About
Research & Development
NVIDIA and Boston University Unveil Nemotron-CC-Math: A New Era for AI’s Mathematical Prowess
TLDR: Researchers developed Nemotron-CC-Math, a 133-billion-token high-quality math pretraining dataset from Common Crawl. Their novel pipeline uses a text-based browser (Lynx) and an LLM...
Research & Development
Hydra: A Unified Approach to Long-Context Language Models
TLDR: Hydra is a proposed 1.6-billion-parameter language model architecture that combines a Mamba-style State-Space Model (SSM) backbone, sparse global attention, chunk-level Mixture-of-Experts (MoE), and...
Research & Development
SemToken: A Smarter Way to Process Text for AI’s Long Conversations
TLDR: SemToken is a new tokenization method for large language models that uses semantic understanding to reduce redundant tokens in long texts. Unlike traditional...
Research & Development
SPARK: Enhancing LLM Efficiency Through Dynamic KV Cache Channel Pruning
TLDR: SPARK is a novel, training-free method designed to address the KV cache memory bottleneck in large language models (LLMs) for long-context inference. It...
Research & Development
Building Robust Quantum AI with Time Crystal Dynamics
TLDR: This research introduces a novel quantum machine learning algorithm, Discrete Time Crystal Quantum Reservoir Computing (DTC-QRC), which utilizes the unique dynamics of discrete...
Research & Development
Understanding Why You Buy: M-LLM3REC’s Deep Dive into User Motivation for Smarter Recommendations
TLDR: M-LLM3REC is a novel recommendation framework that utilizes large language models to understand the underlying motivations of users from limited interactions. It addresses...