Can I read Quantized Distributed Training of Large Models with Convergence Guarantees on EtoBox?

Quantized Distributed Training of Large Models with Convergence Guarantees by Markov, Ilia; Vladu, Adrian; Guo, Qi; Alistarh, Dan is a scholarly article available to read on EtoBox.

What is Quantized Distributed Training of Large Models with Convergence Guarantees about?

Communication-reduction techniques are a popular way to improve scalability in data-parallel training of deep neural networks (DNNs). The recent emergence of large language models such as GPT has created the need for new approaches to exploit data-parallelism. Among these, fully-sharded data parallel (FSDP) training is highly popular, yet it still encounters scalability bottlenecks. One reason is that applying compression techniques to FSDP is challenging: as the vast majority of the communication involves the model's weights, direct compression alters convergence and leads to accuracy loss. We present QSDP, a variant of FSDP which supports both gradient and weight quantization with theoretical guarantees, is simple to implement and has essentially no overheads. To derive QSDP we prove that a natural modification of SGD achieves convergence even when we only maintain quantized weights, and thus the domain over which we train consists of quantized points and is, therefore, highly non-convex. We validate this approach by training GPT-family models with up to 1.3 billion parameters on a multi-node cluster. Experiments show that QSDP preserves model accuracy, while completely removing

Author

Markov, Ilia; Vladu, Adrian; Guo, Qi; Alistarh, Dan

Published

2023

Language

EN