Yulong Ao

TrimTokenator: Towards Adaptive Visual Token Pruning for Large Multimodal Models

Hao Zhang, Mengsi Lyu, Chaoqun He, Yulong Ao et autres

Large Multimodal Models (LMMs) have achieved significant success across various tasks.These models usually encode visual inputs into dense token sequences, which are then concatenated with textual tokens and jointly processed by a language model.However, the increased token count substantially raises computational and …

cn (code pays fourni par la source)

PDTrim: Targeted Pruning for Prefill-Decode Disaggregation in Inference

Hao Zhang, Lyu Mengsi, Zhuo Chen, Yulong Ao et autres

Large Language Models (LLMs) demonstrate exceptional capabilities across various tasks, but their deployment is constrained by high computational and memory costs.Model pruning provides an effective means to alleviate these demands.However, existing methods often ignore the characteristics of prefill-decode (PD) disaggregation in practice.In …

cn (code pays fourni par la source)

RoboOS: A Hierarchical Embodied Framework for Cross-Embodiment and Multi-Agent Collaboration

Huajie Tan, Xiaoshuai Hao, Cheng Chi, Min Lin et autres

The dawn of embodied intelligence has ushered in an unprecedented imperative for resilient, cognition-enabled multi-agent collaboration across next-generation ecosystems, revolutionizing paradigms in autonomous manufacturing, adaptive service robotics, and cyber-physical production architectures. However, current robotic systems face significant limitations, such as limited cross-embodiment …

Infinity-MM: Scaling Multimodal Performance with Large-Scale and High-Quality Instruction Data

Shuhao Gu, Jialing Zhang, Siyuan Zhou, Kevin Yu et autres

Recently, Vision-Language Models (VLMs) have achieved remarkable progress in multimodal tasks, and multimodal instruction data serves as the foundation for enhancing VLM capabilities. Despite the availability of several open-source multimodal datasets, limitations in the scale and quality of open-source instruction data hinder …

Emu3: Next-Token Prediction is All You Need

Xinlong Wang, Xiaosong Zhang, Zhengxiong Luo, Quan Sun et autres

While next-token prediction is considered a promising path towards artificial general intelligence, it has struggled to excel in multimodal tasks, which are still dominated by diffusion models (e.g., Stable Diffusion) and compositional approaches (e.g., CLIP combined with LLMs). In this paper, we …

AquilaMoE: Efficient Training for MoE Models with Scale-Up and Scale-Out Strategies

Bowen Zhang, Liangdong Wang, Ye Yuan, Jijie Li et autres

In recent years, with the rapid application of large language models across various fields, the scale of these models has gradually increased, and the resources required for their pre-training have grown exponentially. Training an LLM from scratch will cost a lot of …

End-to-end Adaptive Distributed Training on PaddlePaddle

Yulong Ao, Zhihua Wu, Dianhai Yu, Weibao Gong et autres

Distributed training has become a pervasive and effective approach for training a large neural network (NN) model with processing massive data. However, it is very challenging to satisfy requirements from various NN models, diverse computing resources, and their dynamic changes during a …

AutoWM: a novel domain-specific tool for universal multi-/many-core accelerations of the WRF cloud microphysics

Peng Zhang, Chao Yang, Yulong Ao

cn (code pays fourni par la source)

Adaptive SpMV/SpMSpV on GPUs for Input Vectors of Varied Sparsity

Min Li, Yulong Ao, Chao Yang

Despite numerous efforts for optimizing the performance of Sparse Matrix and Vector Multiplication (SpMV) on modern hardware architectures, few works are done to its sparse counterpart, Sparse Matrix and Sparse Vector Multiplication (SpMSpV), not to mention dealing with input vectors of varied …

cn (code pays fourni par la source)

Adaptive SpMV/SpMSpV on GPUs for Input Vectors of Varied Sparsity

Min Li, Yulong Ao, Chao Yang

Despite numerous efforts for optimizing the performance of Sparse Matrix and Vector Multiplication (SpMV) on modern hardware architectures, few works are done to its sparse counterpart, Sparse Matrix and Sparse Vector Multiplication (SpMSpV), not to mention dealing with input vectors of varied …

cn (code pays fourni par la source)

Solving a trillion unknowns per second with HPGMG on Sunway TaihuLight

Wenjing Ma, Yulong Ao, Chao Yang, Samuel Williams

cn, us (code pays fourni par la source)

Enabling Highly Efficient k-Means Computations on the SW26010 Many-Core Processor of Sunway TaihuLight

Min Li, Chao Yang, Qiao Sun, Wenjing Ma et autres

cn (code pays fourni par la source)