GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding

Neural network scaling has been critical for improving the model quality in many real-world machine learning applications with vast amounts of training data and compute. Although this trend of scaling is affirmed to be a sure-fire approach for better model quality, there are challenges on the path such as the computation cost, ease of programming, and efficient implementation on parallel devices. GShard is a module composed of a set of lightweight annotation APIs and an extension to the XLA compiler. It provides an elegant way to express a wide range of parallel computation patterns with minimal changes to the existing model code. GShard enabled us to scale up multilingual neural machine translation Transformer model with Sparsely-Gated Mixture-of-Experts beyond 600 billion parameters using automatic sharding. We demonstrate that such a giant model can efficiently be trained on 2048 TPU v3 accelerators in 4 days to achieve far superior quality for translation from 100 languages to English compared to the prior art.

Talking-Heads Attention

We introduce "talking-heads attention" - a variation on multi-head attention which includes linearprojections across the attention-heads dimension, immediately before and after the softmax operation.While inserting only a small number of additional parameters and a moderate amount of additionalcomputation, talking-heads attention leads to better perplexities on masked language modeling tasks, aswell as better quality when transfer-learning to language comprehension and question answering tasks.

How Much Knowledge Can You Pack Into the Parameters of a Language Model?

It has recently been observed that neural language models trained on unstructured text can implicitly store and retrieve knowledge using natural language queries. In this short paper, we measure the practical utility of this approach by fine-tuning pre-trained models to answer questions without access to any external context or knowledge. We show that this approach scales surprisingly well with model size and outperforms models that explicitly look up knowledge on the open-domain variants of Natural Questions and WebQuestions.

GLU Variants Improve Transformer

Gated Linear Units (arXiv:1612.08083) consist of the component-wise product of two linear projections, one of which is first passed through a sigmoid function. Variations on GLU are possible, using different nonlinear (or even linear) functions in place of sigmoid. We test these variants in the feed-forward sublayers of the Transformer (arXiv:1706.03762) sequence-to-sequence model, and find that some of them yield quality improvements over the typically-used ReLU or GELU activations.

Faster Transformer Decoding: N-gram Masked Self-Attention

Motivated by the fact that most of the information relevant to the prediction of target tokens is drawn from the source sentence $S=s_1, \ldots, s_S$, we propose truncating the target-side window used for computing self-attention by making an $N$-gram assumption. Experiments on WMT EnDe and EnFr data sets show that the $N$-gram masked self-attention model loses very little in BLEU score for $N$ values in the range $4, \ldots, 8$, depending on the task.

Fast Transformer Decoding: One Write-Head is All You Need

Multi-head attention layers, as used in the Transformer neural sequence model, are a powerful alternative to RNNs for moving information across and between sequences. While training these layers is generally fast and simple, due to parallelizability across the length of the sequence, incremental inference (where such paralleization is impossible) is often slow, due to the memory-bandwidth cost of repeatedly loading the large "keys" and "values" tensors. We propose a variant called multi-query attention, where the keys and values are shared across all of the different attention "heads", greatly reducing the size of these tensors and hence the memory bandwidth requirements of incremental decoding. We verify experimentally that the resulting models can indeed be much faster to decode, and incur only minor quality degradation from the baseline.

Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer

Transfer learning, where a model is first pre-trained on a data-rich task before being fine-tuned on a downstream task, has emerged as a powerful technique in natural language processing (NLP). The effectiveness of transfer learning has given rise to a diversity of approaches, methodology, and practice. In this paper, we explore the landscape of transfer learning techniques for NLP by introducing a unified framework that converts every language problem into a text-to-text format. Our systematic study compares pre-training objectives, architectures, unlabeled datasets, transfer approaches, and other factors on dozens of language understanding tasks. By combining the insights from our exploration with scale and our new "Colossal Clean Crawled Corpus", we achieve state-of-the-art results on many benchmarks covering summarization, question answering, text classification, and more. To facilitate future work on transfer learning for NLP, we release our dataset, pre-trained models, and code.

High Resolution Medical Image Analysis with Spatial Partitioning

Medical images such as 3D computerized tomography (CT) scans and pathology images, have hundreds of millions or billions of voxels/pixels. It is infeasible to train CNN models directly on such high resolution images, because neural activations of a single image do not fit in the memory of a single GPU/TPU, and naive data and model parallelism approaches do not work. Existing image analysis approaches alleviate this problem by cropping or down-sampling input images, which leads to complicated implementation and sub-optimal performance due to information loss. In this paper, we implement spatial partitioning, which internally distributes the input and output of convolutional layers across GPUs/TPUs. Our implementation is based on the Mesh-TensorFlow framework and the computation distribution is transparent to end users. With this technique, we train a 3D Unet on up to 512 by 512 by 512 resolution data. To the best of our knowledge, this is the first work for handling such high resolution images end-to-end.