Publications (32)
Simple and Scalable Strategies to Continually Pre-train Large Language Models
Adam Ibrahim, Benjamin Thérien, Kshitij Gupta +5
Large language models (LLMs) are routinely pre-trained on billions of tokens, only to start the process over again once new data becomes available. A much more efficient solution i…
Demystifying the Communication Characteristics for Distributed Transformer Models
Quentin Anthony, Benjamin Michalowicz, Jacob Hatef +5
Deep learning (DL) models based on the transformer architecture have revolutionized many DL applications such as large language models (LLMs), vision transformers, audio generation…
Folding Tensor and Sequence Parallelism for Memory-Efficient Transformer Training & Inference
Vasu Shyam, Anna Golubeva, Quentin Anthony
We present tensor and sequence parallelism (TSP), a parallel execution strategy that folds tensor parallelism and sequence parallelism onto a single device axis. In conventional mu…
Exploiting Inter-Layer Expert Affinity for Accelerating Mixture-of-Experts Model Inference
Jinghan Yao, Quentin Anthony, Aamir Shafi +3
In large language models like the Generative Pre-trained Transformer, the Mixture of Experts paradigm has emerged as a powerful technique for enhancing model expressiveness and acc…
Emergent and Predictable Memorization in Large Language Models
Stella Biderman, USVSN Sai Prashanth, Lintang Sutawika +4
Memorization, or the tendency of large language models (LLMs) to output entire sequences from their training data verbatim, is a key concern for safely deploying language models. I…
PyLO: Towards Accessible Learned Optimizers in PyTorch
Paul Janson, Benjamin Therien, Quentin Anthony +3
Learned optimizers have been an active research topic over the past decade, with increasing progress toward practical, general-purpose optimizers that can serve as drop-in replacem…
Eagle and Finch: RWKV with Matrix-Valued States and Dynamic Recurrence
Bo Peng, Daniel Goldstein, Quentin Anthony +27
We present Eagle (RWKV-5) and Finch (RWKV-6), sequence models improving upon the RWKV (RWKV-4) architecture. Our architectural design advancements include multi-headed matrix-value…
MCR-DL: Mix-and-Match Communication Runtime for Deep Learning
Quentin Anthony, Ammar Ahmad Awan, Jeff Rasley +5
In recent years, the training requirements of many state-of-the-art Deep Learning (DL) models have scaled beyond the compute and memory capabilities of a single processor, and nece…
The Zamba2 Suite: Technical Report
Paolo Glorioso, Quentin Anthony, Yury Tokpanov +5
In this technical report, we present the Zamba2 series -- a suite of 1.2B, 2.7B, and 7.4B parameter hybrid Mamba2-transformer models that achieve state of the art performance again…
Accelerating Large Language Model Training with Hybrid GPU-based Compression
Lang Xu, Quentin Anthony, Qinghua Zhou +5
Data Parallelism (DP), Tensor Parallelism (TP), and Pipeline Parallelism (PP) are the three strategies widely adopted to enable fast and efficient Large Language Model (LLM) traini…
HyPar-Flow: Exploiting MPI and Keras for Scalable Hybrid-Parallel DNN Training using TensorFlow
Ammar Ahmad Awan, Arpan Jain, Quentin Anthony +2
To reduce training time of large-scale DNNs, scientists have started to explore parallelization strategies like data-parallelism, model-parallelism, and hybrid-parallelism. While d…
GPT-NeoX-20B: An Open-Source Autoregressive Language Model
Sid Black, Stella Biderman, Eric Hallahan +14
We introduce GPT-NeoX-20B, a 20 billion parameter autoregressive language model trained on the Pile, whose weights will be made freely and openly available to the public through a…
Zyda-2: a 5 Trillion Token High-Quality Dataset
Yury Tokpanov, Paolo Glorioso, Quentin Anthony +1
In this technical report, we present Zyda-2: a five trillion token dataset for language model pretraining. Zyda-2 was used to train our Zamba2 series of models which are state-of-t…
Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs
Kyle O'Brien, Stephen Casper, Quentin Anthony +7
Open-weight AI systems offer unique benefits, including enhanced transparency, open research, and decentralized access. However, they are vulnerable to tampering attacks which can…
RedPajama: an Open Dataset for Training Large Language Models
Maurice Weber, Daniel Fu, Quentin Anthony +16
Large language models are increasingly becoming a cornerstone technology in artificial intelligence, the sciences, and society as a whole, yet the optimal strategies for dataset co…
ZAYA1-8B Technical Report
Robert Washbourne, Rishi Iyer, Tomas Figliolia +15
We present ZAYA1-8B, a reasoning-focused mixture-of-experts (MoE) model with 700M active and 8B total parameters, built on Zyphra's MoE++ architecture. ZAYA1-8B's core pretraining,…
Compressed Convolutional Attention: Efficient Attention in a Compressed Latent Space
Tomas Figliolia, Nicholas Alonso, Rishi Iyer +2
Multi-headed Attention's (MHA) quadratic compute and linearly growing KV-cache make long-context transformers expensive to train and serve. Prior works such as Grouped Query Attent…
Characterizing Communication Patterns in Distributed Large Language Model Inference
Lang Xu, Kaushik Kandadi Suresh, Quentin Anthony +2
Large Language Models (LLMs) built on transformer architectures have transformed natural language processing, achieving remarkable performance across diverse applications. While di…
Zyda: A 1.3T Dataset for Open Language Modeling
Yury Tokpanov, Beren Millidge, Paolo Glorioso +4
The size of large language models (LLMs) has scaled dramatically in recent years and their computational and data requirements have surged correspondingly. State-of-the-art languag…
Training Foundation Models on a Full-Stack AMD Platform: Compute, Networking, and System Design
Quentin Anthony, Yury Tokpanov, Skyler Szot +18
We report on the first large-scale mixture-of-experts (MoE) pretraining study on pure AMD hardware, utilizing both MI300X GPUs and Pollara networking. We distill practical guidance…
CHIRP: A Fine-Grained Benchmark for Open-Ended Response Evaluation in Vision-Language Models
Alexis Roger, Prateek Humane, Daniel Z. Kaplan +7
The proliferation of Vision-Language Models (VLMs) in the past several years calls for rigorous and comprehensive evaluation methods and benchmarks. This work analyzes existing VLM…
Comparative Study of Large Language Model Architectures on Frontier
Junqi Yin, Avishek Bose, Guojing Cong +2
Large language models (LLMs) have garnered significant attention in both the AI community and beyond. Among these, the Generative Pre-trained Transformer (GPT) has emerged as the d…
BlackMamba: Mixture of Experts for State-Space Models
Quentin Anthony, Yury Tokpanov, Paolo Glorioso +1
State-space models (SSMs) have recently demonstrated competitive performance to transformers at large-scale language modeling benchmarks while achieving linear time and memory comp…
Continual Pre-Training of Large Language Models: How to (re)warm your model?
Kshitij Gupta, Benjamin Thérien, Adam Ibrahim +5
Large language models (LLMs) are routinely pre-trained on billions of tokens, only to restart the process over again once new data becomes available. A much cheaper and more effici…
Scaling Large Language Model Training on Frontier with Low-Bandwidth Partitioning
Lang Xu, Quentin Anthony, Jacob Hatef +4
Scaling up Large Language Model(LLM) training involves fitting a tremendous amount of training parameters across a limited number of workers. However, methods like ZeRO-3 that dras…
Zamba: A Compact 7B SSM Hybrid Model
Paolo Glorioso, Quentin Anthony, Yury Tokpanov +4
In this technical report, we present Zamba, a novel 7B SSM-transformer hybrid model which achieves competitive performance against leading open-weight models at a comparable scale.…
Covenant-72B: Pre-Training a 72B LLM with Trustless Peers Over-the-Internet
Joel Lidin, Amir Sarfi, Erfan Miahi +6
Recently, there has been increased interest in globally distributed training, which has the promise to both reduce training costs and democratize participation in building large-sc…
RWKV: Reinventing RNNs for the Transformer Era
Bo Peng, Eric Alcaide, Quentin Anthony +31
Transformers have revolutionized almost all natural language processing (NLP) tasks but suffer from memory and computational complexity that scales quadratically with sequence leng…
Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clusters
Vasudev Shyam, Jonathan Pilault, Emily Shepperd +2
Our formulation reveals that the reduction across the sequence axis can be efficiently computed in parallel through a tree reduction. Our algorithm, called Tree Attention, for para…
Cross-layer Visualization and Profiling of Network and I/O Communication for HPC Clusters
Pouya Kousha, Quentin Anthony, Hari Subramoni +1
Understanding and visualizing the full-stack performance trade-offs and interplay between HPC applications, MPI libraries, the communication fabric, and the file system is a challe…
The Case for Co-Designing Model Architectures with Hardware
Quentin Anthony, Jacob Hatef, Deepak Narayanan +6
While GPUs are responsible for training the vast majority of state-of-the-art deep learning models, the implications of their architecture are often overlooked when designing new d…
Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling
Stella Biderman, Hailey Schoelkopf, Quentin Anthony +10
How do large language models (LLMs) develop and evolve over the course of training? How do these patterns change as models scale? To answer these questions, we introduce \textit{Py…