This paper describes the DSBA submissions to the Prompting Large Language Models as Explainable Metrics shared task, where systems were submitted to two tracks: small and large summarization tracks. With advanced Large Language Models (LLMs) such as GPT-4, evaluating the quality of Natural Language Generation (NLG) has become increasingly paramount. Traditional similarity-based metrics such as BLEU and ROUGE have shown to misalign with human evaluation and are ill-suited for open-ended generation tasks. To address this issue, we explore the potential capability of LLM-based metrics, especially leveraging open-source LLMs. In this study, wide range of prompts and prompting techniques are systematically analyzed with three approaches: prompting strategy, score aggregation, and explainability. Our research focuses on formulating effective prompt templates, determining the granularity of NLG quality scores and assessing the impact of in-context examples on LLM-based evaluation. Furthermore, three aggregation strategies are compared to identify the most reliable method for aggregating NLG quality scores. To examine explainability, we devise a strategy that generates rationales for the scores and analyzes the characteristics of the explanation produced by the open-source LLMs. Extensive experiments provide insights regarding evaluation capabilities of open-source LLMs and suggest effective prompting strategies.
Diffusion model based Text-to-Image has achieved impressive achievements recently. Although current technology for synthesizing images is highly advanced and capable of generating images with high fidelity, it is still possible to give the show away when focusing on the text area in the generated image. To address this issue, we introduce AnyText, a diffusion-based multilingual visual text generation and editing model, that focuses on rendering accurate and coherent text in the image. AnyText comprises a diffusion pipeline with two primary elements: an auxiliary latent module and a text embedding module. The former uses inputs like text glyph, position, and masked image to generate latent features for text generation or editing. The latter employs an OCR model for encoding stroke data as embeddings, which blend with image caption embeddings from the tokenizer to generate texts that seamlessly integrate with the background. We employed text-control diffusion loss and text perceptual loss for training to further enhance writing accuracy. AnyText can write characters in multiple languages, to the best of our knowledge, this is the first work to address multilingual visual text generation. It is worth mentioning that AnyText can be plugged into existing diffusion models from the community for rendering or editing text accurately. After conducting extensive evaluation experiments, our method has outperformed all other approaches by a significant margin. Additionally, we contribute the first large-scale multilingual text images dataset, AnyWord-3M, containing 3 million image-text pairs with OCR annotations in multiple languages. Based on AnyWord-3M dataset, we propose AnyText-benchmark for the evaluation of visual text generation accuracy and quality. Our project will be open-sourced on https://github.com/tyxsspa/AnyText to improve and promote the development of text generation technology.
GPT-Vision has impressed us on a range of vision-language tasks, but it comes with the familiar new challenge: we have little idea of its capabilities and limitations. In our study, we formalize a process that many have instinctively been trying already to develop "grounded intuition" of this new model. Inspired by the recent movement away from benchmarking in favor of example-driven qualitative evaluation, we draw upon grounded theory and thematic analysis in social science and human-computer interaction to establish a rigorous framework for qualitative evaluation in natural language processing. We use our technique to examine alt text generation for scientific figures, finding that GPT-Vision is particularly sensitive to prompting, counterfactual text in images, and relative spatial relationships. Our method and analysis aim to help researchers ramp up their own grounded intuitions of new models while exposing how GPT-Vision can be applied to make information more accessible.
Clinical natural language processing requires methods that can address domain-specific challenges, such as complex medical terminology and clinical contexts. Recently, large language models (LLMs) have shown promise in this domain. Yet, their direct deployment can lead to privacy issues and are constrained by resources. To address this challenge, we delve into synthetic clinical text generation using LLMs for clinical NLP tasks. We propose an innovative, resource-efficient approach, ClinGen, which infuses knowledge into the process. Our model involves clinical knowledge extraction and context-informed LLM prompting. Both clinical topics and writing styles are drawn from external domain-specific knowledge graphs and LLMs to guide data generation. Our extensive empirical study across 7 clinical NLP tasks and 16 datasets reveals that ClinGen consistently enhances performance across various tasks, effectively aligning the distribution of real datasets and significantly enriching the diversity of generated training instances. Our code is available at \url{https://github.com/ritaranx/ClinGen}.
Automatic evaluation of natural language generation has long been an elusive goal in NLP.A recent paradigm fine-tunes pre-trained language models to emulate human judgements for a particular task and evaluation criterion. Inspired by the generalization ability of instruction-tuned models, we propose a learned metric based on instruction tuning. To test our approach, we collected HEAP, a dataset of human judgements across various NLG tasks and evaluation criteria. Our findings demonstrate that instruction tuning language models on HEAP yields good performance on many evaluation tasks, though some criteria are less trivial to learn than others. Further, jointly training on multiple tasks can yield additional performance improvements, which can be beneficial for future tasks with little to no human annotated data.
Styled Handwritten Text Generation (Styled HTG) is an important task in document analysis, aiming to generate text images with the handwriting of given reference images. In recent years, there has been significant progress in the development of deep learning models for tackling this task. Being able to measure the performance of HTG models via a meaningful and representative criterion is key for fostering the development of this research topic. However, despite the current adoption of scores for natural image generation evaluation, assessing the quality of generated handwriting remains challenging. In light of this, we devise the Handwriting Distance (HWD), tailored for HTG evaluation. In particular, it works in the feature space of a network specifically trained to extract handwriting style features from the variable-lenght input images and exploits a perceptual distance to compare the subtle geometric features of handwriting. Through extensive experimental evaluation on different word-level and line-level datasets of handwritten text images, we demonstrate the suitability of the proposed HWD as a score for Styled HTG. The pretrained model used as backbone will be released to ease the adoption of the score, aiming to provide a valuable tool for evaluating HTG models and thus contributing to advancing this important research area.
With an increasing number of parameters and pre-training data, generative large language models (LLMs) have shown remarkable capabilities to solve tasks with minimal or no task-related examples. Notably, LLMs have been successfully employed as evaluation metrics in text generation tasks. Within this context, we introduce the Eval4NLP 2023 shared task that asks participants to explore prompting and score extraction for machine translation (MT) and summarization evaluation. Specifically, we propose a novel competition setting in which we select a list of allowed LLMs and disallow fine-tuning to ensure a focus on prompting. We present an overview of participants' approaches and evaluate them on a new reference-free test set spanning three language pairs for MT and a summarization dataset. Notably, despite the task's restrictions, the best-performing systems achieve results on par with or even surpassing recent reference-free metrics developed using larger models, including GEMBA and Comet-Kiwi-XXL. Finally, as a separate track, we perform a small-scale human evaluation of the plausibility of explanations given by the LLMs.
Previous work adopts large language models (LLMs) as evaluators to evaluate natural language process (NLP) tasks. However, certain shortcomings, e.g., fairness, scope, and accuracy, persist for current LLM evaluators. To analyze whether LLMs can serve as reliable alternatives to humans, we examine the fine-grained alignment between LLM evaluators and human annotators, particularly in understanding the target evaluation tasks and conducting evaluations that meet diverse criteria. This paper explores both conventional tasks (e.g., story generation) and alignment tasks (e.g., math reasoning), each with different evaluation criteria. Our analysis shows that 1) LLM evaluators can generate unnecessary criteria or omit crucial criteria, resulting in a slight deviation from the experts. 2) LLM evaluators excel in general criteria, such as fluency, but face challenges with complex criteria, such as numerical reasoning. We also find that LLM-pre-drafting before human evaluation can help reduce the impact of human subjectivity and minimize annotation outliers in pure human evaluation, leading to more objective evaluation.
A model is considered well-calibrated when its probability estimate aligns with the actual likelihood of the output being correct. Calibrating language models (LMs) is crucial, as it plays a vital role in detecting and mitigating hallucinations of LMs as well as building more trustworthy models. However, standard calibration techniques may not be suited for LM calibration. For instance, post-processing methods such as temperature scaling do not reorder the candidate generations. On the other hand, training-based methods require fine-tuning the entire model, which is impractical for LMs of large scale. We present LitCab, a lightweight calibration mechanism consisting of a single linear layer that takes the input text representation and predicts a bias term, which is then added to the LM output logits. LitCab improves model calibration by only adding < 2% of the original model parameters. For evaluation, we construct CaT, a benchmark consisting of eight text generation tasks, covering responses ranging from short phrases to paragraphs. We test LitCab with Llama2-7B, where it improves calibration across all tasks, reducing the average ECE score by as large as 30%. We further conduct a comprehensive evaluation with multiple popular open-sourced LMs from GPT and LLaMA families, yielding the following key findings: (i) Larger models within the same family exhibit better calibration on tasks with short generation tasks, but not necessarily for longer ones. (ii) GPT-family models show superior calibration compared to LLaMA, Llama2, and Vicuna models, despite having much fewer parameters. (iii) Fine-tuning pretrained model (e.g., LLaMA) with samples of limited purpose (e.g., conversations) may lead to worse calibration, highlighting the importance of fine-tuning setups for calibrating LMs.
Structured dropout approaches, such as attention dropout and DropHead, have been investigated to regularize the multi-head attention mechanism in Transformers. In this paper, we propose a new regularization scheme based on token-level rather than structure-level to reduce overfitting. Specifically, we devise a novel Token-Level Masking (TLM) training strategy for Transformers to regularize the connections of self-attention, which consists of two masking techniques that are effective and easy to implement. The underlying idea is to manipulate the connections between tokens in the multi-head attention via masking, where the networks are forced to exploit partial neighbors' information to produce a meaningful representation. The generality and effectiveness of TLM are thoroughly evaluated via extensive experiments on 4 diversified NLP tasks across 18 datasets, including natural language understanding benchmark GLUE, ChineseGLUE, Chinese Grammatical Error Correction, and data-to-text generation. The results indicate that TLM can consistently outperform attention dropout and DropHead, e.g., it increases by 0.5 points relative to DropHead with BERT-large on GLUE. Moreover, TLM can establish a new record on the data-to-text benchmark Rotowire (18.93 BLEU). Our code will be publicly available at https://github.com/Young1993/tlm.
We present ASPIRO, an approach for structured data verbalisation into short template sentences in zero to few-shot settings. Unlike previous methods, our approach prompts large language models (LLMs) to directly produce entity-agnostic templates, rather than relying on LLMs to faithfully copy the given example entities, or validating/crafting the templates manually. We incorporate LLM re-prompting, triggered by algorithmic parsing checks, as well as the PARENT metric induced consistency validation to identify and rectify template generation problems in real-time. ASPIRO, compared to direct LLM output, averages 66\% parsing error rate reduction in generated verbalisations of RDF triples on the DART dataset. Our best 5-shot text-davinci-003 setup, scoring BLEU of 50.62, METEOR of 45.16, BLEURT of 0.82, NUBIA of 0.87, and PARENT of 0.8962 on the Rel2Text dataset, competes effectively with recent fine-tuned pre-trained language models.
Table-to-text systems generate natural language statements from structured data like tables. While end-to-end techniques suffer from low factual correctness (fidelity), a previous study reported gains when using manual logical forms (LF) that represent the selected content and the semantics of the target text. Given the manual step, it was not clear whether automatic LFs would be effective, or whether the improvement came from content selection alone. We present TlT which, given a table and a selection of the content, first produces LFs and then the textual statement. We show for the first time that automatic LFs improve quality, with an increase in fidelity of 30 points over a comparable system not using LFs. Our experiments allow to quantify the remaining challenges for high factual correctness, with automatic selection of content coming first, followed by better Logic-to-Text generation and, to a lesser extent, better Table-to-Logic parsing.
Hallucination of text ungrounded in the input is a well-known problem in neural data-to-text generation. Many methods have been proposed to mitigate it, but they typically require altering model architecture or collecting additional data, and thus cannot be easily applied to an existing model. In this paper, we explore a new way to mitigate hallucinations by combining the probabilistic output of a generator language model (LM) with the output of a special "text critic" classifier, which guides the generation by assessing the match between the input data and the text generated so far. Our method does not need any changes to the underlying LM's architecture or training procedure and can thus be combined with any model and decoding operating on word probabilities. The critic does not need any additional training data, using the base LM's training data and synthetic negative examples. Our experimental results show that our method improves over the baseline on the WebNLG and OpenDialKG benchmarks.
Recent developments in generative AI have shone a spotlight on high-performance synthetic text generation technologies. The now wide availability and ease of use of such models highlights the urgent need to provide equally powerful technologies capable of identifying synthetic text. With this in mind, we draw inspiration from psychological studies which suggest that people can be driven by emotion and encode emotion in the text they compose. We hypothesize that pretrained language models (PLMs) have an affective deficit because they lack such an emotional driver when generating text and consequently may generate synthetic text which has affective incoherence i.e. lacking the kind of emotional coherence present in human-authored text. We subsequently develop an emotionally aware detector by fine-tuning a PLM on emotion. Experiment results indicate that our emotionally-aware detector achieves improvements across a range of synthetic text generators, various sized models, datasets, and domains. Finally, we compare our emotionally-aware synthetic text detector to ChatGPT in the task of identification of its own output and show substantial gains, reinforcing the potential of emotion as a signal to identify synthetic text. Code, models, and datasets are available at https: //github.com/alanagiasi/emoPLMsynth
Learning from free-text human feedback is essential for dialog systems, but annotated data is scarce and usually covers only a small fraction of error types known in conversational AI. Instead of collecting and annotating new datasets from scratch, recent advances in synthetic dialog generation could be used to augment existing dialog datasets with the necessary annotations. However, to assess the feasibility of such an effort, it is important to know the types and frequency of free-text human feedback included in these datasets. In this work, we investigate this question for a variety of commonly used dialog datasets, including MultiWoZ, SGD, BABI, PersonaChat, Wizards-of-Wikipedia, and the human-bot split of the Self-Feeding Chatbot. Using our observations, we derive new taxonomies for the annotation of free-text human feedback in dialogs and investigate the impact of including such data in response generation for three SOTA language generation models, including GPT-2, LLAMA, and Flan-T5. Our findings provide new insights into the composition of the datasets examined, including error types, user response types, and the relations between them.
In the burgeoning field of natural language processing (NLP), Neural Topic Models (NTMs) , Large Language Models (LLMs) and Diffusion model have emerged as areas of significant research interest. Despite this, NTMs primarily utilize contextual embeddings from LLMs, which are not optimal for clustering or capable for topic based text generation. NTMs have never been combined with diffusion model for text generation. Our study addresses these gaps by introducing a novel framework named Diffusion-Enhanced Topic Modeling using Encoder-Decoder-based LLMs (DeTiME). DeTiME leverages Encoder-Decoder-based LLMs to produce highly clusterable embeddings that could generate topics that exhibit both superior clusterability and enhanced semantic coherence compared to existing methods. Additionally, by exploiting the power of diffusion model, our framework also provides the capability to do topic based text generation. This dual functionality allows users to efficiently produce highly clustered topics and topic based text generation simultaneously. DeTiME's potential extends to generating clustered embeddings as well. Notably, our proposed framework(both encoder-decoder based LLM and diffusion model) proves to be efficient to train and exhibits high adaptability to other LLMs and diffusion model, demonstrating its potential for a wide array of applications.
Security classifiers, designed to detect malicious content in computer systems and communications, can underperform when provided with insufficient training data. In the security domain, it is often easy to find samples of the negative (benign) class, and challenging to find enough samples of the positive (malicious) class to train an effective classifier. This study evaluates the application of natural language text generators to fill this data gap in multiple security-related text classification tasks. We describe a variety of previously-unexamined language-model fine-tuning approaches for this purpose and consider in particular the impact of disproportionate class-imbalances in the training set. Across our evaluation using three state-of-the-art classifiers designed for offensive language detection, review fraud detection, and SMS spam detection, we find that models trained with GPT-3 data augmentation strategies outperform both models trained without augmentation and models trained using basic data augmentation strategies already in common usage. In particular, we find substantial benefits for GPT-3 data augmentation strategies in situations with severe limitations on known positive-class samples.
In this paper, we address the hallucination problem commonly found in natural language generation tasks. Language models often generate fluent and convincing content but can lack consistency with the provided source, resulting in potential inaccuracies. We propose a new decoding method called Fidelity-Enriched Contrastive Search (FECS), which augments the contrastive search framework with context-aware regularization terms. FECS promotes tokens that are semantically similar to the provided source while penalizing repetitiveness in the generated text. We demonstrate its effectiveness across two tasks prone to hallucination: abstractive summarization and dialogue generation. Results show that FECS consistently enhances faithfulness across various language model sizes while maintaining output diversity comparable to well-performing decoding algorithms.
The decoding algorithm is critical for open-ended text generation, transforming latent representations into coherent and meaningful outputs. This paper investigates the self-reinforcement effect in text generation and the effectiveness of a repetition penalty to mitigate it. However, determining the optimal repetition penalty value is challenging. To tackle this, we propose a forgetting mechanism that disregards distant tokens, reducing the burden of penalty selection. In addition, we introduce a length penalty to address overly short sentences caused by excessive penalties. Our penalty decoding approach incorporating three strategies helps resolve issues with sampling methods deviating from factual information. Experimental results demonstrate the efficacy of our approach in generating high-quality sentences resembling human output.
Controllable text generation (CTG) aims to generate text with desired attributes, and decoding-time-based methods have shown promising performance on this task. However, in this paper, we identify the phenomenon of Attribute Collapse for the first time. It causes the fluency of generated text to rapidly decrease when the control strength exceeds a critical value, rendering the text completely unusable. This limitation hinders the effectiveness of decoding methods in achieving high levels of controllability. To address this problem, we propose a novel lightweight decoding framework named Air-Decoding. Its main idea is reconstructing the attribute distributions to balance the weights between attribute words and non-attribute words to generate more fluent text. Specifically, we train prefixes by prefix-tuning to obtain attribute distributions. Then we design a novel attribute distribution reconstruction method to balance the obtained distributions and use the reconstructed distributions to guide language models for generation, effectively avoiding the issue of Attribute Collapse. Experiments on multiple CTG tasks prove that our method achieves a new state-of-the-art control performance.