Kun Xiong

Wei Yang, Yuqing Xie, Aileen Lin et al.

We demonstrate an end-to-end question answering system that integrates BERT with the open-source Anserini information retrieval toolkit. In contrast to most question answering and reading comprehension models today, which operate over small amounts of input text, our system integrates best practices from IR with a BERT-based reader to identify answers from a large corpus of Wikipedia articles in an end-to-end fashion. We report large improvements over previous results on a standard benchmark test collection, showing that fine-tuning pretrained BERT with SQuAD is sufficient to achieve high accuracy in identifying answer spans.

M. Li, H. Bai, L. Tan et al.

Measuring sentence semantic similarity using pre-trained language models such as BERT generally yields unsatisfactory zero-shot performance, and one main reason is ineffective token aggregation methods such as mean pooling. In this paper, we demonstrate under a Bayesian framework that distance between primitive statistics such as the mean of word embeddings are fundamentally flawed for capturing sentence-level semantic similarity. To remedy this issue, we propose to learn a categorical variational autoencoder (VAE) based on off-the-shelf pre-trained language models. We theoretically prove that measuring the distance between the latent categorical mixtures, namely Latte-Mix, can better reflect the true sentence semantic similarity. In addition, our Bayesian framework provides explanations for why models finetuned on labelled sentence pairs have better zero-shot performance. We also empirically demonstrate that these finetuned models could be further improved by Latte-Mix. Our method not only yields the state-of-the-art zero-shot performance on semantic similarity datasets such as STS, but also enjoy the benefits of fast training and having small memory footprints.

Mohan Zhang, Luchen Tan, Zhengkai Tu et al.

In the paraphrase generation task, source sentences often contain phrases that should not be altered. Which phrases, however, can be context dependent and can vary by application. Our solution to this challenge is to provide the user with explicit tags that can be placed around any arbitrary segment of text to mean "don't change me!" when generating a paraphrase; the model learns to explicitly copy these phrases to the output. The contribution of this work is a novel data generation technique using distant supervision that allows us to start with a pretrained sequence-to-sequence model and fine-tune a paraphrase generator that exhibits this behavior, allowing user-controllable paraphrase generation. Additionally, we modify the loss during fine-tuning to explicitly encourage diversity in model output. Our technique is language agnostic, and we report experiments in English and Chinese.

He Bai, Peng Shi, Jimmy Lin et al.

Transformers are powerful for sequence modeling. Nearly all state-of-the-art language models and pre-trained language models are based on the Transformer architecture. However, it distinguishes sequential tokens only with the token position index. We hypothesize that better contextual representations can be generated from the Transformer with richer positional information. To verify this, we propose a segment-aware Transformer (Segatron), by replacing the original token position encoding with a combined position encoding of paragraph, sentence, and token. We first introduce the segment-aware mechanism to Transformer-XL, which is a popular Transformer-based language model with memory extension and relative position encoding. We find that our method can further improve the Transformer-XL base model and large model, achieving 17.1 perplexity on the WikiText-103 dataset. We further investigate the pre-training masked language modeling task with Segatron. Experimental results show that BERT pre-trained with Segatron (SegaBERT) can outperform BERT with vanilla Transformer on various NLP tasks, and outperforms RoBERTa on zero-shot sentence representation learning.

He Bai, Peng Shi, Jimmy Lin et al.

The semantics of a text is manifested not only by what is read, but also by what is not read. In this article, we will study how the implicit "not read" information such as end-of-paragraph (\eop) and end-of-sequence (\eos) affect the quality of text generation. Specifically, we find that the pre-trained language model GPT2 can generate better continuations by learning to generate the \eop in the fine-tuning stage. Experimental results on English story generation show that \eop can lead to higher BLEU score and lower \eos perplexity. We also conduct experiments on a self-collected Chinese essay dataset with Chinese-GPT2, a character level LM without \eop or \eos during pre-training. Experimental results show that the Chinese GPT2 can generate better essay endings with \eop.

Wei Yang, Yuqing Xie, Luchen Tan et al.

Recently, a simple combination of passage retrieval using off-the-shelf IR techniques and a BERT reader was found to be very effective for question answering directly on Wikipedia, yielding a large improvement over the previous state of the art on a standard benchmark dataset. In this paper, we present a data augmentation technique using distant supervision that exploits positive as well as negative examples. We apply a stage-wise approach to fine tuning BERT on multiple datasets, starting with data that is "furthest" from the test data and ending with the "closest". Experimental results show large gains in effectiveness over previous approaches on English QA datasets, and we establish new baselines on two recent Chinese QA datasets.

Kun Xiong, Anqi Cui, Zefeng Zhang et al.

Neural conversational models tend to produce generic or safe responses in different contexts, e.g., reply \textit{"Of course"} to narrative statements or \textit{"I don't know"} to questions. In this paper, we propose an end-to-end approach to avoid such problem in neural generative models. Additional memory mechanisms have been introduced to standard sequence-to-sequence (seq2seq) models, so that context can be considered while generating sentences. Three seq2seq models, which memorize a fix-sized contextual vector from hidden input, hidden input/output and a gated contextual attention structure respectively, have been trained and tested on a dataset of labeled question-answering pairs in Chinese. The model with contextual attention outperforms others including the state-of-the-art seq2seq models on perplexity test. The novel contextual model generates diverse and robust responses, and is able to carry out conversations on a wide range of topics appropriately.