F's significant throughput ...
GitHub

Outlier-Safe Pre-Training

Quantization plays a crucial role in deploying Large Language Models (LLMs) in resource-constrained environments. However, the presence of outlier features significantly hinders low-bit quantization.
IEEE

Pre-Trained Large Model as Distributed PV Power Forecaster: Integrating Scenario-Consistent Retrieval-Augmented Generation and Cross-Modal Semantic Alignment

Abstract: The rapid proliferation of distributed photovoltaic (PV) systems presents significant challenges for accurate power generation forecasting due to their inherent intermittency and ...
The New York Times

Can A.I. Match Molière’s Wit? These Researchers Think So.

Scholars and artists at Sorbonne University trained artificial intelligence to imitate the French playwright’s themes, structures and sense of humor. The result is a new play. By Laura Cappelle ...
IEEE

Steer Your Model: Secure Code Generation with Contrastive Decoding

Abstract: Large Language Models (LLMs) specialized in code have demonstrated impressive capabilities in various programming tasks such as code generation. However, these models often generate ...