Hao Wu

Felix Abecassis, Anjulie Agrusa, Dong Ahn et al. · nvidia

Large Language Models (LLMs) today are powerful problem solvers across many domains, and they continue to get stronger as they scale in model size, training set size, and training set quality, as shown by extensive research and experimentation across the industry. Training a frontier model today requires on the order of tens to hundreds of yottaflops, which is a massive investment of time, compute, and energy. Improving pretraining efficiency is therefore essential to enable the next generation of even more capable LLMs. While 8-bit floating point (FP8) training is now widely adopted, transitioning to even narrower precision, such as 4-bit floating point (FP4), could unlock additional improvements in computational speed and resource utilization. However, quantization at this level poses challenges to training stability, convergence, and implementation, notably for large-scale models trained on long token horizons. In this study, we introduce a novel approach for stable and accurate training of large language models (LLMs) using the NVFP4 format. Our method integrates Random Hadamard transforms (RHT) to bound block-level outliers, employs a two-dimensional quantization scheme for consistent representations across both the forward and backward passes, utilizes stochastic rounding for unbiased gradient estimation, and incorporates selective high-precision layers. We validate our approach by training a 12-billion-parameter model on 10 trillion tokens -- the longest publicly documented training run in 4-bit precision to date. Our results show that the model trained with our NVFP4-based pretraining technique achieves training loss and downstream task accuracies comparable to an FP8 baseline. These findings highlight that NVFP4, when combined with our training approach, represents a major step forward in narrow-precision LLM training algorithms.

Hongkang Yang, Zehao Lin, Wenjin Wang et al.

The training and inference of large language models (LLMs) are together a costly process that transports knowledge from raw data to meaningful computation. Inspired by the memory hierarchy of the human brain, we reduce this cost by equipping LLMs with explicit memory, a memory format cheaper than model parameters and text retrieval-augmented generation (RAG). Conceptually, with most of its knowledge externalized to explicit memories, the LLM can enjoy a smaller parameter size, training cost, and inference cost, all proportional to the amount of remaining "abstract knowledge". As a preliminary proof of concept, we train from scratch a 2.4B LLM, which achieves better performance than much larger LLMs as well as RAG models, and maintains higher decoding speed than RAG. The model is named $\text{Memory}^3$, since explicit memory is the third form of memory in LLMs after implicit memory (model parameters) and working memory (context key-values). We introduce a memory circuitry theory to support the externalization of knowledge, and present novel techniques including a memory sparsification mechanism that makes storage tractable and a two-stage pretraining scheme that facilitates memory formation.

Hao Wu, Yuxuan Liang, Wei Xiong et al.

Efficiently modeling spatio-temporal (ST) physical processes and observations presents a challenging problem for the deep learning community. Many recent studies have concentrated on meticulously reconciling various advantages, leading to designed models that are neither simple nor practical. To address this issue, this paper presents a systematic study on existing shortcomings faced by off-the-shelf models, including lack of local fidelity, poor prediction performance over long time-steps,low scalability, and inefficiency. To systematically address the aforementioned problems, we propose an EarthFarseer, a concise framework that combines parallel local convolutions and global Fourier-based transformer architectures, enabling dynamically capture the local-global spatial interactions and dependencies. EarthFarseer also incorporates a multi-scale fully convolutional and Fourier architectures to efficiently and effectively capture the temporal evolution. Our proposal demonstrates strong adaptability across various tasks and datasets, with fast convergence and better local fidelity in long time-steps predictions. Extensive experiments and visualizations over eight human society physical and natural physical datasets demonstrates the state-of-the-art performance of EarthFarseer. We release our code at https://github.com/easylearningscores/EarthFarseer.

Xiaoqing Zhang, Jilu Zhao, Yan Li et al.

Pathological myopia (PM) is the leading ocular disease for impaired vision worldwide. Clinically, the characteristic of pathology distribution in PM is global-local on the fundus image, which plays a significant role in assisting clinicians in diagnosing PM. However, most existing deep neural networks focused on designing complex architectures but rarely explored the pathology distribution prior of PM. To tackle this issue, we propose an efficient pyramid channel attention (EPCA) module, which fully leverages the potential of the clinical pathology prior of PM with pyramid pooling and multi-scale context fusion. Then, we construct EPCA-Net for automatic PM recognition based on fundus images by stacking a sequence of EPCA modules. Moreover, motivated by the recent pretraining-and-finetuning paradigm, we attempt to adapt pre-trained natural image models for PM recognition by freezing them and treating the EPCA and other attention modules as adapters. In addition, we construct a PM recognition benchmark termed PM-fundus by collecting fundus images of PM from publicly available datasets. The comprehensive experiments demonstrate the superiority of our EPCA-Net over state-of-the-art methods in the PM recognition task. The results also show that our method based on the pretraining-and-finetuning paradigm achieves competitive performance through comparisons to part of previous methods based on traditional fine-tuning paradigm with fewer tunable parameters, which has the potential to leverage more natural image foundation models to address the PM recognition task in limited medical data regime.

Yuanjie Lyu, Zhiyu Li, Simin Niu et al.

Retrieval-Augmented Generation (RAG) is a technique that enhances the capabilities of large language models (LLMs) by incorporating external knowledge sources. This method addresses common LLM limitations, including outdated information and the tendency to produce inaccurate "hallucinated" content. However, the evaluation of RAG systems is challenging, as existing benchmarks are limited in scope and diversity. Most of the current benchmarks predominantly assess question-answering applications, overlooking the broader spectrum of situations where RAG could prove advantageous. Moreover, they only evaluate the performance of the LLM component of the RAG pipeline in the experiments, and neglect the influence of the retrieval component and the external knowledge database. To address these issues, this paper constructs a large-scale and more comprehensive benchmark, and evaluates all the components of RAG systems in various RAG application scenarios. Specifically, we have categorized the range of RAG applications into four distinct types-Create, Read, Update, and Delete (CRUD), each representing a unique use case. "Create" refers to scenarios requiring the generation of original, varied content. "Read" involves responding to intricate questions in knowledge-intensive situations. "Update" focuses on revising and rectifying inaccuracies or inconsistencies in pre-existing texts. "Delete" pertains to the task of summarizing extensive texts into more concise forms. For each of these CRUD categories, we have developed comprehensive datasets to evaluate the performance of RAG systems. We also analyze the effects of various components of the RAG system, such as the retriever, the context length, the knowledge base construction, and the LLM. Finally, we provide useful insights for optimizing the RAG technology for different scenarios.