Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.

/pdf/language-models-are-few-shot-learners-2fg8gvia7m.pdf

Language Models are Few-Shot Learners

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

We present the Natural Questions corpus, a question answering data set. Questions consist of real anonymized, aggregated queries issued to the Google search engine. An annotator is presented with a...

/pdf/natural-questions-a-benchmark-for-question-answering-10mm1ytgmc.pdf

Natural Questions: A Benchmark for Question Answering Research

Unsupervised word representations are very useful in NLP tasks both as inputs to learning algorithms and as extra word features in NLP systems. However, most of these models are built with only local context and one representation per word. This is problematic because words are often polysemous and global context can also provide useful information for learning word meanings. We present a new neural network architecture which 1) learns word embeddings that better capture the semantics of words by incorporating both local and global document context, and 2) accounts for homonymy and polysemy by learning multiple embeddings per word. We introduce a new dataset with human judgments on pairs of words in sentential context, and evaluate our model on it, showing that our model outperforms competitive baselines and other neural language models.

/pdf/improving-word-representations-via-global-context-and-4x4fd6wohu.pdf

Improving Word Representations via Global Context and Multiple Word Prototypes

While achieving state-of-the-art results in multiple tasks and languages, translation-based cross-lingual transfer is often overlooked in favour of massively multilingual pre-trained encoders. Arguably, this is due to its main limitations: 1) translation errors percolating to the classification phase and 2) the insufficient expressiveness of the maximum-likelihood translation. To remedy this, we propose a new technique that integrates both steps of the traditional pipeline (translation and classification) into a single model, by treating the intermediate translations as a latent random variable. As a result, 1) the neural machine translation system can be fine-tuned with a variant of Minimum Risk Training where the reward is the accuracy of the downstream task classifier. Moreover, 2) multiple samples can be drawn to approximate the expected loss across all possible translations during inference. We evaluate our novel latent translation-based model on a series of multilingual NLU tasks, including commonsense reasoning, paraphrase identification, and natural language inference. We report gains for both zero-shot and few-shot learning setups, up to 2.7 accuracy points on average, which are even more prominent for low-resource languages (e.g., Haitian Creole). Finally, we carry out in-depth analyses comparing different underlying NMT models and assessing the impact of alternative translations on the downstream performance.

Modelling Latent Translations for Cross-Lingual Transfer

This paper describes a neural semantic parser that maps natural language utterances onto logical forms which can be executed against a task-specific environment, such as a knowledge base or a database, to produce a response. The parser generates tree-structured logical forms with a transition-based approach which combines a generic tree-generation algorithm with domain-general operations defined by the logical language. The generation process is modeled by structured recurrent neural networks, which provide a rich encoding of the sentential context and generation history for making predictions. To tackle mismatches between natural language and logical form tokens, various attention mechanisms are explored. Finally, we consider different training settings for the neural semantic parser, including a fully supervised training where annotated logical forms are given, weakly-supervised training where denotations are provided, and distant supervision where only unlabeled sentences and a knowledge base are available. Experiments across a wide range of datasets demonstrate the effectiveness of our parser.

Learning an Executable Neural Semantic Parser

Syntax is fundamental to our thinking about language. Although neural networks are very successful in many tasks, they do not explicitly model syntactic structure. Failing to capture the structure of inputs could lead to generalization problems and over-parametrization. In the present work, we propose a new syntax-aware language model: Syntactic Ordered Memory (SOM). The model explicitly models the structure with a one-step look-ahead parser and maintains the conditional probability setting of the standard language model. Experiments show that SOM can achieve strong results in language modeling and syntactic generalization tests, while using fewer parameters then other models.

Explicitly Modeling Syntax in Language Model improves Generalization.

In this paper, we propose a new domain adaptation method called $\textit{back-training}$, a superior alternative to self-training. While self-training results in synthetic training data of the form quality inputs aligned with noisy outputs, back-training results in noisy inputs aligned with quality outputs. Our experimental results on unsupervised domain adaptation of question generation and passage retrieval models from $\textit{Natural Questions}$ domain to the machine learning domain show that back-training outperforms self-training by a large margin: 9.3 BLEU-1 points on generation, and 7.9 accuracy points on top-1 retrieval. We release $\textit{MLQuestions}$, a domain-adaptation dataset for the machine learning domain containing 50K unaligned passages and 35K unaligned questions, and 3K aligned passage and question pairs. Our data and code are available at this https URL

Back-Training excels Self-Training at Unsupervised Domain Adaptation of Question Generation and Passage Retrieval

To explain NLP models, many methods inform which inputs tokens are important for a prediction. However, an open question is if these methods accurately reflect the model's logic, a property often called faithfulness. In this work, we adapt and improve a recently proposed faithfulness benchmark from computer vision called ROAR (RemOve And Retrain), by Hooker et al. (2019). 
We improve ROAR by recursively removing dataset redundancies, which otherwise interfere with ROAR. We adapt and apply ROAR, to popular NLP importance measures, namely attention, gradient, and integrated gradients. Additionally, we use mutual information as an additional baseline. Evaluation is done on a suite of classification tasks often used in the faithfulness of attention literature. Finally, we propose a scalar faithfulness metric, which makes it easy to compare results across papers. 
We find that, importance measures considered to be unfaithful for computer vision tasks perform favorably for NLP tasks, the faithfulness of an importance measure is task-dependent, and the computational overhead of integrated gradient is rarely justified.

Siva Reddy

Papers

Modelling Latent Translations for Cross-Lingual Transfer

Learning an Executable Neural Semantic Parser

Explicitly Modeling Syntax in Language Model improves Generalization.

Back-Training excels Self-Training at Unsupervised Domain Adaptation of Question Generation and Passage Retrieval

Evaluating the Faithfulness of Importance Measures in NLP by Recursively Masking Allegedly Important Tokens and Retraining