The performance of abstractive text summarization has been greatly boosted by pre-trained language models recently. The main concern of existing abstractive summarization methods is the factual inconsistency problem of their generated summary. To alleviate the problem, many efforts have focused on developing effective factuality evaluation metrics based on natural language inference and question answering et al. However, they have limitations of high computational complexity and relying on annotated data. Most recently, large language models such as ChatGPT have shown strong ability in not only natural language understanding but also natural language inference. In this paper, we study the factual inconsistency evaluation ability of ChatGPT under the zero-shot setting by evaluating it on the coarse-grained and fine-grained factuality evaluation tasks including binary natural language inference (NLI), summary ranking, and consistency rating. Experimental results show that ChatGPT outperforms previous SOTA evaluation metrics on 6/9 datasets across three tasks, demonstrating its great potential for assessing factual inconsistency in the zero-shot setting. The results also highlight the importance of prompt design and the need for future efforts to address ChatGPT's limitations on evaluation bias, wrong reasoning, and hallucination.
相關內容
Factual consistency evaluation is often conducted using Natural Language Inference (NLI) models, yet these models exhibit limited success in evaluating summaries. Previous work improved such models with synthetic training data. However, the data is typically based on perturbed human-written summaries, which often differ in their characteristics from real model-generated summaries and have limited coverage of possible factual errors. Alternatively, large language models (LLMs) have recently shown promising results in directly evaluating generative tasks, but are too computationally expensive for practical use. Motivated by these limitations, we introduce TrueTeacher, a method for generating synthetic data by annotating diverse model-generated summaries using a LLM. Unlike prior work, TrueTeacher does not rely on human-written summaries, and is multilingual by nature. Experiments on the TRUE benchmark show that a student model trained using our data, substantially outperforms both the state-of-the-art model with similar capacity, and the LLM teacher. In a systematic study, we compare TrueTeacher to existing synthetic data generation methods and demonstrate its superiority and robustness to domain-shift. Using the the mFACE dataset, we also show that our method generalizes to multilingual scenarios. Finally, we release a large-scale synthetic dataset with 1.4M examples generated using TrueTeacher.
We present evidence that language models can learn meaning despite being trained only to perform next token prediction on text, specifically a corpus of programs. Each program is preceded by a specification in the form of (textual) input-output examples. Working with programs enables us to precisely define concepts relevant to meaning in language (e.g., correctness and semantics), making program synthesis well-suited as an intermediate testbed for characterizing the presence (or absence) of meaning in language models. We first train a Transformer model on the corpus of programs, then probe the trained model's hidden states as it completes a program given a specification. Despite providing no inductive bias toward learning the semantics of the language, we find that a linear probe is able to extract abstractions of both current and future program states from the model states. Moreover, there is a strong, statistically significant correlation between the accuracy of the probe and the model's ability to generate a program that implements the specification. To evaluate whether the semantics are represented in the model states rather than learned by the probe, we design a novel experimental procedure that intervenes on the semantics of the language while preserving the lexicon and syntax. We also demonstrate that the model learns to generate correct programs that are, on average, shorter than those in the training set, which is evidence that language model outputs may differ from the training distribution in semantically meaningful ways. In summary, this paper does not propose any new techniques for training language models, but develops an experimental framework for and provides insights into the acquisition and representation of (formal) meaning in language models.
The ability of artificial intelligence (AI) systems to perceive and comprehend audio signals is crucial for many applications. Although significant progress has been made in this area since the development of AudioSet, most existing models are designed to map audio inputs to pre-defined, discrete sound label sets. In contrast, humans possess the ability to not only classify sounds into coarse-grained categories, but also to listen to the details of the sounds, explain the reason for the predictions, think what the sound infers, and understand the scene and what action needs to be taken. Such capabilities beyond perception are not yet present in existing audio models. On the other hand, modern large language models (LLMs) exhibit emerging reasoning ability but they lack audio perception capabilities. Therefore, we ask the question: can we build an AI model that has both audio perception and a reasoning ability? In this paper, we propose a novel audio foundation model, called LTU (Listen, Think, and Understand). To train LTU, we created a new OpenAQA-5M dataset consisting of 1.9 million closed-ended and 3.7 million open-ended, diverse (audio, question, answer) tuples, and used an autoregressive training framework and a perception-to-understanding curriculum. LTU demonstrates strong performance and generalization ability on conventional audio tasks such as classification and captioning. Moreover, it exhibits remarkable reasoning and comprehension abilities in the audio domain. To the best of our knowledge, LTU is the first audio-enabled large language model that bridges audio perception with advanced reasoning.
Despite substantial progress in abstractive text summarization to generate fluent and informative texts, the factual inconsistency in the generated summaries remains an important yet challenging problem to be solved. In this paper, we construct causal graphs for abstractive text summarization and identify the intrinsic causes of the factual inconsistency, i.e., the language bias and irrelevancy bias, and further propose a debiasing framework, named CoFactSum, to alleviate the causal effects of these biases by counterfactual estimation. Specifically, the proposed CoFactSum provides two counterfactual estimation strategies, i.e., Explicit Counterfactual Masking with an explicit dynamic masking strategy, and Implicit Counterfactual Training with an implicit discriminative cross-attention mechanism. Meanwhile, we design a Debiasing Degree Adjustment mechanism to dynamically adapt the debiasing degree at each decoding step. Extensive experiments on two widely-used summarization datasets demonstrate the effectiveness of CoFactSum in enhancing the factual consistency of generated summaries compared with several baselines.
An important problem of the sequence-to-sequence neural models widely used in abstractive summarization is exposure bias. To alleviate this problem, re-ranking systems have been applied in recent years. Despite some performance improvements, this approach remains underexplored. Previous works have mostly specified the rank through the ROUGE score and aligned candidate summaries, but there can be quite a large gap between the lexical overlap metric and semantic similarity. In this paper, we propose a novel training method in which a re-ranker balances the lexical and semantic quality. We further newly define false positives in ranking and present a strategy to reduce their influence. Experiments on the CNN/DailyMail and XSum datasets show that our method can estimate the meaning of summaries without seriously degrading the lexical aspect. More specifically, it achieves an 89.67 BERTScore on the CNN/DailyMail dataset, reaching new state-of-the-art performance. Our code is publicly available at //github.com/jeewoo1025/BalSum.
The success of deep learning models on multi-hop fact verification has prompted researchers to understand the behavior behind their veracity. One possible way is erasure search: obtaining the rationale by entirely removing a subset of input without compromising the veracity prediction. Although extensively explored, existing approaches fall within the scope of the single-granular (tokens or sentences) explanation, which inevitably leads to explanation redundancy and inconsistency. To address such issues, this paper explores the viability of multi-granular rationale extraction with consistency and faithfulness for explainable multi-hop fact verification. In particular, given a pretrained veracity prediction model, both the token-level explainer and sentence-level explainer are trained simultaneously to obtain multi-granular rationales via differentiable masking. Meanwhile, three diagnostic properties (fidelity, consistency, salience) are introduced and applied to the training process, to ensure that the extracted rationales satisfy faithfulness and consistency. Experimental results on three multi-hop fact verification datasets show that the proposed approach outperforms some state-of-the-art baselines.
We present a new pre-trained language model (PLM) for modern Hebrew, termed AlephBERTGimmel, which employs a much larger vocabulary (128K items) than standard Hebrew PLMs before. We perform a contrastive analysis of this model against all previous Hebrew PLMs (mBERT, heBERT, AlephBERT) and assess the effects of larger vocabularies on task performance. Our experiments show that larger vocabularies lead to fewer splits, and that reducing splits is better for model performance, across different tasks. All in all this new model achieves new SOTA on all available Hebrew benchmarks, including Morphological Segmentation, POS Tagging, Full Morphological Analysis, NER, and Sentiment Analysis. Subsequently we advocate for PLMs that are larger not only in terms of number of layers or training data, but also in terms of their vocabulary. We release the new model publicly for unrestricted use.
While recent studies on semi-supervised learning have shown remarkable progress in leveraging both labeled and unlabeled data, most of them presume a basic setting of the model is randomly initialized. In this work, we consider semi-supervised learning and transfer learning jointly, leading to a more practical and competitive paradigm that can utilize both powerful pre-trained models from source domain as well as labeled/unlabeled data in the target domain. To better exploit the value of both pre-trained weights and unlabeled target examples, we introduce adaptive consistency regularization that consists of two complementary components: Adaptive Knowledge Consistency (AKC) on the examples between the source and target model, and Adaptive Representation Consistency (ARC) on the target model between labeled and unlabeled examples. Examples involved in the consistency regularization are adaptively selected according to their potential contributions to the target task. We conduct extensive experiments on several popular benchmarks including CUB-200-2011, MIT Indoor-67, MURA, by fine-tuning the ImageNet pre-trained ResNet-50 model. Results show that our proposed adaptive consistency regularization outperforms state-of-the-art semi-supervised learning techniques such as Pseudo Label, Mean Teacher, and MixMatch. Moreover, our algorithm is orthogonal to existing methods and thus able to gain additional improvements on top of MixMatch and FixMatch. Our code is available at //github.com/SHI-Labs/Semi-Supervised-Transfer-Learning.
Recent work pre-training Transformers with self-supervised objectives on large text corpora has shown great success when fine-tuned on downstream NLP tasks including text summarization. However, pre-training objectives tailored for abstractive text summarization have not been explored. Furthermore there is a lack of systematic evaluation across diverse domains. In this work, we propose pre-training large Transformer-based encoder-decoder models on massive text corpora with a new self-supervised objective. In PEGASUS, important sentences are removed/masked from an input document and are generated together as one output sequence from the remaining sentences, similar to an extractive summary. We evaluated our best PEGASUS model on 12 downstream summarization tasks spanning news, science, stories, instructions, emails, patents, and legislative bills. Experiments demonstrate it achieves state-of-the-art performance on all 12 downstream datasets measured by ROUGE scores. Our model also shows surprising performance on low-resource summarization, surpassing previous state-of-the-art results on 6 datasets with only 1000 examples. Finally we validated our results using human evaluation and show that our model summaries achieve human performance on multiple datasets.
Pre-training text representations has recently been shown to significantly improve the state-of-the-art in many natural language processing tasks. The central goal of pre-training is to learn text representations that are useful for subsequent tasks. However, existing approaches are optimized by minimizing a proxy objective, such as the negative log likelihood of language modeling. In this work, we introduce a learning algorithm which directly optimizes model's ability to learn text representations for effective learning of downstream tasks. We show that there is an intrinsic connection between multi-task pre-training and model-agnostic meta-learning with a sequence of meta-train steps. The standard multi-task learning objective adopted in BERT is a special case of our learning algorithm where the depth of meta-train is zero. We study the problem in two settings: unsupervised pre-training and supervised pre-training with different pre-training objects to verify the generality of our approach.Experimental results show that our algorithm brings improvements and learns better initializations for a variety of downstream tasks.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191