Related papers: Failure Tolerant Training with Persistent Memory D…

Large Language Model (LLM) inference uses an autoregressive manner to generate one token at a time, which exhibits notably lower operational intensity compared to earlier Machine Learning (ML) models such as encoder-only transformers and…

We introduce Prompt Curriculum Learning (PCL), a lightweight reinforcement learning (RL) algorithm that selects intermediate-difficulty prompts using a learned value model to post-train language models. Since post-training LLMs via RL…

High-Performance Computing (HPC) and Artificial Intelligence (AI) workloads typically demand substantial memory bandwidth and, to a degree, memory capacity. CXL memory expansion modules, also known as CXL "type-3" devices, enable…

Current HPC systems provide memory resources that are statically configured and tightly coupled with compute nodes. However, workloads on HPC systems are evolving. Diverse workloads lead to a need for configurable memory resources to…

Disaggregated LLM serving improves resource efficiency by separating the compute-intensive prefill phase from the latency-critical decode phase. However, this architecture introduces a fundamental bottleneck: key/value (KV) tensors…

A common challenge in continual learning (CL) is catastrophic forgetting, where the performance on old tasks drops after new, additional tasks are learned. In this paper, we propose a novel framework called ReCL to slow down forgetting in…

LLM training is scaled up to 10Ks of GPUs by a mix of data-(DP) and model-parallel (MP) execution. Critical to achieving efficiency is tensor-parallel (TP; a form of MP) execution within tightly-coupled subsets of GPUs, referred to as a…

As Large Language Models (LLMs) are increasingly adopted in edge intelligence to power domain-specific applications and personalized services, the quality and efficiency of the LLM post-training phase-including fine-tuning and inference,…

As emerging deep neural network (DNN) models continue to grow in size, using large GPU clusters to train DNNs is becoming an essential requirement to achieving acceptable training times. In this paper, we consider the case where future…

Maximizing training throughput and cost-efficiency of RL for LLMs is essential to democratize this advanced technique. One promising but challenging approach is to deploy such a computational workflow over heterogeneous GPUs. Unlike…

Parallelism is a ubiquitous method for accelerating machine learning algorithms. However, theoretical analysis of parallel learning is usually done in an algorithm- and protocol-specific setting, giving little insight about how changes in…

When training large language models (LLMs), it is common practice to track downstream task performance throughout the training process and select the checkpoint with the highest validation score. However, downstream metrics often exhibit…

Training machine learning (ML) algorithms is a computationally intensive process, which is frequently memory-bound due to repeatedly accessing large training datasets. As a result, processor-centric systems (e.g., CPU, GPU) suffer from…

Microservice architectures are increasingly used to modularize IoT applications and deploy them in distributed and heterogeneous edge computing environments. Over time, these microservice-based IoT applications are susceptible to…

We introduce a new training paradigm that enforces interval constraints on neural network parameter space to control forgetting. Contemporary Continual Learning (CL) methods focus on training neural networks efficiently from a stream of…

Datacenter applications often rely on remote procedure calls (RPCs) for fast, efficient, and secure communication. However, RPCs are slow, inefficient, and hard to use as they require expensive serialization and compression to communicate…

Model checkpoints are critical Deep Learning (DL) artifacts that enable fault tolerance for training and downstream applications, such as inference. However, writing checkpoints to persistent storage, and other I/O aspects of DL training,…

Collective communication is becoming increasingly important in data center and supercomputer workloads with an increase in distributed AI related jobs. However, existing libraries that provide collective support such as NCCL, RCCL, and…

Deep learning (DL) applications are increasingly being deployed on HPC systems, to leverage the massive parallelism and computing power of those systems for DL model training. While significant effort has been put to facilitate distributed…

This paper proposes ScalePool, a novel cluster architecture designed to interconnect numerous accelerators using unified hardware interconnects rather than traditional long-distance networking. ScalePool integrates Accelerator-Centric Links…