Abstract
Automatically open-ended long text generation poses significant challenges due to semantic incoherence and plot implausibility. Previous works usually alleviate this problem through outlines in the form of short phrases or abstractive signals by designing unsupervised tasks, which tend to be unstable and weakly interpretable. Assuming that a summary serves as a mature outline, we introduce a two-stage, summary-enhanced outline supervised generation framework. This framework leverages the dual characteristics of the summarization task to improve outline prediction, resulting in more explicit and plausible outlines. Furthermore, we identify an underutilization issue in outline-based generation with both standard pretrained language models (e.g., GPT-2, BART) and large language models (e.g., Vicuna, ChatGPT). To address this, we propose a novel explicit outline control method for more effective utilization of generated outlines.
Language Model Self-improvement by Reinforcement Learning Contemplation
Abstract
Large Language Models (LLMs) have exhibited remarkable performance across various natural language processing (NLP) tasks. However, fine-tuning these models often necessitates substantial supervision, which can be expensive and time-consuming to obtain. This paper introduces a novel unsupervised method called LanguageModel Self-Improvement by Reinforcement Learning Contemplation (SIRLC) that improves LLMs without reliance on external labels. Our approach is grounded in the observation that it is simpler for language models to assess text quality than to generate text. Building on this insight, SIRLC assigns LLMs dual roles as both student and teacher. As a student, the LLM generates answers to unlabeled questions, while as a teacher, it evaluates the generated text and assigns scores accordingly. The model parameters are updated using reinforcement learning to maximize the evaluation score. We demonstrate that SIRLC can be applied to various NLP tasks, such as reasoning problems, text generation, and machine translation. Our experiments show that SIRLC effectively improves LLM performance without external supervision, resulting in a 5.6% increase in answering accuracy for reasoning tasks and a rise in BERTScore from 0.82 to 0.86 for translation tasks. Furthermore, SIRLC can be applied to models of different sizes, showcasing its broad applicability.
Meta-review Generation with Checklist-guided Iterative Introspection
Authors: Qi Zeng, Mankeerat Sidhu, Hou Pong Chan, Lu Wang, Heng Ji
Abstract
Opinions in the scientific domain can be divergent, leading to controversy or consensus among reviewers. However, current opinion summarization datasets mostly focus on product review domains, which do not account for this variability under the assumption that the input opinions are non-controversial. To address this gap, we propose the task of scientific opinion summarization, where research paper reviews are synthesized into meta-reviews. To facilitate this task, we introduce a new ORSUM dataset covering 10,989 paper meta-reviews and 40,903 paper reviews from 39 conferences. Furthermore, we propose the Checklist-guided Iterative Introspection (CGI$^2$) approach, which breaks down the task into several stages and iteratively refines the summary under the guidance of questions from a checklist. We conclude that (1) human-written summaries are not always reliable since many do not follow the guideline, and (2) the combination of task decomposition and iterative self-refinement shows promising discussion involvement ability and can be applied to other complex text generation using black-box LLM.
In-Context Demonstration Selection with Cross Entropy Difference
Authors: Dan Iter, Reid Pryzant, Ruochen Xu, Shuohang Wang, Yang Liu, Yichong Xu, Chenguang Zhu
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Large language models (LLMs) can use in-context demonstrations to improve performance on zero-shot tasks. However, selecting the best in-context examples is challenging because model performance can vary widely depending on the selected examples. We present a cross-entropy difference (CED) method for selecting in-context demonstrations. Our method is based on the observation that the effectiveness of in-context demonstrations negatively correlates with the perplexity of the test example by a language model that was finetuned on that demonstration. We utilize parameter efficient finetuning to train small models on training data that are used for computing the cross-entropy difference between a test example and every candidate in-context demonstration. This metric is used to rank and select in-context demonstrations independently for each test input. We evaluate our method on a mix-domain dataset that combines 8 benchmarks, representing 4 text generation tasks, showing that CED for in-context demonstration selection can improve performance for a variety of LLMs.
Faithful Low-Resource Data-to-Text Generation through Cycle Training
Abstract
Methods to generate text from structured data have advanced significantly in recent years, primarily due to fine-tuning of pre-trained language models on large datasets. However, such models can fail to produce output faithful to the input data, particularly on out-of-domain data. Sufficient annotated data is often not available for specific domains, leading us to seek an unsupervised approach to improve the faithfulness of output text. Since the problem is fundamentally one of consistency between the representations of the structured data and text, we evaluate the effectiveness of cycle training in this work. Cycle training uses two models which are inverses of each other: one that generates text from structured data, and one which generates the structured data from natural language text. We show that cycle training, when initialized with a small amount of supervised data (100 samples in our case), achieves nearly the same performance as fully supervised approaches for the data-to-text generation task on the WebNLG, E2E, WTQ, and WSQL datasets. We perform extensive empirical analysis with automated evaluation metrics and a newly designed human evaluation schema to reveal different cycle training strategies' effectiveness of reducing various types of generation errors. Our code is publicly available at https://github.com/Edillower/CycleNLG.
Trade-Offs Between Fairness and Privacy in Language Modeling
Authors: Cleo Matzken, Steffen Eger, Ivan Habernal
Abstract
Protecting privacy in contemporary NLP models is gaining in importance. So does the need to mitigate social biases of such models. But can we have both at the same time? Existing research suggests that privacy preservation comes at the price of worsening biases in classification tasks. In this paper, we explore the extent to which this tradeoff really holds when we incorporate both privacy preservation and de-biasing techniques into training text generation models. How does improving the model along one dimension affect the other dimension as well as the utility of the model? We conduct an extensive set of experiments that include bias detection, privacy attacks, language modeling, and performance on downstream tasks.
Large Language Models are Effective Table-to-Text Generators, Evaluators, and Feedback Providers
Abstract
Large language models (LLMs) have shown remarkable ability on controllable text generation. However, the potential of LLMs in generating text from structured tables remains largely under-explored. In this paper, we study the capabilities of LLMs for table-to-text generation tasks, particularly aiming to investigate their performance in generating natural language statements that can be logically entailed by a provided table. First, we investigate how LLMs compare to state-of-the-art table-to-text fine-tuned models, and demonstrate that LLMs can generate statements with higher faithfulness compared with previous state-of-the-art fine-tuned models. Given this finding, we next explore whether LLMs can serve as faithfulness-level automated evaluation metrics. Through human evaluation, we show that evaluation metrics adopted from LLMs correlates better with human judgments compared with existing faithfulness-level metrics. Finally, we demonstrate that LLMs using chain-of-thought prompting can generate high-fidelity natural language feedback for other table-to-text models' generations, provide insights for future work regarding the distillation of text generation capabilities from LLMs to smaller models.
MuLER: Detailed and Scalable Reference-based Evaluation
Authors: Taelin Karidi, Leshem Choshen, Gal Patel, Omri Abend
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
We propose a novel methodology (namely, MuLER) that transforms any reference-based evaluation metric for text generation, such as machine translation (MT) into a fine-grained analysis tool. Given a system and a metric, MuLER quantifies how much the chosen metric penalizes specific error types (e.g., errors in translating names of locations). MuLER thus enables a detailed error analysis which can lead to targeted improvement efforts for specific phenomena. We perform experiments in both synthetic and naturalistic settings to support MuLER's validity and showcase its usability in MT evaluation, and other tasks, such as summarization. Analyzing all submissions to WMT in 2014-2020, we find consistent trends. For example, nouns and verbs are among the most frequent POS tags. However, they are among the hardest to translate. Performance on most POS tags improves with overall system performance, but a few are not thus correlated (their identity changes from language to language). Preliminary experiments with summarization reveal similar trends.
Abstract
The field of text generation suffers from a severe shortage of labeled data due to the extremely expensive and time consuming process involved in manual annotation. A natural approach for coping with this problem is active learning (AL), a well-known machine learning technique for improving annotation efficiency by selectively choosing the most informative examples to label. However, while AL has been well-researched in the context of text classification, its application to text generation remained largely unexplored. In this paper, we present a first systematic study of active learning for text generation, considering a diverse set of tasks and multiple leading AL strategies. Our results indicate that existing AL strategies, despite their success in classification, are largely ineffective for the text generation scenario, and fail to consistently surpass the baseline of random example selection. We highlight some notable differences between the classification and generation scenarios, and analyze the selection behaviors of existing AL strategies. Our findings motivate exploring novel approaches for applying AL to NLG tasks.
Inference-Time Policy Adapters (IPA): Tailoring Extreme-Scale LMs without Fine-tuning
Abstract
Large language models excel at a variety of language tasks when prompted with examples or instructions. Yet controlling these models through prompting alone is limited. Tailoring language models through fine-tuning (e.g., via reinforcement learning) can be effective, but it is expensive and requires model access. We propose Inference-time Policy Adapters (IPA), which efficiently tailors a language model such as GPT-3 without fine-tuning it. IPA guides a large base model during decoding time through a lightweight policy adaptor trained to optimize an arbitrary user objective with reinforcement learning. On five challenging text generation tasks, such as toxicity reduction and open-domain generation, IPA consistently brings significant improvements over off-the-shelf language models. It outperforms competitive baseline methods, sometimes even including expensive fine-tuning. In particular, tailoring GPT-2 with IPA can outperform GPT-3, while tailoring GPT- 3 with IPA brings a major performance boost over GPT-3 (and sometimes even over GPT-4). Our promising results highlight the potential of IPA as a lightweight alternative to tailoring extreme-scale language models.
Peek Across: Improving Multi-Document Modeling via Cross-Document Question-Answering
Authors: Avi Caciularu, Matthew E. Peters, Jacob Goldberger, Ido Dagan, Arman Cohan
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
The integration of multi-document pre-training objectives into language models has resulted in remarkable improvements in multi-document downstream tasks. In this work, we propose extending this idea by pre-training a generic multi-document model from a novel cross-document question answering pre-training objective. To that end, given a set (or cluster) of topically-related documents, we systematically generate semantically-oriented questions from a salient sentence in one document and challenge the model, during pre-training, to answer these questions while "peeking" into other topically-related documents. In a similar manner, the model is also challenged to recover the sentence from which the question was generated, again while leveraging cross-document information. This novel multi-document QA formulation directs the model to better recover cross-text informational relations, and introduces a natural augmentation that artificially increases the pre-training data. Further, unlike prior multi-document models that focus on either classification or summarization tasks, our pre-training objective formulation enables the model to perform tasks that involve both short text generation (e.g., QA) and long text generation (e.g., summarization). Following this scheme, we pre-train our model -- termed QAmden -- and evaluate its performance across several multi-document tasks, including multi-document QA, summarization, and query-focused summarization, yielding improvements of up to 7%, and significantly outperforms zero-shot GPT-3.5 and GPT-4.
Keyword: machine translation
Language Model Self-improvement by Reinforcement Learning Contemplation
Abstract
Large Language Models (LLMs) have exhibited remarkable performance across various natural language processing (NLP) tasks. However, fine-tuning these models often necessitates substantial supervision, which can be expensive and time-consuming to obtain. This paper introduces a novel unsupervised method called LanguageModel Self-Improvement by Reinforcement Learning Contemplation (SIRLC) that improves LLMs without reliance on external labels. Our approach is grounded in the observation that it is simpler for language models to assess text quality than to generate text. Building on this insight, SIRLC assigns LLMs dual roles as both student and teacher. As a student, the LLM generates answers to unlabeled questions, while as a teacher, it evaluates the generated text and assigns scores accordingly. The model parameters are updated using reinforcement learning to maximize the evaluation score. We demonstrate that SIRLC can be applied to various NLP tasks, such as reasoning problems, text generation, and machine translation. Our experiments show that SIRLC effectively improves LLM performance without external supervision, resulting in a 5.6% increase in answering accuracy for reasoning tasks and a rise in BERTScore from 0.82 to 0.86 for translation tasks. Furthermore, SIRLC can be applied to models of different sizes, showcasing its broad applicability.
CMOT: Cross-modal Mixup via Optimal Transport for Speech Translation
Authors: Yan Zhou, Qingkai Fang, Yang Feng
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD); Audio and Speech Processing (eess.AS)
Abstract
End-to-end speech translation (ST) is the task of translating speech signals in the source language into text in the target language. As a cross-modal task, end-to-end ST is difficult to train with limited data. Existing methods often try to transfer knowledge from machine translation (MT), but their performances are restricted by the modality gap between speech and text. In this paper, we propose Cross-modal Mixup via Optimal Transport CMOT to overcome the modality gap. We find the alignment between speech and text sequences via optimal transport and then mix up the sequences from different modalities at a token level using the alignment. Experiments on the MuST-C ST benchmark demonstrate that CMOT achieves an average BLEU of 30.0 in 8 translation directions, outperforming previous methods. Further analysis shows CMOT can adaptively find the alignment between modalities, which helps alleviate the modality gap between speech and text. Code is publicly available at https://github.com/ictnlp/CMOT.
Abstract
This survey paper provides a comprehensive review of the use of diffusion models in natural language processing (NLP). Diffusion models are a class of mathematical models that aim to capture the diffusion of information or signals across a network or manifold. In NLP, diffusion models have been used in a variety of applications, such as natural language generation, sentiment analysis, topic modeling, and machine translation. This paper discusses the different formulations of diffusion models used in NLP, their strengths and limitations, and their applications. We also perform a thorough comparison between diffusion models and alternative generative models, specifically highlighting the autoregressive (AR) models, while also examining how diverse architectures incorporate the Transformer in conjunction with diffusion models. Compared to AR models, diffusion models have significant advantages for parallel generation, text interpolation, token-level controls such as syntactic structures and semantic contents, and robustness. Exploring further permutations of integrating Transformers into diffusion models would be a valuable pursuit. Also, the development of multimodal diffusion models and large-scale diffusion language models with notable capabilities for few-shot learning would be important directions for the future advance of diffusion models in NLP.
Leveraging GPT-4 for Automatic Translation Post-Editing
Authors: Vikas Raunak, Amr Sharaf, Hany Hassan Awadallah, Arul Menezes
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
While Neural Machine Translation (NMT) represents the leading approach to Machine Translation (MT), the outputs of NMT models still require translation post-editing to rectify errors and enhance quality, particularly under critical settings. In this work, we formalize the task of translation post-editing with Large Language Models (LLMs) and explore the use of GPT-4 to automatically post-edit NMT outputs across several language pairs. Our results demonstrate that GPT-4 is adept at translation post-editing and produces meaningful edits even when the target language is not English. Notably, we achieve state-of-the-art performance on WMT-22 English-Chinese, English-German, Chinese-English and German-English language pairs using GPT-4 based post-editing, as evaluated by state-of-the-art MT quality metrics.
Dolphin: A Challenging and Diverse Benchmark for Arabic NLG
Authors: El Moatez Billah Nagoudi, Ahmed El-Shangiti, AbdelRahim Elmadany, Muhammad Abdul-Mageed
Abstract
We present Dolphin, a novel benchmark that addresses the need for an evaluation framework for the wide collection of Arabic languages and varieties. The proposed benchmark encompasses a broad range of 13 different NLG tasks, including text summarization, machine translation, question answering, and dialogue generation, among others. Dolphin comprises a substantial corpus of 40 diverse and representative public datasets across 50 test splits, carefully curated to reflect real-world scenarios and the linguistic richness of Arabic. It sets a new standard for evaluating the performance and generalization capabilities of Arabic and multilingual models, promising to enable researchers to push the boundaries of current methodologies. We provide an extensive analysis of Dolphin, highlighting its diversity and identifying gaps in current Arabic NLG research. We also evaluate several Arabic and multilingual models on our benchmark, allowing us to set strong baselines against which researchers can compare.
MuLER: Detailed and Scalable Reference-based Evaluation
Authors: Taelin Karidi, Leshem Choshen, Gal Patel, Omri Abend
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
We propose a novel methodology (namely, MuLER) that transforms any reference-based evaluation metric for text generation, such as machine translation (MT) into a fine-grained analysis tool. Given a system and a metric, MuLER quantifies how much the chosen metric penalizes specific error types (e.g., errors in translating names of locations). MuLER thus enables a detailed error analysis which can lead to targeted improvement efforts for specific phenomena. We perform experiments in both synthetic and naturalistic settings to support MuLER's validity and showcase its usability in MT evaluation, and other tasks, such as summarization. Analyzing all submissions to WMT in 2014-2020, we find consistent trends. For example, nouns and verbs are among the most frequent POS tags. However, they are among the hardest to translate. Performance on most POS tags improves with overall system performance, but a few are not thus correlated (their identity changes from language to language). Preliminary experiments with summarization reveal similar trends.
Not All Metrics Are Guilty: Improving NLG Evaluation with LLM Paraphrasing
Abstract
Most research about natural language generation (NLG) relies on evaluation benchmarks with limited references for a sample, which may result in poor correlations with human judgements. The underlying reason is that one semantic meaning can actually be expressed in different forms, and the evaluation with a single or few references may not accurately reflect the quality of the model's hypotheses. To address this issue, this paper presents a novel method, named Para-Ref, to enhance existing evaluation benchmarks by enriching the number of references. We leverage large language models (LLMs) to paraphrase a single reference into multiple high-quality ones in diverse expressions. Experimental results on representative NLG tasks of machine translation, text summarization, and image caption demonstrate that our method can effectively improve the correlation with human evaluation for sixteen automatic evaluation metrics by +7.82% in ratio. We release the code and data at https://github.com/RUCAIBox/Para-Ref.
Unit-based Speech-to-Speech Translation Without Parallel Data
Authors: Anuj Diwan, Anirudh Srinivasan, David Harwath, Eunsol Choi
Subjects: Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
Abstract
We propose an unsupervised speech-to-speech translation (S2ST) system that does not rely on parallel data between the source and target languages. Our approach maps source and target language speech signals into automatically discovered, discrete units and reformulates the problem as unsupervised unit-to-unit machine translation. We develop a three-step training procedure that involves (a) pre-training an unit-based encoder-decoder language model with a denoising objective (b) training it with word-by-word translated utterance pairs created by aligning monolingual text embedding spaces and (c) running unsupervised backtranslation bootstrapping off of the initial translation model. Our approach avoids mapping the speech signal into text and uses speech-to-unit and unit-to-speech models instead of automatic speech recognition and text to speech models. We evaluate our model on synthetic-speaker Europarl-ST English-German and German-English evaluation sets, finding that unit-based translation is feasible under this constrained scenario, achieving 9.29 ASR-BLEU in German to English and 8.07 in English to German.
Keyword: non-autoregressive
NAIL: Lexical Retrieval Indices with Efficient Non-Autoregressive Decoders
Authors: Livio Baldini Soares, Daniel Gillick, Jeremy R. Cole, Tom Kwiatkowski
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR)
Abstract
Neural document rerankers are extremely effective in terms of accuracy. However, the best models require dedicated hardware for serving, which is costly and often not feasible. To avoid this serving-time requirement, we present a method of capturing up to 86% of the gains of a Transformer cross-attention model with a lexicalized scoring function that only requires 10-6% of the Transformer's FLOPs per document and can be served using commodity CPUs. When combined with a BM25 retriever, this approach matches the quality of a state-of-the art dual encoder retriever, that still requires an accelerator for query encoding. We introduce NAIL (Non-Autoregressive Indexing with Language models) as a model architecture that is compatible with recent encoder-decoder and decoder-only large language models, such as T5, GPT-3 and PaLM. This model architecture can leverage existing pre-trained checkpoints and can be fine-tuned for efficiently constructing document representations that do not require neural processing of queries.
Keyword: abstractive summarization
Improving Factuality of Abstractive Summarization without Sacrificing Summary Quality
Authors: Tanay Dixit, Fei Wang, Muhao Chen
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Improving factual consistency of abstractive summarization has been a widely studied topic. However, most of the prior works on training factuality-aware models have ignored the negative effect it has on summary quality. We propose EFACTSUM (i.e., Effective Factual Summarization), a candidate summary generation and ranking technique to improve summary factuality without sacrificing summary quality. We show that using a contrastive learning framework with our refined candidate summaries leads to significant gains on both factuality and similarity-based metrics. Specifically, we propose a ranking strategy in which we effectively combine two metrics, thereby preventing any conflict during training. Models trained using our approach show up to 6 points of absolute improvement over the base model with respect to FactCC on XSUM and 11 points on CNN/DM, without negatively affecting either similarity-based metrics or absractiveness.
Keyword: factual
Exploring Contrast Consistency of Open-Domain Question Answering Systems on Minimally Edited Questions
Abstract
Contrast consistency, the ability of a model to make consistently correct predictions in the presence of perturbations, is an essential aspect in NLP. While studied in tasks such as sentiment analysis and reading comprehension, it remains unexplored in open-domain question answering (OpenQA) due to the difficulty of collecting perturbed questions that satisfy factuality requirements. In this work, we collect minimally edited questions as challenging contrast sets to evaluate OpenQA models. Our collection approach combines both human annotation and large language model generation. We find that the widely used dense passage retriever (DPR) performs poorly on our contrast sets, despite fitting the training set well and performing competitively on standard test sets. To address this issue, we introduce a simple and effective query-side contrastive loss with the aid of data augmentation to improve DPR training. Our experiments on the contrast sets demonstrate that DPR's contrast consistency is improved without sacrificing its accuracy on the standard test sets.
LLMs as Factual Reasoners: Insights from Existing Benchmarks and Beyond
Authors: Philippe Laban, Wojciech Kryściński, Divyansh Agarwal, Alexander R. Fabbri, Caiming Xiong, Shafiq Joty, Chien-Sheng Wu
Abstract
With the recent appearance of LLMs in practical settings, having methods that can effectively detect factual inconsistencies is crucial to reduce the propagation of misinformation and improve trust in model outputs. When testing on existing factual consistency benchmarks, we find that a few large language models (LLMs) perform competitively on classification benchmarks for factual inconsistency detection compared to traditional non-LLM methods. However, a closer analysis reveals that most LLMs fail on more complex formulations of the task and exposes issues with existing evaluation benchmarks, affecting evaluation precision. To address this, we propose a new protocol for inconsistency detection benchmark creation and implement it in a 10-domain benchmark called SummEdits. This new benchmark is 20 times more cost-effective per sample than previous benchmarks and highly reproducible, as we estimate inter-annotator agreement at about 0.9. Most LLMs struggle on SummEdits, with performance close to random chance. The best-performing model, GPT-4, is still 8\% below estimated human performance, highlighting the gaps in LLMs' ability to reason about facts and detect inconsistencies when they occur.
Interpretable Automatic Fine-grained Inconsistency Detection in Text Summarization
Abstract
Existing factual consistency evaluation approaches for text summarization provide binary predictions and limited insights into the weakness of summarization systems. Therefore, we propose the task of fine-grained inconsistency detection, the goal of which is to predict the fine-grained types of factual errors in a summary. Motivated by how humans inspect factual inconsistency in summaries, we propose an interpretable fine-grained inconsistency detection model, FineGrainFact, which explicitly represents the facts in the documents and summaries with semantic frames extracted by semantic role labeling, and highlights the related semantic frames to predict inconsistency. The highlighted semantic frames help verify predicted error types and correct inconsistent summaries. Experiment results demonstrate that our model outperforms strong baselines and provides evidence to support or refute the summary.
Self-Checker: Plug-and-Play Modules for Fact-Checking with Large Language Models
Abstract
Fact-checking is an essential task in NLP that is commonly utilized for validating the factual accuracy of claims. Prior work has mainly focused on fine-tuning pre-trained languages models on specific datasets, which can be computationally intensive and time-consuming. With the rapid development of large language models (LLMs), such as ChatGPT and GPT-3, researchers are now exploring their in-context learning capabilities for a wide range of tasks. In this paper, we aim to assess the capacity of LLMs for fact-checking by introducing Self-Checker, a framework comprising a set of plug-and-play modules that facilitate fact-checking by purely prompting LLMs in an almost zero-shot setting. This framework provides a fast and efficient way to construct fact-checking systems in low-resource environments. Empirical results demonstrate the potential of Self-Checker in utilizing LLMs for fact-checking. However, there is still significant room for improvement compared to SOTA fine-tuned models, which suggests that LLM adoption could be a promising approach for future fact-checking research.
Enabling Large Language Models to Generate Text with Citations
Authors: Tianyu Gao, Howard Yen, Jiatong Yu, Danqi Chen
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Abstract
Large language models (LLMs) have emerged as a widely-used tool for information seeking, but their generated outputs are prone to hallucination. In this work, we aim to enable LLMs to generate text with citations, improving their factual correctness and verifiability. Existing work mainly relies on commercial search engines and human evaluation, making it challenging to reproduce and compare with different modeling approaches. We propose ALCE, the first benchmark for Automatic LLMs' Citation Evaluation. ALCE collects a diverse set of questions and retrieval corpora and requires building end-to-end systems to retrieve supporting evidence and generate answers with citations. We build automatic metrics along three dimensions -- fluency, correctness, and citation quality -- and demonstrate their strong correlation with human judgements. Our experiments with state-of-the-art LLMs and novel prompting strategies show that current systems have considerable room for improvements -- for example, on the ELI5 dataset, even the best model has 49% of its generations lacking complete citation support. Our extensive analyses further highlight promising future directions, including developing better retrievers, advancing long-context LLMs, and improving the ability to synthesize information from multiple sources.
Mixture of Prompt Experts for Generalizable and Interpretable Question Answering
Authors: Chenglei Si, Weijia Shi, Chen Zhao, Luke Zettlemoyer, Jordan Boyd-Graber
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
One of the ultimate quests of question answering (QA) is to deploy a system that can answer any type of question from the users, and refrain from answering when it does not know the answer. While recent advancements in scaling large language models (LLMs) brought significant improvements on various QA datasets, it remains difficult for a single model to generalize across question types that require distinct reasoning abilities. In this paper, we first provide empirical evidence that state-of-the-art LLMs such as Codex suffer from poor generalizability on question types beyond those seen in the prompt. To address this, we propose a Mixture-of-Prompt-Experts (MOPE) system that ensembles multiple specialized LLMs. We first implement each specialized model based on the same backbone model (Codex) but with prompts optimized for different reasoning categories including factual, multihop, mathematical, and commonsense reasoning. By strategically selecting the best specialized model for each given question, our MOPE system significantly outperforms any single specialized model on a collection of 12 QA datasets from four reasoning types. Moreover, the attribution and agreement among specialized expert models offer greater interpretability, allowing for better selective question answering. Our human study further confirms that presenting the expert predictions and answer selection process helps annotators more accurately decide when to trust the system's output. We release all code and data to facilitate future work.
Analyzing Influential Factors in Human Preference Judgments via GPT-4
Abstract
Pairwise human judgments are pivotal in guiding large language models (LLMs) to generate outputs that align with human preferences. They are also often used in summarization evaluation, complementing existing automatic metrics. Despite their significance, however, there has been limited research probing these pairwise human judgments. The collective impact and respective weights of factors such as informativeness, coherence, fluency, and factual consistency remain elusive. The impact of hidden factors on the final judgment is also unclear. In this paper, we conduct an in-depth examination of a dataset of pairwise human judgments released by OpenAI. Utilizing the Bradley-Terry-Luce model, we identify key factors that could potentially influence human judgments. Our research uncovers the inherent preferences embedded in human judgments and suggests strategies to boost sample efficiency. Finally, we provide insights on the construction of balanced datasets for human judgment evaluations, a crucial step in shaping the behaviors of future LLMs.
The student becomes the master: Matching GPT3 on Scientific Factual Error Correction
Authors: Dhananjay Ashok, Atharva Kulkarni, Hai Pham, Barnabás Póczos
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Due to the prohibitively high cost of creating error correction datasets, most Factual Claim Correction methods rely on a powerful verification model to guide the correction process. This leads to a significant drop in performance in domains like Scientific Claim Correction, where good verification models do not always exist. In this work, we introduce a claim correction system that makes no domain assumptions and does not require a verifier but is able to outperform existing methods by an order of magnitude -- achieving 94% correction accuracy on the SciFact dataset, and 62.5% on the SciFact-Open dataset, compared to the next best methods 0.5% and 1.50% respectively. Our method leverages the power of prompting with LLMs during training to create a richly annotated dataset that can be used for fully supervised training and regularization. We additionally use a claim-aware decoding procedure to improve the quality of corrected claims. Our method is competitive with the very LLM that was used to generate the annotated dataset -- with GPT3.5 achieving 89.5% and 60% correction accuracy on SciFact and SciFact-Open, despite using 1250 times as many parameters as our model.
Trusting Your Evidence: Hallucinate Less with Context-aware Decoding
Authors: Weijia Shi, Xiaochuang Han, Mike Lewis, Yulia Tsvetkov, Luke Zettlemoyer, Scott Wen-tau Yih
Abstract
Language models (LMs) often struggle to pay enough attention to the input context, and generate texts that are unfaithful or contain hallucinations. To mitigate this issue, we present context-aware decoding (CAD), which follows a contrastive output distribution that amplifies the difference between the output probabilities when a model is used with and without context. Our experiments show that CAD, without additional training, significantly improves the faithfulness of different LM families, including OPT, GPT, LLaMA and FLAN-T5 for summarization tasks (e.g., 14.3% gain for LLaMA in factuality metrics). Furthermore, CAD is particularly effective in overriding a model's prior knowledge when it contradicts the provided context, leading to substantial improvements in tasks where resolving the knowledge conflict is essential.
Measuring the Knowledge Acquisition-Utilization Gap in Pretrained Language Models
Authors: Amirhossein Kazemnejad, Mehdi Rezagholizadeh, Prasanna Parthasarathi, Sarath Chandar
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
While pre-trained language models (PLMs) have shown evidence of acquiring vast amounts of knowledge, it remains unclear how much of this parametric knowledge is actually usable in performing downstream tasks. We propose a systematic framework to measure parametric knowledge utilization in PLMs. Our framework first extracts knowledge from a PLM's parameters and subsequently constructs a downstream task around this extracted knowledge. Performance on this task thus depends exclusively on utilizing the model's possessed knowledge, avoiding confounding factors like insufficient signal. As an instantiation, we study factual knowledge of PLMs and measure utilization across 125M to 13B parameter PLMs. We observe that: (1) PLMs exhibit two gaps - in acquired vs. utilized knowledge, (2) they show limited robustness in utilizing knowledge under distribution shifts, and (3) larger models close the acquired knowledge gap but the utilized knowledge gap remains. Overall, our study provides insights into PLMs' capabilities beyond their acquired knowledge.
Large Language Models as Counterfactual Generator: Strengths and Weaknesses
Abstract
Large language models (LLMs) have demonstrated remarkable performance in a range of natural language understanding and generation tasks. Yet, their ability to generate counterfactuals, which can be used for areas like data augmentation, remains under-explored. This study aims to investigate the counterfactual generation capabilities of LLMs and analysis factors that influence this ability. First, we evaluate how effective are LLMs in counterfactual generation through data augmentation experiments for small language models (SLMs) across four tasks: sentiment analysis, natural language inference, named entity recognition, and relation extraction. While LLMs show promising enhancements in various settings, they struggle in complex tasks due to their self-limitations and the lack of logical guidance to produce counterfactuals that align with commonsense. Second, our analysis reveals the pivotal role of providing accurate task definitions and detailed step-by-step instructions to LLMs in generating counterfactuals. Interestingly, we also find that LLMs can generate reasonable counterfactuals even with unreasonable demonstrations, which illustrates that demonstrations are primarily to regulate the output format.This study provides the first comprehensive insight into counterfactual generation abilities of LLMs, and offers a novel perspective on utilizing LLMs for data augmentation to enhance SLMs.
Privacy Implications of Retrieval-Based Language Models
Abstract
Retrieval-based language models (LMs) have demonstrated improved interpretability, factuality, and adaptability compared to their parametric counterparts, by incorporating retrieved text from external datastores. While it is well known that parametric models are prone to leaking private data, it remains unclear how the addition of a retrieval datastore impacts model privacy. In this work, we present the first study of privacy risks in retrieval-based LMs, particularly $k$NN-LMs. Our goal is to explore the optimal design and training procedure in domains where privacy is of concern, aiming to strike a balance between utility and privacy. Crucially, we find that $k$NN-LMs are more susceptible to leaking private information from their private datastore than parametric models. We further explore mitigations of privacy risks. When privacy information is targeted and readily detected in the text, we find that a simple sanitization step would completely eliminate the risks, while decoupling query and key encoders achieves an even better utility-privacy trade-off. Otherwise, we consider strategies of mixing public and private data in both datastore and encoder training. While these methods offer modest improvements, they leave considerable room for future work. Together, our findings provide insights for practitioners to better understand and mitigate privacy risks in retrieval-based LMs. Our code is available at: https://github.com/Princeton-SysML/kNNLM_privacy .
Getting Sick After Seeing a Doctor? Diagnosing and Mitigating Knowledge Conflicts in Event Temporal Reasoning
Abstract
Event temporal reasoning aims at identifying the temporal relations between two or more events. However, knowledge conflicts arise when there is a mismatch between the actual temporal relations of events in the context and the prior knowledge or biases learned by the model. We first systematically define distinct kinds of bias in event temporal reasoning, which include event relation prior bias, tense bias, narrative bias, and dependency bias, as indicators to study knowledge conflicts. To mitigate such event-related knowledge conflict, we introduce a Counterfactual Data Augmentation based method that can be applied to both Pre-trained Language Models (PLMs) and Large Language Models (LLMs) either as additional training data or demonstrations for In-Context Learning. Experiments suggest the importance of mitigating knowledge conflicts in event temporal reasoning tasks for reducing hallucination and highlight the potential of counterfactual data augmentation for improving model performance.
Improving Factuality of Abstractive Summarization without Sacrificing Summary Quality
Authors: Tanay Dixit, Fei Wang, Muhao Chen
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Improving factual consistency of abstractive summarization has been a widely studied topic. However, most of the prior works on training factuality-aware models have ignored the negative effect it has on summary quality. We propose EFACTSUM (i.e., Effective Factual Summarization), a candidate summary generation and ranking technique to improve summary factuality without sacrificing summary quality. We show that using a contrastive learning framework with our refined candidate summaries leads to significant gains on both factuality and similarity-based metrics. Specifically, we propose a ranking strategy in which we effectively combine two metrics, thereby preventing any conflict during training. Models trained using our approach show up to 6 points of absolute improvement over the base model with respect to FactCC on XSUM and 11 points on CNN/DM, without negatively affecting either similarity-based metrics or absractiveness.
RefGPT: Reference -> Truthful & Customized Dialogues Generation by GPTs and for GPTs
Abstract
General chat models, like ChatGPT, have attained impressive capability to resolve a wide range of NLP tasks by tuning Large Language Models (LLMs) with high-quality instruction data. However, collecting human-written high-quality data, especially multi-turn dialogues, is expensive and unattainable for most people. Though previous studies have used powerful LLMs to generate the dialogues automatically, but they all suffer from generating untruthful dialogues because of the LLMs hallucination. Therefore, we propose a method called RefGPT to generate enormous truthful and customized dialogues without worrying about factual errors caused by the model hallucination. RefGPT solves the model hallucination in dialogue generation by restricting the LLMs to leverage the given reference instead of reciting their own knowledge to generate dialogues. Additionally, RefGPT adds detailed controls on every utterances to enable highly customization capability, which previous studies have ignored. On the basis of RefGPT, we also propose two high-quality dialogue datasets generated by GPT-4, namely RefGPT-Fact and RefGPT-Code. RefGPT-Fact is 100k multi-turn dialogue datasets based on factual knowledge and RefGPT-Code is 76k multi-turn dialogue dataset covering a wide range of coding scenarios. Our code and datasets are released in https://github.com/ziliwangnlp/RefGPT
Understanding Arithmetic Reasoning in Language Models using Causal Mediation Analysis
Abstract
Mathematical reasoning in large language models (LLMs) has garnered attention in recent research, but there is limited understanding of how these models process and store information related to arithmetic tasks. In this paper, we present a mechanistic interpretation of LLMs for arithmetic-based questions using a causal mediation analysis framework. By intervening on the activations of specific model components and measuring the resulting changes in predicted probabilities, we identify the subset of parameters responsible for specific predictions. We analyze two pre-trained language models with different sizes (2.8B and 6B parameters). Experimental results reveal that a small set of mid-late layers significantly affect predictions for arithmetic-based questions, with distinct activation patterns for correct and wrong predictions. We also investigate the role of the attention mechanism and compare the model's activation patterns for arithmetic queries with the prediction of factual knowledge. Our findings provide insights into the mechanistic interpretation of LLMs for arithmetic tasks and highlight the specific components involved in arithmetic reasoning.
Meta-Learning Online Adaptation of Language Models
Authors: Nathan Hu, Eric Mitchell, Christopher D. Manning, Chelsea Finn
Abstract
Large language models encode surprisingly broad knowledge about the world into their parameters. However, the knowledge in static language models can fall out of date, limiting the model's effective "shelf life." While online fine-tuning can reduce this degradation, we find that fine-tuning on a stream of documents using standard optimizers such as Adam leads to a disappointingly low level of information uptake. We hypothesize that online fine-tuning does not sufficiently 'attend' to important information. That is, the gradient signal from important tokens representing factual information is drowned out by the gradient from inherently noisy tokens, suggesting a dynamic, context-aware learning rate may be beneficial. To test this hypothesis, we meta-train a small, autoregressive model to reweight the language modeling loss for each token during online fine-tuning, with the objective of maximizing the out-of-date base language model's ability to answer questions about a document after a single weighted gradient step. We call this approach Context-aware Meta-learned Loss Scaling (CaMeLS). Across three different distributions of documents, our experiments find that fine-tuning on streams of thousands of documents with CaMeLS substantially improves knowledge retention compared to standard online fine-tuning. Finally, we find that the meta-learned weights are general, and that a single reweighting model can be used to enhance the online adaptation of many LMs.
ASPER: Answer Set Programming Enhanced Neural Network Models for Joint Entity-Relation Extraction
Authors: Trung Hoang Le, Huiping Cao, Tran Cao Son
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
A plethora of approaches have been proposed for joint entity-relation (ER) extraction. Most of these methods largely depend on a large amount of manually annotated training data. However, manual data annotation is time consuming, labor intensive, and error prone. Human beings learn using both data (through induction) and knowledge (through deduction). Answer Set Programming (ASP) has been a widely utilized approach for knowledge representation and reasoning that is elaboration tolerant and adept at reasoning with incomplete information. This paper proposes a new approach, ASP-enhanced Entity-Relation extraction (ASPER), to jointly recognize entities and relations by learning from both data and domain knowledge. In particular, ASPER takes advantage of the factual knowledge (represented as facts in ASP) and derived knowledge (represented as rules in ASP) in the learning process of neural network models. We have conducted experiments on two real datasets and compare our method with three baselines. The results show that our ASPER model consistently outperforms the baselines.
Keyword: knowledge distillation
Large Language Model Distillation Doesn't Need a Teacher
Authors: Ananya Harsh Jha, Dirk Groeneveld, Emma Strubell, Iz Beltagy
Abstract
Knowledge distillation trains a smaller student model to match the output distribution of a larger teacher to maximize the end-task performance under computational constraints. However, existing literature on language model distillation primarily focuses on compressing encoder-only models that are then specialized by task-specific supervised finetuning. We need to rethink this setup for more recent large language models with tens to hundreds of billions of parameters. Task-specific finetuning is impractical at this scale, and model performance is often measured using zero/few-shot prompting. Thus, in this work, we advocate for task-agnostic zero-shot evaluated distillation for large language models without access to end-task finetuning data. We propose a teacher-free task-agnostic distillation method, which uses a truncated version of the larger model for initialization, and continues pretraining this model using a language modeling objective. Our teacher-free method shines in a distillation regime where it is infeasible to fit both the student and teacher into the GPU memory. Despite its simplicity, our method can effectively reduce the model size by 50\%, matching or outperforming the vanilla distillation method on perplexity and accuracy on 13 zero-shot end-tasks while being 1.5x computationally efficient.
CoLaDa: A Collaborative Label Denoising Framework for Cross-lingual Named Entity Recognition
Authors: Tingting Ma, Qianhui Wu, Huiqiang Jiang, Börje F. Karlsson, Tiejun Zhao, Chin-Yew Lin
Abstract
Cross-lingual named entity recognition (NER) aims to train an NER system that generalizes well to a target language by leveraging labeled data in a given source language. Previous work alleviates the data scarcity problem by translating source-language labeled data or performing knowledge distillation on target-language unlabeled data. However, these methods may suffer from label noise due to the automatic labeling process. In this paper, we propose CoLaDa, a Collaborative Label Denoising Framework, to address this problem. Specifically, we first explore a model-collaboration-based denoising scheme that enables models trained on different data sources to collaboratively denoise pseudo labels used by each other. We then present an instance-collaboration-based strategy that considers the label consistency of each token's neighborhood in the representation space for denoising. Experiments on different benchmark datasets show that the proposed CoLaDa achieves superior results compared to previous methods, especially when generalizing to distant languages.
How to Distill your BERT: An Empirical Study on the Impact of Weight Initialisation and Distillation Objectives
Authors: Xinpeng Wang, Leonie Weissweiler, Hinrich Schütze, Barbara Plank
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Recently, various intermediate layer distillation (ILD) objectives have been shown to improve compression of BERT models via Knowledge Distillation (KD). However, a comprehensive evaluation of the objectives in both task-specific and task-agnostic settings is lacking. To the best of our knowledge, this is the first work comprehensively evaluating distillation objectives in both settings. We show that attention transfer gives the best performance overall. We also study the impact of layer choice when initializing the student from the teacher layers, finding a significant impact on the performance in task-specific distillation. For vanilla KD and hidden states transfer, initialisation with lower layers of the teacher gives a considerable improvement over higher layers, especially on the task of QNLI (up to an absolute percentage change of 17.8 in accuracy). Attention transfer behaves consistently under different initialisation settings. We release our code as an efficient transformer-based model distillation framework for further studies.
Keyword: Hallucination
Sources of Hallucination by Large Language Models on Inference Tasks
Authors: Nick McKenna, Tianyi Li, Liang Cheng, Mohammad Javad Hosseini, Mark Johnson, Mark Steedman
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Large Language Models (LLMs) are claimed to be capable of Natural Language Inference (NLI), necessary for applied tasks like question answering and summarization, yet this capability is under-explored. We present a series of behavioral studies on several LLM families (LLaMA, GPT-3.5, and PaLM) which probe their behavior using controlled experiments. We establish two factors which predict much of their performance, and propose that these are major sources of hallucination in generative LLM. First, the most influential factor is memorization of the training data. We show that models falsely label NLI test samples as entailing when the hypothesis is attested in the training text, regardless of the premise. We further show that named entity IDs are used as "indices" to access the memorized data. Second, we show that LLMs exploit a further corpus-based heuristic using the relative frequencies of words. We show that LLMs score significantly worse on NLI test samples which do not conform to these factors than those which do; we also discuss a tension between the two factors, and a performance trade-off.
Enabling Large Language Models to Generate Text with Citations
Authors: Tianyu Gao, Howard Yen, Jiatong Yu, Danqi Chen
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Abstract
Large language models (LLMs) have emerged as a widely-used tool for information seeking, but their generated outputs are prone to hallucination. In this work, we aim to enable LLMs to generate text with citations, improving their factual correctness and verifiability. Existing work mainly relies on commercial search engines and human evaluation, making it challenging to reproduce and compare with different modeling approaches. We propose ALCE, the first benchmark for Automatic LLMs' Citation Evaluation. ALCE collects a diverse set of questions and retrieval corpora and requires building end-to-end systems to retrieve supporting evidence and generate answers with citations. We build automatic metrics along three dimensions -- fluency, correctness, and citation quality -- and demonstrate their strong correlation with human judgements. Our experiments with state-of-the-art LLMs and novel prompting strategies show that current systems have considerable room for improvements -- for example, on the ELI5 dataset, even the best model has 49% of its generations lacking complete citation support. Our extensive analyses further highlight promising future directions, including developing better retrievers, advancing long-context LLMs, and improving the ability to synthesize information from multiple sources.
Trusting Your Evidence: Hallucinate Less with Context-aware Decoding
Authors: Weijia Shi, Xiaochuang Han, Mike Lewis, Yulia Tsvetkov, Luke Zettlemoyer, Scott Wen-tau Yih
Abstract
Language models (LMs) often struggle to pay enough attention to the input context, and generate texts that are unfaithful or contain hallucinations. To mitigate this issue, we present context-aware decoding (CAD), which follows a contrastive output distribution that amplifies the difference between the output probabilities when a model is used with and without context. Our experiments show that CAD, without additional training, significantly improves the faithfulness of different LM families, including OPT, GPT, LLaMA and FLAN-T5 for summarization tasks (e.g., 14.3% gain for LLaMA in factuality metrics). Furthermore, CAD is particularly effective in overriding a model's prior knowledge when it contradicts the provided context, leading to substantial improvements in tasks where resolving the knowledge conflict is essential.
Abstract
Existing text summarization systems have made significant progress in recent years but typically generates summaries in a single step. The one-shot summarization setting is sometimes inadequate, however, as the generated summary may contain hallucinations or overlook important details related to the reader's interests. In this paper, we address this limitation by proposing SummIt, an iterative text summarization framework based on large language models like ChatGPT. Our framework enables the model to refine the generated summary iteratively through self-evaluation and feedback, closely resembling the iterative process humans undertake when drafting and revising summaries. We also explore using in-context learning to guide the rationale generation and summary refinement. Furthermore, we explore the potential benefits of integrating knowledge and topic extractors into the framework to enhance summary faithfulness and controllability. We evaluate the performance of our framework on three benchmark summarization datasets through empirical and qualitative analyses. We also conduct a human evaluation to validate the effectiveness of the model's refinements and find a potential issue of over-correction. Our code is available at \url{https://github.com/hpzhang94/summ_it}.
PURR: Efficiently Editing Language Model Hallucinations by Denoising Language Model Corruptions
Abstract
The remarkable capabilities of large language models have been accompanied by a persistent drawback: the generation of false and unsubstantiated claims commonly known as "hallucinations". To combat this issue, recent research has introduced approaches that involve editing and attributing the outputs of language models, particularly through prompt-based editing. However, the inference cost and speed of using large language models for editing currently bottleneck prompt-based methods. These bottlenecks motivate the training of compact editors, which is challenging due to the scarcity of training data for this purpose. To overcome these challenges, we exploit the power of large language models to introduce corruptions (i.e., noise) into text and subsequently fine-tune compact editors to denoise the corruptions by incorporating relevant evidence. Our methodology is entirely unsupervised and provides us with faux hallucinations for training in any domain. Our Petite Unsupervised Research and Revision model, PURR, not only improves attribution over existing editing methods based on fine-tuning and prompting, but also achieves faster execution times by orders of magnitude.
Getting Sick After Seeing a Doctor? Diagnosing and Mitigating Knowledge Conflicts in Event Temporal Reasoning
Abstract
Event temporal reasoning aims at identifying the temporal relations between two or more events. However, knowledge conflicts arise when there is a mismatch between the actual temporal relations of events in the context and the prior knowledge or biases learned by the model. We first systematically define distinct kinds of bias in event temporal reasoning, which include event relation prior bias, tense bias, narrative bias, and dependency bias, as indicators to study knowledge conflicts. To mitigate such event-related knowledge conflict, we introduce a Counterfactual Data Augmentation based method that can be applied to both Pre-trained Language Models (PLMs) and Large Language Models (LLMs) either as additional training data or demonstrations for In-Context Learning. Experiments suggest the importance of mitigating knowledge conflicts in event temporal reasoning tasks for reducing hallucination and highlight the potential of counterfactual data augmentation for improving model performance.
RefGPT: Reference -> Truthful & Customized Dialogues Generation by GPTs and for GPTs
Abstract
General chat models, like ChatGPT, have attained impressive capability to resolve a wide range of NLP tasks by tuning Large Language Models (LLMs) with high-quality instruction data. However, collecting human-written high-quality data, especially multi-turn dialogues, is expensive and unattainable for most people. Though previous studies have used powerful LLMs to generate the dialogues automatically, but they all suffer from generating untruthful dialogues because of the LLMs hallucination. Therefore, we propose a method called RefGPT to generate enormous truthful and customized dialogues without worrying about factual errors caused by the model hallucination. RefGPT solves the model hallucination in dialogue generation by restricting the LLMs to leverage the given reference instead of reciting their own knowledge to generate dialogues. Additionally, RefGPT adds detailed controls on every utterances to enable highly customization capability, which previous studies have ignored. On the basis of RefGPT, we also propose two high-quality dialogue datasets generated by GPT-4, namely RefGPT-Fact and RefGPT-Code. RefGPT-Fact is 100k multi-turn dialogue datasets based on factual knowledge and RefGPT-Code is 76k multi-turn dialogue dataset covering a wide range of coding scenarios. Our code and datasets are released in https://github.com/ziliwangnlp/RefGPT
Abstract
Large Language Models (LLMs), like LLaMA, have exhibited remarkable performances across various tasks. Nevertheless, when deployed to specific domains such as law or medicine, the models still confront the challenge of a deficiency in domain-specific knowledge and an inadequate capability to leverage that knowledge to resolve domain-related problems. In this paper, we focus on the legal domain and explore how to inject domain knowledge during the continual training stage and how to design proper supervised finetune tasks to help the model tackle practical issues. Moreover, to alleviate the hallucination problem during model's generation, we add a retrieval module and extract relevant articles before the model answers any queries. Augmenting with the extracted evidence, our model could generate more reliable responses. We release our data and model at https://github.com/AndrewZhe/lawyer-llama.
SciReviewGen: A Large-scale Dataset for Automatic Literature Review Generation
Abstract
Automatic literature review generation is one of the most challenging tasks in natural language processing. Although large language models have tackled literature review generation, the absence of large-scale datasets has been a stumbling block to the progress. We release SciReviewGen, consisting of over 10,000 literature reviews and 690,000 papers cited in the reviews. Based on the dataset, we evaluate recent transformer-based summarization models on the literature review generation task, including Fusion-in-Decoder extended for literature review generation. Human evaluation results show that some machine-generated summaries are comparable to human-written reviews, while revealing the challenges of automatic literature review generation such as hallucinations and a lack of detailed information. Our dataset and code are available at https://github.com/tetsu9923/SciReviewGen.
Enhancing Retrieval-Augmented Large Language Models with Iterative Retrieval-Generation Synergy
Abstract
Large language models are powerful text processors and reasoners, but are still subject to limitations including outdated knowledge and hallucinations, which necessitates connecting them to the world. Retrieval-augmented large language models have raised extensive attention for grounding model generation on external knowledge. However, retrievers struggle to capture relevance, especially for queries with complex information needs. Recent work has proposed to improve relevance modeling by having large language models actively involved in retrieval, i.e., to improve retrieval with generation. In this paper, we show that strong performance can be achieved by a method we call Iter-RetGen, which synergizes retrieval and generation in an iterative manner. A model output shows what might be needed to finish a task, and thus provides an informative context for retrieving more relevant knowledge which in turn helps generate a better output in the next iteration. Compared with recent work which interleaves retrieval with generation when producing an output, Iter-RetGen processes all retrieved knowledge as a whole and largely preserves the flexibility in generation without structural constraints. We evaluate Iter-RetGen on multi-hop question answering, fact verification, and commonsense reasoning, and show that it can flexibly leverage parametric knowledge and non-parametric knowledge, and is superior to or competitive with state-of-the-art retrieval-augmented baselines while causing fewer overheads of retrieval and generation. We can further improve performance via generation-augmented retrieval adaptation.
Gorilla: Large Language Model Connected with Massive APIs
Authors: Shishir G. Patil, Tianjun Zhang, Xin Wang, Joseph E. Gonzalez
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Large Language Models (LLMs) have seen an impressive wave of advances recently, with models now excelling in a variety of tasks, such as mathematical reasoning and program synthesis. However, their potential to effectively use tools via API calls remains unfulfilled. This is a challenging task even for today's state-of-the-art LLMs such as GPT-4, largely due to their inability to generate accurate input arguments and their tendency to hallucinate the wrong usage of an API call. We release Gorilla, a finetuned LLaMA-based model that surpasses the performance of GPT-4 on writing API calls. When combined with a document retriever, Gorilla demonstrates a strong capability to adapt to test-time document changes, enabling flexible user updates or version changes. It also substantially mitigates the issue of hallucination, commonly encountered when prompting LLMs directly. To evaluate the model's ability, we introduce APIBench, a comprehensive dataset consisting of HuggingFace, TorchHub, and TensorHub APIs. The successful integration of the retrieval system with Gorilla demonstrates the potential for LLMs to use tools more accurately, keep up with frequently updated documentation, and consequently increase the reliability and applicability of their outputs. Gorilla's code, model, data, and demo are available at https://gorilla.cs.berkeley.edu
Keyword: evaluation
Is Information Extraction Solved by ChatGPT? An Analysis of Performance, Evaluation Criteria, Robustness and Errors
Authors: Ridong Han, Tao Peng, Chaohao Yang, Benyou Wang, Lu Liu, Xiang Wan
Abstract
ChatGPT has stimulated the research boom in the field of large language models. In this paper, we assess the capabilities of ChatGPT from four perspectives including Performance, Evaluation Criteria, Robustness and Error Types. Specifically, we first evaluate ChatGPT's performance on 17 datasets with 14 IE sub-tasks under the zero-shot, few-shot and chain-of-thought scenarios, and find a huge performance gap between ChatGPT and SOTA results. Next, we rethink this gap and propose a soft-matching strategy for evaluation to more accurately reflect ChatGPT's performance. Then, we analyze the robustness of ChatGPT on 14 IE sub-tasks, and find that: 1) ChatGPT rarely outputs invalid responses; 2) Irrelevant context and long-tail target types greatly affect ChatGPT's performance; 3) ChatGPT cannot understand well the subject-object relationships in RE task. Finally, we analyze the errors of ChatGPT, and find that "unannotated spans" is the most dominant error type. This raises concerns about the quality of annotated data, and indicates the possibility of annotating data with ChatGPT. The data and code are released at Github site.
On Robustness of Finetuned Transformer-based NLP Models
Abstract
Transformer-based pretrained models like BERT, GPT-2 and T5 have been finetuned for a large number of natural language processing (NLP) tasks, and have been shown to be very effective. However, while finetuning, what changes across layers in these models with respect to pretrained checkpoints is under-studied. Further, how robust are these models to perturbations in input text? Does the robustness vary depending on the NLP task for which the models have been finetuned? While there exists some work on studying robustness of BERT finetuned for a few NLP tasks, there is no rigorous study which compares this robustness across encoder only, decoder only and encoder-decoder models. In this paper, we study the robustness of three language models (BERT, GPT-2 and T5) with eight different text perturbations on the General Language Understanding Evaluation (GLUE) benchmark. Also, we use two metrics (CKA and STIR) to quantify changes between pretrained and finetuned language model representations across layers. GPT-2 representations are more robust than BERT and T5 across multiple types of input perturbation. Although models exhibit good robustness broadly, dropping nouns, verbs or changing characters are the most impactful. Overall, this study provides valuable insights into perturbation-specific weaknesses of popular Transformer-based models which should be kept in mind when passing inputs.
Pre-training Language Models for Comparative Reasoning
Abstract
In this paper, we propose a novel framework to pre-train language models for enhancing their abilities of comparative reasoning over texts. While recent research has developed models for NLP tasks that require comparative reasoning, they suffer from costly manual data labeling and limited generalizability to different tasks. Our approach involves a scalable method for collecting data for text-based entity comparison, which leverages both structured and unstructured data, and the design of three novel pre-training tasks. Evaluation on a range of downstream tasks including comparative question answering, question generation, and summarization shows that our pre-training framework significantly improves the comparative reasoning abilities of language models, especially under low-resource conditions. This work also releases the first integrated benchmark for comparative reasoning over texts.
Dancing Between Success and Failure: Edit-level Simplification Evaluation using SALSA
Authors: David Heineman, Yao Dou, Mounica Maddela, Wei Xu
Abstract
Large language models (e.g., GPT-3.5) are uniquely capable of producing highly rated text simplification, yet current human evaluation methods fail to provide a clear understanding of systems' specific strengths and weaknesses. To address this limitation, we introduce SALSA, an edit-based human annotation framework that enables holistic and fine-grained text simplification evaluation. We develop twenty one linguistically grounded edit types, covering the full spectrum of success and failure across dimensions of conceptual, syntactic and lexical simplicity. Using SALSA, we collect 12K edit annotations on 700 simplifications, revealing discrepancies in the distribution of transformation approaches performed by fine-tuned models, few-shot LLMs and humans, and finding GPT-3.5 performs more quality edits than humans, but still exhibits frequent errors. Using our fine-grained annotations, we develop LENS-SALSA, a reference-free automatic simplification metric, trained to predict sentence- and word-level quality simultaneously. Additionally, we introduce word-level quality estimation for simplification and report promising baseline results. Our training material, annotation toolkit, and data are released at this http URL
Towards Massively Multi-domain Multilingual Readability Assessment
Authors: Tarek Naous, Michael J. Ryan, Mohit Chandra, Wei Xu
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
We present ReadMe++, a massively multi-domain multilingual dataset for automatic readability assessment. Prior work on readability assessment has been mostly restricted to the English language and one or two text domains. Additionally, the readability levels of sentences used in many previous datasets are assumed on the document-level other than sentence-level, which raises doubt about the quality of previous evaluations. We address those gaps in the literature by providing an annotated dataset of 6,330 sentences in Arabic, English, and Hindi collected from 64 different domains of text. Unlike previous datasets, ReadMe++ offers more domain and language diversity and is manually annotated at a sentence level using the Common European Framework of Reference for Languages (CEFR) and through a Rank-and-Rate annotation framework that reduces subjectivity in annotation. Our experiments demonstrate that models fine-tuned using ReadMe++ achieve strong cross-lingual transfer capabilities and generalization to unseen domains. ReadMe++ will be made publicly available to the research community.
CGCE: A Chinese Generative Chat Evaluation Benchmark for General and Financial Domains
Abstract
Generative chat models, such as ChatGPT and GPT-4, have revolutionized natural language generation (NLG) by incorporating instructions and human feedback to achieve significant performance improvements. However, the lack of standardized evaluation benchmarks for chat models, particularly for Chinese and domain-specific models, hinders their assessment and progress. To address this gap, we introduce the Chinese Generative Chat Evaluation (CGCE) benchmark, focusing on general and financial domains. The CGCE benchmark encompasses diverse tasks, including 200 questions in the general domain and 150 specific professional questions in the financial domain. Manual scoring evaluates factors such as accuracy, coherence, expression clarity, and completeness. The CGCE benchmark provides researchers with a standardized framework to assess and compare Chinese generative chat models, fostering advancements in NLG research.
Language Model Self-improvement by Reinforcement Learning Contemplation
Abstract
Large Language Models (LLMs) have exhibited remarkable performance across various natural language processing (NLP) tasks. However, fine-tuning these models often necessitates substantial supervision, which can be expensive and time-consuming to obtain. This paper introduces a novel unsupervised method called LanguageModel Self-Improvement by Reinforcement Learning Contemplation (SIRLC) that improves LLMs without reliance on external labels. Our approach is grounded in the observation that it is simpler for language models to assess text quality than to generate text. Building on this insight, SIRLC assigns LLMs dual roles as both student and teacher. As a student, the LLM generates answers to unlabeled questions, while as a teacher, it evaluates the generated text and assigns scores accordingly. The model parameters are updated using reinforcement learning to maximize the evaluation score. We demonstrate that SIRLC can be applied to various NLP tasks, such as reasoning problems, text generation, and machine translation. Our experiments show that SIRLC effectively improves LLM performance without external supervision, resulting in a 5.6% increase in answering accuracy for reasoning tasks and a rise in BERTScore from 0.82 to 0.86 for translation tasks. Furthermore, SIRLC can be applied to models of different sizes, showcasing its broad applicability.
Sociocultural Norm Similarities and Differences via Situational Alignment and Explainable Textual Entailment
Abstract
Designing systems that can reason across cultures requires that they are grounded in the norms of the contexts in which they operate. However, current research on developing computational models of social norms has primarily focused on American society. Here, we propose a novel approach to discover and compare descriptive social norms across Chinese and American cultures. We demonstrate our approach by leveraging discussions on a Chinese Q&A platform-Zhihu-and the existing SocialChemistry dataset as proxies for contrasting cultural axes, align social situations cross-culturally, and extract social norms from texts using in-context learning. Embedding Chain-of-Thought prompting in a human-AI collaborative framework, we build a high-quality dataset of 3,069 social norms aligned with social situations across Chinese and American cultures alongside corresponding free-text explanations. To test the ability of models to reason about social norms across cultures, we introduce the task of explainable social norm entailment, showing that existing models under 3B parameters have significant room for improvement in both automatic and human evaluation. Further analysis of cross-cultural norm differences based on our dataset shows empirical alignment with the social orientations framework, revealing several situational and descriptive nuances in norms across these cultures.
Self-Polish: Enhance Reasoning in Large Language Models via Problem Refinement
Abstract
Prompting methods such as Chain-of-Thought (CoT) have shed new light on enhancing the reasoning capabilities of large language models, and researchers have extensively explored the generation process of rationales and answers. However, they have overlooked the potential challenges posed by the poor quality of reasoning problems, which may influence the reasoning performance significantly. In this work, we propose Self-Polish (SP), a novel method that facilitates the model's problem-solving process by prompting them to progressively refine the given problems to be more comprehensible and solvable. Specifically, the method teaches models to eliminate irrelevant information, rearrange the logic structure and organize local conditions into new ones parallelly. SP is orthogonal to all other prompting methods, making it convenient to integrate with state-of-the-art techniques for further improvement. We conduct thorough experiments on five benchmarks to illustrate the effectiveness of the proposed method. For example, with Text-davinci-003, our method boosts the performance of standard few-shot prompting by $8.0\%$ on GSM8K and $17.8\%$ on MultiArith; it also improves the performance of CoT by $6.0\%$ on GSM8K and $6.0\%$ on MathQA, respectively. Furthermore, our method also showcases impressive performance on robustness evaluation.
How to Choose How to Choose Your Chatbot: A Massively Multi-System MultiReference Data Set for Dialog Metric Evaluation
Authors: Huda Khayrallah, Zuhaib Akhtar, Edward Cohen, João Sedoc
Abstract
We release MMSMR, a Massively Multi-System MultiReference dataset to enable future work on metrics and evaluation for dialog. Automatic metrics for dialogue evaluation should be robust proxies for human judgments; however, the verification of robustness is currently far from satisfactory. To quantify the robustness correlation and understand what is necessary in a test set, we create and release an 8-reference dialog dataset by extending single-reference evaluation sets and introduce this new language learning conversation dataset. We then train 1750 systems and evaluate them on our novel test set and the DailyDialog dataset. We release the novel test set, and model hyper parameters, inference outputs, and metric scores for each system on a variety of datasets.
LLMs as Factual Reasoners: Insights from Existing Benchmarks and Beyond
Authors: Philippe Laban, Wojciech Kryściński, Divyansh Agarwal, Alexander R. Fabbri, Caiming Xiong, Shafiq Joty, Chien-Sheng Wu
Abstract
With the recent appearance of LLMs in practical settings, having methods that can effectively detect factual inconsistencies is crucial to reduce the propagation of misinformation and improve trust in model outputs. When testing on existing factual consistency benchmarks, we find that a few large language models (LLMs) perform competitively on classification benchmarks for factual inconsistency detection compared to traditional non-LLM methods. However, a closer analysis reveals that most LLMs fail on more complex formulations of the task and exposes issues with existing evaluation benchmarks, affecting evaluation precision. To address this, we propose a new protocol for inconsistency detection benchmark creation and implement it in a 10-domain benchmark called SummEdits. This new benchmark is 20 times more cost-effective per sample than previous benchmarks and highly reproducible, as we estimate inter-annotator agreement at about 0.9. Most LLMs struggle on SummEdits, with performance close to random chance. The best-performing model, GPT-4, is still 8\% below estimated human performance, highlighting the gaps in LLMs' ability to reason about facts and detect inconsistencies when they occur.
Interpretable Automatic Fine-grained Inconsistency Detection in Text Summarization
Abstract
Existing factual consistency evaluation approaches for text summarization provide binary predictions and limited insights into the weakness of summarization systems. Therefore, we propose the task of fine-grained inconsistency detection, the goal of which is to predict the fine-grained types of factual errors in a summary. Motivated by how humans inspect factual inconsistency in summaries, we propose an interpretable fine-grained inconsistency detection model, FineGrainFact, which explicitly represents the facts in the documents and summaries with semantic frames extracted by semantic role labeling, and highlights the related semantic frames to predict inconsistency. The highlighted semantic frames help verify predicted error types and correct inconsistent summaries. Experiment results demonstrate that our model outperforms strong baselines and provides evidence to support or refute the summary.
Unraveling ChatGPT: A Critical Analysis of AI-Generated Goal-Oriented Dialogues and Annotations
Authors: Tiziano Labruna, Sofia Brenna, Andrea Zaninello, Bernardo Magnini
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Large pre-trained language models have exhibited unprecedented capabilities in producing high-quality text via prompting techniques. This fact introduces new possibilities for data collection and annotation, particularly in situations where such data is scarce, complex to gather, expensive, or even sensitive. In this paper, we explore the potential of these models to generate and annotate goal-oriented dialogues, and conduct an in-depth analysis to evaluate their quality. Our experiments employ ChatGPT, and encompass three categories of goal-oriented dialogues (task-oriented, collaborative, and explanatory), two generation modes (interactive and one-shot), and two languages (English and Italian). Based on extensive human-based evaluations, we demonstrate that the quality of generated dialogues and annotations is on par with those generated by humans.
Evaluating OpenAI's Whisper ASR for Punctuation Prediction and Topic Modeling of life histories of the Museum of the Person
Authors: Lucas Rafael Stefanel Gris, Ricardo Marcacini, Arnaldo Candido Junior, Edresson Casanova, Anderson Soares, Sandra Maria Aluísio
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Automatic speech recognition (ASR) systems play a key role in applications involving human-machine interactions. Despite their importance, ASR models for the Portuguese language proposed in the last decade have limitations in relation to the correct identification of punctuation marks in automatic transcriptions, which hinder the use of transcriptions by other systems, models, and even by humans. However, recently Whisper ASR was proposed by OpenAI, a general-purpose speech recognition model that has generated great expectations in dealing with such limitations. This chapter presents the first study on the performance of Whisper for punctuation prediction in the Portuguese language. We present an experimental evaluation considering both theoretical aspects involving pausing points (comma) and complete ideas (exclamation, question, and fullstop), as well as practical aspects involving transcript-based topic modeling - an application dependent on punctuation marks for promising performance. We analyzed experimental results from videos of Museum of the Person, a virtual museum that aims to tell and preserve people's life histories, thus discussing the pros and cons of Whisper in a real-world scenario. Although our experiments indicate that Whisper achieves state-of-the-art results, we conclude that some punctuation marks require improvements, such as exclamation, semicolon and colon.
Contextualized Topic Coherence Metrics
Authors: Hamed Rahimi, Jacob Louis Hoover, David Mimno, Hubert Naacke, Camelia Constantin, Bernd Amann
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR)
Abstract
The recent explosion in work on neural topic modeling has been criticized for optimizing automated topic evaluation metrics at the expense of actual meaningful topic identification. But human annotation remains expensive and time-consuming. We propose LLM-based methods inspired by standard human topic evaluations, in a family of metrics called Contextualized Topic Coherence (CTC). We evaluate both a fully automated version as well as a semi-automated CTC that allows human-centered evaluation of coherence while maintaining the efficiency of automated methods. We evaluate CTC relative to five other metrics on six topic models and find that it outperforms automated topic coherence methods, works well on short documents, and is not susceptible to meaningless but high-scoring topics.
OpenPI2.0: An Improved Dataset for Entity Tracking in Texts
Authors: Li Zhang, Hainiu Xu, Abhinav Kommula, Niket Tandon, Chris Callison-Burch
Abstract
Representing texts as information about entities has long been deemed effective in event reasoning. We propose OpenPI2.0, an improved dataset for tracking entity states in procedural texts. OpenPI2.0 features not only canonicalized entities that facilitate evaluation, but also salience annotations including both manual labels and automatic predictions. Regarding entity salience, we provide a survey on annotation subjectivity, modeling feasibility, and downstream applications in tasks such as question answering and classical planning.
This Land is {Your, My} Land: Evaluating Geopolitical Biases in Language Models
Abstract
We introduce the notion of geopolitical bias -- a tendency to report different geopolitical knowledge depending on the linguistic context. As a case study, we consider territorial disputes between countries. For example, for the widely contested Spratly Islands, would an LM be more likely to say they belong to China if asked in Chinese, vs. to the Philippines if asked in Tagalog? To evaluate if such biases exist, we first collect a dataset of territorial disputes from Wikipedia, then associate each territory with a set of multilingual, multiple-choice questions. This dataset, termed BorderLines, consists of 250 territories with questions in 45 languages. We pose these question sets to language models, and analyze geopolitical bias in their responses through several proposed quantitative metrics. The metrics compare between responses in different question languages as well as to the actual geopolitical situation. The phenomenon of geopolitical bias is a uniquely cross-lingual evaluation, contrasting with prior work's monolingual (mostly English) focus on bias evaluation. Its existence shows that the knowledge of LMs, unlike multilingual humans, is inconsistent across languages.
KNN-LM Does Not Improve Open-ended Text Generation
Abstract
In this paper, we study the generation quality of interpolation-based retrieval-augmented language models (LMs). These methods, best exemplified by the KNN-LM, interpolate the LM's predicted distribution of the next word with a distribution formed from the most relevant retrievals for a given prefix. While the KNN-LM and related methods yield impressive decreases in perplexity, we discover that they do not exhibit corresponding improvements in open-ended generation quality, as measured by both automatic evaluation metrics (e.g., MAUVE) and human evaluations. Digging deeper, we find that interpolating with a retrieval distribution actually increases perplexity compared to a baseline Transformer LM for the majority of tokens in the WikiText-103 test set, even though the overall perplexity is lower due to a smaller number of tokens for which perplexity dramatically decreases after interpolation. However, when decoding a long sequence at inference time, significant improvements on this smaller subset of tokens are washed out by slightly worse predictions on most tokens. Furthermore, we discover that the entropy of the retrieval distribution increases faster than that of the base LM as the generated sequence becomes longer, which indicates that retrieval is less reliable when using model-generated text as queries (i.e., is subject to exposure bias). We hope that our analysis spurs future work on improved decoding algorithms and interpolation strategies for retrieval-augmented language models.
Enabling Large Language Models to Generate Text with Citations
Authors: Tianyu Gao, Howard Yen, Jiatong Yu, Danqi Chen
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Abstract
Large language models (LLMs) have emerged as a widely-used tool for information seeking, but their generated outputs are prone to hallucination. In this work, we aim to enable LLMs to generate text with citations, improving their factual correctness and verifiability. Existing work mainly relies on commercial search engines and human evaluation, making it challenging to reproduce and compare with different modeling approaches. We propose ALCE, the first benchmark for Automatic LLMs' Citation Evaluation. ALCE collects a diverse set of questions and retrieval corpora and requires building end-to-end systems to retrieve supporting evidence and generate answers with citations. We build automatic metrics along three dimensions -- fluency, correctness, and citation quality -- and demonstrate their strong correlation with human judgements. Our experiments with state-of-the-art LLMs and novel prompting strategies show that current systems have considerable room for improvements -- for example, on the ELI5 dataset, even the best model has 49% of its generations lacking complete citation support. Our extensive analyses further highlight promising future directions, including developing better retrievers, advancing long-context LLMs, and improving the ability to synthesize information from multiple sources.
Evaluate What You Can't Evaluate: Unassessable Generated Responses Quality
Abstract
LLMs (large language models) such as ChatGPT have shown remarkable language understanding and generation capabilities. Although reference-free evaluators based on LLMs show better human alignment than traditional reference-based evaluators, there are many challenges in using reference-free evaluators based on LLMs. Reference-free evaluators are more suitable for open-ended examples with different semantics responses. But not all examples are open-ended. For closed-ended examples with unique correct semantic response, reference-free evaluators will still consider it high quality when giving a response that is inconsistent with the facts and the semantic of reference. In order to comprehensively evaluate the reliability of evaluators based on LLMs, we construct two adversarial meta-evaluation dialogue generation datasets KdConv-ADV and DSTC7-ADV based on KdConv and DSTC7-AVSD, respectively. Compared to previous meta-evaluation benchmarks, KdConv-ADV and DSTC7-ADV are much more challenging since they requires evaluators to be able to reasonably evaluate closed-ended examples with the help of external knowledge or even its own knowledge. Empirical results show that the ability of LLMs to identify unreasonable responses is insufficient. There are risks in using eference-free evaluators based on LLMs to evaluate the quality of dialogue responses.
ExpertPrompting: Instructing Large Language Models to be Distinguished Experts
Authors: Benfeng Xu, An Yang, Junyang Lin, Quan Wang, Chang Zhou, Yongdong Zhang, Zhendong Mao
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
The answering quality of an aligned large language model (LLM) can be drastically improved if treated with proper crafting of prompts. In this paper, we propose ExpertPrompting to elicit the potential of LLMs to answer as distinguished experts. We first utilize In-Context Learning to automatically synthesize detailed and customized descriptions of the expert identity for each specific instruction, and then ask LLMs to provide answer conditioned on such agent background. Based on this augmented prompting strategy, we produce a new set of instruction-following data using GPT-3.5, and train a competitive open-source chat assistant called ExpertLLaMA. We employ GPT4-based evaluation to show that 1) the expert data is of significantly higher quality than vanilla answers, and 2) ExpertLLaMA outperforms existing open-source opponents and achieves 96\% of the original ChatGPT's capability. All data and the ExpertLLaMA model will be made publicly available at \url{https://github.com/OFA-Sys/ExpertLLaMA}.
Analyzing Influential Factors in Human Preference Judgments via GPT-4
Abstract
Pairwise human judgments are pivotal in guiding large language models (LLMs) to generate outputs that align with human preferences. They are also often used in summarization evaluation, complementing existing automatic metrics. Despite their significance, however, there has been limited research probing these pairwise human judgments. The collective impact and respective weights of factors such as informativeness, coherence, fluency, and factual consistency remain elusive. The impact of hidden factors on the final judgment is also unclear. In this paper, we conduct an in-depth examination of a dataset of pairwise human judgments released by OpenAI. Utilizing the Bradley-Terry-Luce model, we identify key factors that could potentially influence human judgments. Our research uncovers the inherent preferences embedded in human judgments and suggests strategies to boost sample efficiency. Finally, we provide insights on the construction of balanced datasets for human judgment evaluations, a crucial step in shaping the behaviors of future LLMs.
Gender Biases in Automatic Evaluation Metrics: A Case Study on Image Captioning
Abstract
Pretrained model-based evaluation metrics have demonstrated strong performance with high correlations with human judgments in various natural language generation tasks such as image captioning. Despite the impressive results, their impact on fairness is under-explored -- it is widely acknowledged that pretrained models can encode societal biases, and utilizing them for evaluation purposes may inadvertently manifest and potentially amplify biases. In this paper, we conduct a systematic study in gender biases of model-based evaluation metrics with a focus on image captioning tasks. Specifically, we first identify and quantify gender biases in different evaluation metrics regarding profession, activity, and object concepts. Then, we demonstrate the negative consequences of using these biased metrics, such as favoring biased generation models in deployment and propagating the biases to generation models through reinforcement learning. We also present a simple but effective alternative to reduce gender biases by combining n-gram matching-based and pretrained model-based evaluation metrics.
I Spy a Metaphor: Large Language Models and Diffusion Models Co-Create Visual Metaphors
Abstract
Visual metaphors are powerful rhetorical devices used to persuade or communicate creative ideas through images. Similar to linguistic metaphors, they convey meaning implicitly through symbolism and juxtaposition of the symbols. We propose a new task of generating visual metaphors from linguistic metaphors. This is a challenging task for diffusion-based text-to-image models, such as DALL$\cdot$E 2, since it requires the ability to model implicit meaning and compositionality. We propose to solve the task through the collaboration between Large Language Models (LLMs) and Diffusion Models: Instruct GPT-3 (davinci-002) with Chain-of-Thought prompting generates text that represents a visual elaboration of the linguistic metaphor containing the implicit meaning and relevant objects, which is then used as input to the diffusion-based text-to-image models.Using a human-AI collaboration framework, where humans interact both with the LLM and the top-performing diffusion model, we create a high-quality dataset containing 6,476 visual metaphors for 1,540 linguistic metaphors and their associated visual elaborations. Evaluation by professional illustrators shows the promise of LLM-Diffusion Model collaboration for this task.To evaluate the utility of our Human-AI collaboration framework and the quality of our dataset, we perform both an intrinsic human-based evaluation and an extrinsic evaluation using visual entailment as a downstream task.
SenteCon: Leveraging Lexicons to Learn Human-Interpretable Language Representations
Abstract
Although deep language representations have become the dominant form of language featurization in recent years, in many settings it is important to understand a model's decision-making process. This necessitates not only an interpretable model but also interpretable features. In particular, language must be featurized in a way that is interpretable while still characterizing the original text well. We present SenteCon, a method for introducing human interpretability in deep language representations. Given a passage of text, SenteCon encodes the text as a layer of interpretable categories in which each dimension corresponds to the relevance of a specific category. Our empirical evaluations indicate that encoding language with SenteCon provides high-level interpretability at little to no cost to predictive performance on downstream tasks. Moreover, we find that SenteCon outperforms existing interpretable language representations with respect to both its downstream performance and its agreement with human characterizations of the text.
Mastering the ABCDs of Complex Questions: Answer-Based Claim Decomposition for Fine-grained Self-Evaluation
Authors: Nishant Balepur, Jie Huang, Samraj Moorjani, Hari Sundaram, Kevin Chen-Chuan Chang
Abstract
When answering complex questions, large language models (LLMs) may produce answers that do not satisfy all criteria of the question. While existing self-evaluation techniques aim to detect if such answers are correct, these techniques are unable to determine which criteria of the question are satisfied by the generated answers. To address this issue, we propose answer-based claim decomposition (ABCD), a prompting strategy that decomposes questions into a series of true/false claims that can be used to verify which criteria of the input question an answer satisfies. Using the decomposed ABCD claims, we perform fine-grained self-evaluation. Through preliminary experiments on three datasets, including a newly-collected challenge dataset ObscureQA, we find that GPT-3.5 has some ability to determine to what extent its answer satisfies the criteria of the input question, and can give insights into the errors and knowledge gaps of the model.
Don't Take This Out of Context! On the Need for Contextual Models and Evaluations for Stylistic Rewriting
Authors: Akhila Yerukola, Xuhui Zhou, Maarten Sap
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Most existing stylistic text rewriting methods operate on a sentence level, but ignoring the broader context of the text can lead to generic, ambiguous, and incoherent rewrites. In this paper, we propose the integration of preceding textual context into both the rewriting and evaluation stages of stylistic text rewriting, focusing on formality, toxicity, and sentiment transfer tasks. We conduct a comparative evaluation of rewriting through few-shot prompting of GPT-3.5 and GPT NeoX, comparing non-contextual rewrites to contextual rewrites. Our experiments show that humans often prefer contextual rewrites over non-contextual ones, but automatic metrics (e.g., BLEU, sBERT) do not. To bridge this gap, we propose context-infused versions of common automatic metrics, and show that these better reflect human preferences. Overall, our paper highlights the importance of integrating preceding textual context into both the rewriting and evaluation stages of stylistic text rewriting.
Human-Centered Metrics for Dialog System Evaluation
Authors: Salvatore Giorgi, Shreya Havaldar, Farhan Ahmed, Zuhaib Akhtar, Shalaka Vaidya, Gary Pan, Lyle H. Ungar, H. Andrew Schwartz, Joao Sedoc
Abstract
We present metrics for evaluating dialog systems through a psychologically-grounded "human" lens: conversational agents express a diversity of both states (short-term factors like emotions) and traits (longer-term factors like personality) just as people do. These interpretable metrics consist of five measures from established psychology constructs that can be applied both across dialogs and on turns within dialogs: emotional entropy, linguistic style and emotion matching, as well as agreeableness and empathy. We compare these human metrics against 6 state-of-the-art automatic metrics (e.g. BARTScore and BLEURT) on 7 standard dialog system data sets. We also introduce a novel data set, the Three Bot Dialog Evaluation Corpus, which consists of annotated conversations from ChatGPT, GPT-3, and BlenderBot. We demonstrate the proposed human metrics offer novel information, are uncorrelated with automatic metrics, and lead to increased accuracy beyond existing automatic metrics for predicting crowd-sourced dialog judgements. The interpretability and unique signal of our proposed human-centered framework make it a valuable tool for evaluating and improving dialog systems.
Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Models
Abstract
The escalating debate on AI's capabilities warrants developing reliable metrics to assess machine "intelligence". Recently, many anecdotal examples were used to suggest that newer large language models (LLMs) like ChatGPT and GPT-4 exhibit Neural Theory-of-Mind (N-ToM); however, prior work reached conflicting conclusions regarding those abilities. We investigate the extent of LLMs' N-ToM through an extensive evaluation on 6 tasks and find that while LLMs exhibit certain N-ToM abilities, this behavior is far from being robust. We further examine the factors impacting performance on N-ToM tasks and discover that LLMs struggle with adversarial examples, indicating reliance on shallow heuristics rather than robust ToM abilities. We caution against drawing conclusions from anecdotal examples, limited benchmark testing, and using human-designed psychological tests to evaluate models.
Using Natural Language Explanations to Rescale Human Judgments
Abstract
The rise of large language models (LLMs) has brought a critical need for high-quality human-labeled data, particularly for processes like human feedback and evaluation. A common practice is to label data via consensus annotation over the judgments of multiple crowdworkers. However, different annotators may have different interpretations of labeling schemes unless given extensive training, and for subjective NLP tasks, even trained expert annotators can diverge heavily. We show that these nuances can be captured by high quality natural language explanations, and propose a method to rescale ordinal annotation in the presence of disagreement using LLMs. Specifically, we feed Likert ratings and corresponding natural language explanations into an LLM and prompt it to produce a numeric score. This score should reflect the underlying assessment of the example by the annotator. The presence of explanations allows the LLM to homogenize ratings across annotators in spite of scale usage differences. We explore our technique in the context of a document-grounded question answering task on which large language models achieve near-human performance. Among questions where annotators identify incompleteness in the answers, our rescaling improves correlation between nearly all annotator pairs, improving pairwise correlation on these examples by an average of 0.2 Kendall's tau.
ChatGPT and Simple Linguistic Inferences: Blind Spots and Blinds
Authors: Victoria Basmov, Yoav Goldberg, Reut Tsarfaty
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
This paper sheds light on the limitations of ChatGPT's understanding capabilities, focusing on simple inference tasks that are typically easy for humans but appear to be challenging for the model. Specifically, we target (i) grammatically-specified entailments, (ii) premises with evidential adverbs of uncertainty, and (iii) monotonicity entailments. We present expert-designed evaluation sets for these inference types and conduct experiments in a zero-shot setup. Our results show that the model struggles with these types of inferences, exhibiting moderate to low accuracy. Moreover, while ChatGPT demonstrates knowledge of the underlying linguistic concepts when prompted directly, it often fails to incorporate this knowledge to make correct inferences. Even more strikingly, further experiments show that embedding the premise under presupposition triggers or non-factive verbs causes the model to predict entailment more frequently {regardless} of the correct semantic label. Overall these results suggest that, despite GPT's celebrated language understanding capacity, ChatGPT has blindspots with respect to certain types of entailment, and that certain entailment-cancelling features act as ``blinds'' overshadowing the semantics of the embedded premise. Our analyses emphasize the need for further research into the linguistic comprehension and reasoning capabilities of LLMs, in order to improve their reliability, and establish their trustworthiness for real-world applications.
Faithful Low-Resource Data-to-Text Generation through Cycle Training
Abstract
Methods to generate text from structured data have advanced significantly in recent years, primarily due to fine-tuning of pre-trained language models on large datasets. However, such models can fail to produce output faithful to the input data, particularly on out-of-domain data. Sufficient annotated data is often not available for specific domains, leading us to seek an unsupervised approach to improve the faithfulness of output text. Since the problem is fundamentally one of consistency between the representations of the structured data and text, we evaluate the effectiveness of cycle training in this work. Cycle training uses two models which are inverses of each other: one that generates text from structured data, and one which generates the structured data from natural language text. We show that cycle training, when initialized with a small amount of supervised data (100 samples in our case), achieves nearly the same performance as fully supervised approaches for the data-to-text generation task on the WebNLG, E2E, WTQ, and WSQL datasets. We perform extensive empirical analysis with automated evaluation metrics and a newly designed human evaluation schema to reveal different cycle training strategies' effectiveness of reducing various types of generation errors. Our code is publicly available at https://github.com/Edillower/CycleNLG.
MQuAKE: Assessing Knowledge Editing in Language Models via Multi-Hop Questions
Authors: Zexuan Zhong, Zhengxuan Wu, Christopher D. Manning, Christopher Potts, Danqi Chen
Abstract
The information stored in large language models (LLMs) falls out of date quickly, and retraining from scratch is often not an option. This has recently given rise to a range of techniques for injecting new facts through updating model weights. Current evaluation paradigms are extremely limited, mainly validating the recall of edited facts, but changing one fact should cause rippling changes to the model's related beliefs. If we edit the UK Prime Minister to now be Rishi Sunak, then we should get a different answer to Who is married to the British Prime Minister? In this work, we present a benchmark MQuAKE (Multi-hop Question Answering for Knowledge Editing) comprising multi-hop questions that assess whether edited models correctly answer questions where the answer should change as an entailed consequence of edited facts. While we find that current knowledge-editing approaches can recall edited facts accurately, they fail catastrophically on the constructed multi-hop questions. We thus propose a simple memory-based approach, MeLLo, which stores all edited facts externally while prompting the language model iteratively to generate answers that are consistent with the edited facts. While MQuAKE remains challenging, we show that MeLLo scales well with LLMs (up to 175B) and outperforms previous model editors by a large margin.
Abstract
Existing text summarization systems have made significant progress in recent years but typically generates summaries in a single step. The one-shot summarization setting is sometimes inadequate, however, as the generated summary may contain hallucinations or overlook important details related to the reader's interests. In this paper, we address this limitation by proposing SummIt, an iterative text summarization framework based on large language models like ChatGPT. Our framework enables the model to refine the generated summary iteratively through self-evaluation and feedback, closely resembling the iterative process humans undertake when drafting and revising summaries. We also explore using in-context learning to guide the rationale generation and summary refinement. Furthermore, we explore the potential benefits of integrating knowledge and topic extractors into the framework to enhance summary faithfulness and controllability. We evaluate the performance of our framework on three benchmark summarization datasets through empirical and qualitative analyses. We also conduct a human evaluation to validate the effectiveness of the model's refinements and find a potential issue of over-correction. Our code is available at \url{https://github.com/hpzhang94/summ_it}.
ComSL: A Composite Speech-Language Model for End-to-End Speech-to-Text Translation
Authors: Chenyang Le, Yao Qian, Long Zhou, Shujie Liu, Michael Zeng, Xuedong Huang
Subjects: Computation and Language (cs.CL); Sound (cs.SD); Audio and Speech Processing (eess.AS)
Abstract
Joint speech-language training is challenging due to the large demand for training data and GPU consumption, as well as the modality gap between speech and language. We present ComSL, a speech-language model built atop a composite architecture of public pretrained speech-only and language-only models and optimized data-efficiently for spoken language tasks. Particularly, we propose to incorporate cross-modality learning into transfer learning and conduct them simultaneously for downstream tasks in a multi-task learning manner. Our approach has demonstrated effectiveness in end-to-end speech-to-text translation tasks, achieving a new state-of-the-art average BLEU score of 31.5 on the multilingual speech to English text translation task for 21 languages, as measured on the public CoVoST2 evaluation set.
Exploring Sentiment Analysis Techniques in Natural Language Processing: A Comprehensive Review
Authors: Karthick Prasad Gunasekaran
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Sentiment analysis (SA) is the automated process of detecting and understanding the emotions conveyed through written text. Over the past decade, SA has gained significant popularity in the field of Natural Language Processing (NLP). With the widespread use of social media and online platforms, SA has become crucial for companies to gather customer feedback and shape their marketing strategies. Additionally, researchers rely on SA to analyze public sentiment on various topics. In this particular research study, a comprehensive survey was conducted to explore the latest trends and techniques in SA. The survey encompassed a wide range of methods, including lexicon-based, graph-based, network-based, machine learning, deep learning, ensemble-based, rule-based, and hybrid techniques. The paper also addresses the challenges and opportunities in SA, such as dealing with sarcasm and irony, analyzing multi-lingual data, and addressing ethical concerns. To provide a practical case study, Twitter was chosen as one of the largest online social media platforms. Furthermore, the researchers shed light on the diverse application areas of SA, including social media, healthcare, marketing, finance, and politics. The paper also presents a comparative and comprehensive analysis of existing trends and techniques, datasets, and evaluation metrics. The ultimate goal is to offer researchers and practitioners a systematic review of SA techniques, identify existing gaps, and suggest possible improvements. This study aims to enhance the efficiency and accuracy of SA processes, leading to smoother and error-free outcomes.
BUFFET: Benchmarking Large Language Models for Few-shot Cross-lingual Transfer
Authors: Akari Asai, Sneha Kudugunta, Xinyan Velocity Yu, Terra Blevins, Hila Gonen, Machel Reid, Yulia Tsvetkov, Sebastian Ruder, Hannaneh Hajishirzi
Abstract
Despite remarkable advancements in few-shot generalization in natural language processing, most models are developed and evaluated primarily in English. To facilitate research on few-shot cross-lingual transfer, we introduce a new benchmark, called BUFFET, which unifies 15 diverse tasks across 54 languages in a sequence-to-sequence format and provides a fixed set of few-shot examples and instructions. BUFFET is designed to establish a rigorous and equitable evaluation framework for few-shot cross-lingual transfer across a broad range of tasks and languages. Using BUFFET, we perform thorough evaluations of state-of-the-art multilingual large language models with different transfer methods, namely in-context learning and fine-tuning. Our findings reveal significant room for improvement in few-shot in-context cross-lingual transfer. In particular, ChatGPT with in-context learning often performs worse than much smaller mT5-base models fine-tuned on English task data and few-shot in-language examples. Our analysis suggests various avenues for future research in few-shot cross-lingual transfer, such as improved pretraining, understanding, and future evaluations.
From Words to Wires: Generating Functioning Electronic Devices from Natural Language Descriptions
Authors: Peter Jansen
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
In this work, we show that contemporary language models have a previously unknown skill -- the capacity for electronic circuit design from high-level textual descriptions, akin to code generation. We introduce two benchmarks: Pins100, assessing model knowledge of electrical components, and Micro25, evaluating a model's capability to design common microcontroller circuits and code in the Arduino ecosystem that involve input, output, sensors, motors, protocols, and logic -- with models such as GPT-4 and Claude-V1 achieving between 60% to 96% Pass@1 on generating full devices. We include six case studies of using language models as a design assistant for moderately complex devices, such as a radiation-powered random number generator, an emoji keyboard, a visible spectrometer, and several assistive devices, while offering a qualitative analysis performance, outlining evaluation challenges, and suggesting areas of development to improve complex circuit design and practical utility. With this work, we aim to spur research at the juncture of natural language processing and electronic design.
ByteSized32: A Corpus and Challenge Task for Generating Task-Specific World Models Expressed as Text Games
Authors: Ruoyao Wang, Graham Todd, Eric Yuan, Ziang Xiao, Marc-Alexandre Côté, Peter Jansen
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
In this work we examine the ability of language models to generate explicit world models of scientific and common-sense reasoning tasks by framing this as a problem of generating text-based games. To support this, we introduce ByteSized32, a corpus of 32 highly-templated text games written in Python totaling 24k lines of code, each centered around a particular task, and paired with a set of 16 unseen text game specifications for evaluation. We propose a suite of automatic and manual metrics for assessing simulation validity, compliance with task specifications, playability, winnability, and alignment with the physical world. In a single-shot evaluation of GPT-4 on this simulation-as-code-generation task, we find it capable of producing runnable games in 27% of cases, highlighting the difficulty of this challenge task. We discuss areas of future improvement, including GPT-4's apparent capacity to perform well at simulating near canonical task solutions, with performance dropping off as simulations include distractors or deviate from canonical solutions in the action space.
Evaluating NLG Evaluation Metrics: A Measurement Theory Perspective
Authors: Ziang Xiao, Susu Zhang, Vivian Lai, Q. Vera Liao
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
We address the fundamental challenge in Natural Language Generation (NLG) model evaluation, the design and validation of evaluation metrics. Recognizing the limitations of existing metrics and issues with human judgment, we propose using measurement theory, the foundation of test design, as a framework for conceptualizing and evaluating the validity and reliability of NLG evaluation metrics. This approach offers a systematic method for defining "good" metrics, developing robust metrics, and assessing metric performance. In this paper, we introduce core concepts in measurement theory in the context of NLG evaluation and key methods to evaluate the performance of NLG metrics. Through this framework, we aim to promote the design, evaluation, and interpretation of valid and reliable metrics, ultimately contributing to the advancement of robust and effective NLG models in real-world settings.
Text encoders are performance bottlenecks in contrastive vision-language models
Authors: Amita Kamath, Jack Hessel, Kai-Wei Chang
Subjects: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
Performant vision-language (VL) models like CLIP represent captions using a single vector. How much information about language is lost in this bottleneck? We first curate CompPrompts, a set of increasingly compositional image captions that VL models should be able to capture (e.g., single object, to object+property, to multiple interacting objects). Then, we train text-only recovery probes that aim to reconstruct captions from single-vector text representations produced by several VL models. This approach doesn't require images, allowing us to test on a broader range of scenes compared to prior work. We find that: 1) CLIP's text encoder falls short on object relationships, attribute-object association, counting, and negations; 2) some text encoders work significantly better than others; and 3) text-only recovery performance predicts multi-modal matching performance on ControlledImCaps: a new evaluation benchmark we collect+release consisting of fine-grained compositional images+captions. Specifically -- our results suggest text-only recoverability is a necessary (but not sufficient) condition for modeling compositional factors in contrastive vision+language models. We release data+code.
A Fair and In-Depth Evaluation of Existing End-to-End Entity Linking Systems
Abstract
Existing evaluations of entity linking systems often say little about how the system is going to perform for a particular application. There are four fundamental reasons for this: many benchmarks focus on named entities; it is hard to define which other entities to include; there are ambiguities in entity recognition and entity linking; many benchmarks have errors or artifacts that invite overfitting or lead to evaluation results of limited meaningfulness. We provide a more meaningful and fair in-depth evaluation of a variety of existing end-to-end entity linkers. We characterize the strengths and weaknesses of these linkers and how well the results from the respective publications can be reproduced. Our evaluation is based on several widely used benchmarks, which exhibit the problems mentioned above to various degrees, as well as on two new benchmarks, which address these problems.
Do LLMs Understand Social Knowledge? Evaluating the Sociability of Large Language Models with SocKET Benchmark
Abstract
Large language models (LLMs) have been shown to perform well at a variety of syntactic, discourse, and reasoning tasks. While LLMs are increasingly deployed in many forms including conversational agents that interact with humans, we lack a grounded benchmark to measure how well LLMs understand \textit{social} language. Here, we introduce a new theory-driven benchmark, SocKET, that contains 58 NLP tasks testing social knowledge which we group into five categories: humor & sarcasm, offensiveness, sentiment & emotion, and trustworthiness. In tests on the benchmark, we demonstrate that current models attain only moderate performance but reveal significant potential for task transfer among different types and categories of tasks, which were predicted from theory. Through zero-shot evaluations, we show that pretrained models already possess some innate but limited capabilities of social language understanding and training on one category of tasks can improve zero-shot testing on others. Our benchmark provides a systematic way to analyze model performance on an important dimension of language and points to clear room for improvement to build more socially-aware LLMs. The associated resources are released at https://github.com/minjechoi/SOCKET.
How Predictable Are Large Language Model Capabilities? A Case Study on BIG-bench
Authors: Qinyuan Ye, Harvey Yiyun Fu, Xiang Ren, Robin Jia
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Abstract
We investigate the predictability of large language model (LLM) capabilities: given records of past experiments using different model families, numbers of parameters, tasks, and numbers of in-context examples, can we accurately predict LLM performance on new experiment configurations? Answering this question has practical implications for LLM users (e.g., deciding which models to try), developers (e.g., prioritizing evaluation on representative tasks), and the research community (e.g., identifying hard-to-predict capabilities that warrant further investigation). We study the performance prediction problem on experiment records from BIG-bench. On a random train-test split, an MLP-based predictor achieves RMSE below 5%, demonstrating the presence of learnable patterns within the experiment records. Further, we formulate the problem of searching for "small-bench," an informative subset of BIG-bench tasks from which the performance of the full set can be maximally recovered, and find a subset as informative for evaluating new model families as BIG-bench Hard, while being 3x smaller.
Abstract
Memory editing methods for updating encyclopedic knowledge in transformers have received increasing attention for their efficacy, specificity, and generalization advantages. However, it remains unclear if such methods can be adapted for the more nuanced domain of commonsense knowledge. We propose $MEMIT{CSK}$, an adaptation of MEMIT to edit commonsense mistakes in GPT-2 Large and XL. We extend editing to various token locations and employ a robust layer selection strategy. Models edited by $MEMIT{CSK}$ outperforms the fine-tuning baselines by 10.97% and 10.73% F1 scores on subsets of PEP3k and 20Q. We further propose a novel evaluation dataset, MEMIT-CSK-PROBE, that contains unaffected neighborhood, affected neighborhood, affected paraphrase, and affected reasoning challenges. $MEMIT_{CSK}$ demonstrates favorable semantic generalization, outperforming fine-tuning baselines by 13.72% and 5.57% overall scores on MEMIT-CSK-PROBE. These results suggest a compelling future direction of incorporating context-specific user feedback concerning commonsense in GPT by direct model editing, rectifying and customizing model behaviors via human-in-the-loop systems.
Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback
Authors: Katherine Tian, Eric Mitchell, Allan Zhou, Archit Sharma, Rafael Rafailov, Huaxiu Yao, Chelsea Finn, Christopher D. Manning
Abstract
A trustworthy real-world prediction system should be well-calibrated; that is, its confidence in an answer is indicative of the likelihood that the answer is correct, enabling deferral to a more expensive expert in cases of low-confidence predictions. While recent studies have shown that unsupervised pre-training produces large language models (LMs) that are remarkably well-calibrated, the most widely-used LMs in practice are fine-tuned with reinforcement learning with human feedback (RLHF-LMs) after the initial unsupervised pre-training stage, and results are mixed as to whether these models preserve the well-calibratedness of their ancestors. In this paper, we conduct a broad evaluation of computationally feasible methods for extracting confidence scores from LLMs fine-tuned with RLHF. We find that with the right prompting strategy, RLHF-LMs verbalize probabilities that are much better calibrated than the model's conditional probabilities, enabling fairly well-calibrated predictions. Through a combination of prompting strategy and temperature scaling, we find that we can reduce the expected calibration error of RLHF-LMs by over 50%.
GPTAraEval: A Comprehensive Evaluation of ChatGPT on Arabic NLP
Authors: Md Tawkat Islam Khondaker, Abdul Waheed, El Moatez Billah Nagoudi, Muhammad Abdul-Mageed
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Abstract
The recent emergence of ChatGPT has brought a revolutionary change in the landscape of NLP. Although ChatGPT has consistently shown impressive performance on English benchmarks, its exact capabilities on most other languages remain largely unknown. To better understand ChatGPT's capabilities on Arabic, we present a large-scale evaluation of the model on a broad range of Arabic NLP tasks. Namely, we evaluate ChatGPT on 32 diverse natural language understanding and generation tasks on over 60 different datasets. To the best of our knowledge, our work offers the first performance analysis of ChatGPT on Arabic NLP at such a massive scale. Our results show that, despite its success on English benchmarks, ChatGPT trained in-context (few-shot) is consistently outperformed by much smaller dedicated models finetuned on Arabic. These results suggest that there is significant place for improvement for instruction-tuned LLMs such as ChatGPT.
Large Language Models are Effective Table-to-Text Generators, Evaluators, and Feedback Providers
Abstract
Large language models (LLMs) have shown remarkable ability on controllable text generation. However, the potential of LLMs in generating text from structured tables remains largely under-explored. In this paper, we study the capabilities of LLMs for table-to-text generation tasks, particularly aiming to investigate their performance in generating natural language statements that can be logically entailed by a provided table. First, we investigate how LLMs compare to state-of-the-art table-to-text fine-tuned models, and demonstrate that LLMs can generate statements with higher faithfulness compared with previous state-of-the-art fine-tuned models. Given this finding, we next explore whether LLMs can serve as faithfulness-level automated evaluation metrics. Through human evaluation, we show that evaluation metrics adopted from LLMs correlates better with human judgments compared with existing faithfulness-level metrics. Finally, we demonstrate that LLMs using chain-of-thought prompting can generate high-fidelity natural language feedback for other table-to-text models' generations, provide insights for future work regarding the distillation of text generation capabilities from LLMs to smaller models.
Dolphin: A Challenging and Diverse Benchmark for Arabic NLG
Authors: El Moatez Billah Nagoudi, Ahmed El-Shangiti, AbdelRahim Elmadany, Muhammad Abdul-Mageed
Abstract
We present Dolphin, a novel benchmark that addresses the need for an evaluation framework for the wide collection of Arabic languages and varieties. The proposed benchmark encompasses a broad range of 13 different NLG tasks, including text summarization, machine translation, question answering, and dialogue generation, among others. Dolphin comprises a substantial corpus of 40 diverse and representative public datasets across 50 test splits, carefully curated to reflect real-world scenarios and the linguistic richness of Arabic. It sets a new standard for evaluating the performance and generalization capabilities of Arabic and multilingual models, promising to enable researchers to push the boundaries of current methodologies. We provide an extensive analysis of Dolphin, highlighting its diversity and identifying gaps in current Arabic NLG research. We also evaluate several Arabic and multilingual models on our benchmark, allowing us to set strong baselines against which researchers can compare.
MuLER: Detailed and Scalable Reference-based Evaluation
Authors: Taelin Karidi, Leshem Choshen, Gal Patel, Omri Abend
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
We propose a novel methodology (namely, MuLER) that transforms any reference-based evaluation metric for text generation, such as machine translation (MT) into a fine-grained analysis tool. Given a system and a metric, MuLER quantifies how much the chosen metric penalizes specific error types (e.g., errors in translating names of locations). MuLER thus enables a detailed error analysis which can lead to targeted improvement efforts for specific phenomena. We perform experiments in both synthetic and naturalistic settings to support MuLER's validity and showcase its usability in MT evaluation, and other tasks, such as summarization. Analyzing all submissions to WMT in 2014-2020, we find consistent trends. For example, nouns and verbs are among the most frequent POS tags. However, they are among the hardest to translate. Performance on most POS tags improves with overall system performance, but a few are not thus correlated (their identity changes from language to language). Preliminary experiments with summarization reveal similar trends.
An Examination of the Robustness of Reference-Free Image Captioning Evaluation Metrics
Authors: Saba Ahmadi, Aishwarya Agrawal
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
Recently, reference-free metrics such as CLIPScore (Hessel et al., 2021) and UMIC (Lee et al., 2021) have been proposed for automatic evaluation of image captions, demonstrating a high correlation with human judgment. In this work, our focus lies in evaluating the robustness of these metrics in scenarios that require distinguishing between two captions with high lexical overlap but very different meanings. Our findings reveal that despite their high correlation with human judgment, both CLIPScore and UMIC struggle to identify fine-grained errors in captions. However, when comparing different types of fine-grained errors, both metrics exhibit limited sensitivity to implausibility of captions and strong sensitivity to lack of sufficient visual grounding. Probing further into the visual grounding aspect, we found that both CLIPScore and UMIC are impacted by the size of image-relevant objects mentioned in the caption, and that CLIPScore is also sensitive to the number of mentions of image-relevant objects in the caption. In terms of linguistic aspects of a caption, we found that both metrics lack the ability to comprehend negation, UMIC is sensitive to caption lengths, and CLIPScore is insensitive to the structure of the sentence. We hope our findings will serve as a valuable guide towards improving reference-free evaluation in image captioning.
Sentiment Analysis in the Era of Large Language Models: A Reality Check
Abstract
Sentiment analysis (SA) has been a long-standing research area in natural language processing. It can offer rich insights into human sentiments and opinions and has thus seen considerable interest from both academia and industry. With the advent of large language models (LLMs) such as ChatGPT, there is a great potential for their employment on SA problems. However, the extent to which existing LLMs can be leveraged for different sentiment analysis tasks remains unclear. This paper aims to provide a comprehensive investigation into the capabilities of LLMs in performing various sentiment analysis tasks, from conventional sentiment classification to aspect-based sentiment analysis and multifaceted analysis of subjective texts. We evaluate performance across 13 tasks on 26 datasets and compare the results against small language models (SLMs) trained on domain-specific datasets. Our study reveals that while LLMs demonstrate satisfactory performance in simpler tasks, they lag behind in more complex tasks requiring deeper understanding or structured sentiment information. However, LLMs significantly outperform SLMs in few-shot learning settings, suggesting their potential when annotation resources are limited. We also highlight the limitations of current evaluation practices in assessing LLMs' SA abilities and propose a novel benchmark, \textsc{SentiEval}, for a more comprehensive and realistic evaluation. Data and code during our investigations are available at \url{https://github.com/DAMO-NLP-SG/LLM-Sentiment}.
Bactrian-X : A Multilingual Replicable Instruction-Following Model with Low-Rank Adaptation
Abstract
Instruction tuning has shown great promise in the field of natural language processing. However, the research on multilingual instruction tuning has been limited due to the scarcity of high-quality instruction-response datasets. To address this gap, we present Bactrian-X, a comprehensive multilingual parallel dataset of 3.4 million instruction-response pairs across 52 languages. Leveraging this dataset, we train a set of adapters using low-rank adaptation (LoRA), which are lightweight components seamlessly integrated with foundational models. These adapters have a significantly smaller parameter count than the base model, making them easily replaceable and usable as plug-ins for different languages or language groups. Through extensive experiments on 52 languages, we demonstrate the superior performance of our models in various multilingual evaluation settings. Our proposed models outperform both the vanilla models and the existing instruction-tuned models. The code and models are publicly available at https://github.com/mbzuai-nlp/bactrian-x.
An Efficient Multilingual Language Model Compression through Vocabulary Trimming
Authors: Asahi Ushio, Yi Zhou, Jose Camacho-Collados
Abstract
Multilingual language model (LM) have become a powerful tool in NLP especially for non-English languages. Nevertheless, model parameters of multilingual LMs remain large due to the larger embedding matrix of the vocabulary covering tokens in different languages. On the contrary, monolingual LMs can be trained in a target language with the language-specific vocabulary only, but this requires a large budget and availability of reliable corpora to achieve a high-quality LM from scratch. In this paper, we propose vocabulary-trimming (VT), a method to reduce a multilingual LM vocabulary to a target language by deleting irrelevant tokens from its vocabulary. In theory, VT can compress any existing multilingual LM to build monolingual LMs in any language covered by the multilingual LM. In our experiments, we show that VT can retain the original performance of the multilingual LM, while being smaller in size (in general around 50% of the original vocabulary size is enough) than the original multilingual LM. The evaluation is performed over four NLP tasks (two generative and two classification tasks) among four widely used multilingual LMs in seven languages. Finally, we show that this methodology can keep the best of both monolingual and multilingual worlds by keeping a small size as monolingual models without the need for specifically retraining them, and even limiting potentially harmful social biases.
ImageNetVC: Zero-Shot Visual Commonsense Evaluation on 1000 ImageNet Categories
Authors: Heming Xia, Qingxiu Dong, Lei Li, Jingjing Xu, Ziwei Qin, Zhifang Sui
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Abstract
Recently, Pretrained Language Models (PLMs) have been serving as general-purpose interfaces, posing a significant demand for comprehensive visual knowledge. However, it remains unclear how well current PLMs and their visually augmented counterparts (VaLMs) can master visual commonsense knowledge. To investigate this, we propose ImageNetVC, a fine-grained, human-annotated dataset specifically designed for zero-shot visual commonsense evaluation across 1,000 ImageNet categories. Utilizing ImageNetVC, we delve into the fundamental visual commonsense knowledge of both unimodal PLMs and VaLMs, uncovering the scaling law and the influence of the backbone model on VaLMs. Furthermore, we investigate the factors affecting the visual commonsense knowledge of large-scale models, providing insights into the development of language models enriched with visual commonsense knowledge. Our code and dataset are available at https://github.com/hemingkx/ImageNetVC.
How to Distill your BERT: An Empirical Study on the Impact of Weight Initialisation and Distillation Objectives
Authors: Xinpeng Wang, Leonie Weissweiler, Hinrich Schütze, Barbara Plank
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Recently, various intermediate layer distillation (ILD) objectives have been shown to improve compression of BERT models via Knowledge Distillation (KD). However, a comprehensive evaluation of the objectives in both task-specific and task-agnostic settings is lacking. To the best of our knowledge, this is the first work comprehensively evaluating distillation objectives in both settings. We show that attention transfer gives the best performance overall. We also study the impact of layer choice when initializing the student from the teacher layers, finding a significant impact on the performance in task-specific distillation. For vanilla KD and hidden states transfer, initialisation with lower layers of the teacher gives a considerable improvement over higher layers, especially on the task of QNLI (up to an absolute percentage change of 17.8 in accuracy). Attention transfer behaves consistently under different initialisation settings. We release our code as an efficient transformer-based model distillation framework for further studies.
Self-ICL: Zero-Shot In-Context Learning with Self-Generated Demonstrations
Abstract
Large language models (LMs) have exhibited superior in-context learning (ICL) ability to adopt to target tasks by prompting with a few input-output demonstrations. Towards better ICL, different methods are proposed to select representative demonstrations from existing training corpora. However, such a setting is not aligned with real-world practices, as end-users usually query LMs without accesses to demonstration pools. Inspired by evidence suggesting LMs' zero-shot capabilities are underrated, and the role of demonstrations are primarily for exposing models' intrinsic functionalities, we introduce Self-ICL, a simple framework for zero-shot ICL. Given a test input, Self-ICL first prompts the model to generate pseudo-inputs. Next, the model predicts pseudo-labels for the pseudo-inputs via zero-shot prompting. Finally, we construct pseudo-demonstrations from pseudo-input-label pairs, and perform ICL for the test input. Evaluation on BIG-Bench Hard shows Self-ICL steadily surpasses zero-shot and zero-shot chain-of-thought baselines on head-to-head and all-task average performance. Our findings suggest the possibility to bootstrap LMs' intrinsic capabilities towards better zero-shot performance.
Abstract
As large language models (LLMs) have demonstrated their powerful capabilities in plenty of domains and tasks, including context understanding, code generation, language generation, data storytelling, etc., many data analysts may raise concerns if their jobs will be replaced by AI. This controversial topic has drawn a lot of attention in public. However, we are still at a stage of divergent opinions without any definitive conclusion. Motivated by this, we raise the research question of "is GPT-4 a good data analyst?" in this work and aim to answer it by conducting head-to-head comparative studies. In detail, we regard GPT-4 as a data analyst to perform end-to-end data analysis with databases from a wide range of domains. We propose a framework to tackle the problems by carefully designing the prompts for GPT-4 to conduct experiments. We also design several task-specific evaluation metrics to systematically compare the performance between several professional human data analysts and GPT-4. Experimental results show that GPT-4 can achieve comparable performance to humans. We also provide in-depth discussions about our results to shed light on further studies before we reach the conclusion that GPT-4 can replace data analysts.
Is Summary Useful or Not? An Extrinsic Human Evaluation of Text Summaries on Downstream Tasks
Abstract
Research on automated text summarization relies heavily on human and automatic evaluation. While recent work on human evaluation mainly adopted intrinsic evaluation methods, judging the generic quality of text summaries, e.g. informativeness and coherence, our work focuses on evaluating the usefulness of text summaries with extrinsic methods. We carefully design three different downstream tasks for extrinsic human evaluation of summaries, i.e., question answering, text classification and text similarity assessment. We carry out experiments using system rankings and user behavior data to evaluate the performance of different summarization models. We find summaries are particularly useful in tasks that rely on an overall judgment of the text, while being less effective for question answering tasks. The results show that summaries generated by fine-tuned models lead to higher consistency in usefulness across all three tasks, as rankings of fine-tuned summarization systems are close across downstream tasks according to the proposed extrinsic metrics. Summaries generated by models in the zero-shot setting, however, are found to be biased towards the text classification and similarity assessment tasks, due to its general and less detailed summary style. We further evaluate the correlation of 14 intrinsic automatic metrics with human criteria and show that intrinsic automatic metrics perform well in evaluating the usefulness of summaries in the question-answering task, but are less effective in the other two tasks. This highlights the limitations of relying solely on intrinsic automatic metrics in evaluating the performance and usefulness of summaries.
SETI: Systematicity Evaluation of Textual Inference
Abstract
We propose SETI (Systematicity Evaluation of Textual Inference), a novel and comprehensive benchmark designed for evaluating pre-trained language models (PLMs) for their systematicity capabilities in the domain of textual inference. Specifically, SETI offers three different NLI tasks and corresponding datasets to evaluate various types of systematicity in reasoning processes. In order to solve these tasks, models are required to perform compositional inference based on known primitive constituents. We conduct experiments of SETI on six widely used PLMs. Results show that various PLMs are able to solve unseen compositional inferences when having encountered the knowledge of how to combine primitives, with good performance. However, they are considerably limited when this knowledge is unknown to the model (40-100% points decrease). Furthermore, we find that PLMs can improve drastically once exposed to crucial compositional knowledge in minimalistic shots. These findings position SETI as the first benchmark for measuring the future progress of PLMs in achieving systematicity generalization in the textual inference.
Ranger: A Toolkit for Effect-Size Based Multi-Task Evaluation
Authors: Mete Sertkan, Sophia Althammer, Sebastian Hofstätter
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR)
Abstract
In this paper, we introduce Ranger - a toolkit to facilitate the easy use of effect-size-based meta-analysis for multi-task evaluation in NLP and IR. We observed that our communities often face the challenge of aggregating results over incomparable metrics and scenarios, which makes conclusions and take-away messages less reliable. With Ranger, we aim to address this issue by providing a task-agnostic toolkit that combines the effect of a treatment on multiple tasks into one statistical evaluation, allowing for comparison of metrics and computation of an overall summary effect. Our toolkit produces publication-ready forest plots that enable clear communication of evaluation results over multiple tasks. Our goal with the ready-to-use Ranger toolkit is to promote robust, effect-size-based evaluation and improve evaluation standards in the community. We provide two case studies for common IR and NLP settings to highlight Ranger's benefits.
Not All Metrics Are Guilty: Improving NLG Evaluation with LLM Paraphrasing
Abstract
Most research about natural language generation (NLG) relies on evaluation benchmarks with limited references for a sample, which may result in poor correlations with human judgements. The underlying reason is that one semantic meaning can actually be expressed in different forms, and the evaluation with a single or few references may not accurately reflect the quality of the model's hypotheses. To address this issue, this paper presents a novel method, named Para-Ref, to enhance existing evaluation benchmarks by enriching the number of references. We leverage large language models (LLMs) to paraphrase a single reference into multiple high-quality ones in diverse expressions. Experimental results on representative NLG tasks of machine translation, text summarization, and image caption demonstrate that our method can effectively improve the correlation with human evaluation for sixteen automatic evaluation metrics by +7.82% in ratio. We release the code and data at https://github.com/RUCAIBox/Para-Ref.
Have LLMs Advanced Enough? A Challenging Problem Solving Benchmark For Large Language Models
Abstract
The performance on Large Language Models (LLMs) on existing reasoning benchmarks has shot up considerably over the past years. In response, we present JEEBench, a considerably more challenging benchmark dataset for evaluating the problem solving abilities of LLMs. We curate 450 challenging pre-engineering mathematics, physics and chemistry problems from the IIT JEE-Advanced exam. Long-horizon reasoning on top of deep in-domain knowledge is essential for solving problems in this benchmark. Our evaluation on the GPT series of models reveals that although performance improves with newer models, the best being GPT-4, the highest performance, even after using techniques like Self-Consistency and Chain-of-Thought prompting is less than 40 percent. Our analysis demonstrates that errors in algebraic manipulation and failure in retrieving relevant domain specific concepts are primary contributors to GPT4's low performance. Given the challenging nature of the benchmark, we hope that it can guide future research in problem solving using LLMs. Our code and dataset is available here.
HuatuoGPT, towards Taming Language Model to Be a Doctor
Abstract
In this paper, we present HuatuoGPT, a large language model (LLM) for medical consultation. The core recipe of HuatuoGPT is to leverage both \textit{distilled data from ChatGPT} and \textit{real-world data from doctors} in the supervised fine-tuned stage. The responses of ChatGPT are usually detailed, well-presented and informative while it cannot perform like a doctor in many aspects, e.g. for integrative diagnosis. We argue that real-world data from doctors would be complementary to distilled data in the sense the former could tame a distilled language model to perform like doctors. To better leverage the strengths of both data, we train a reward model to align the language model with the merits that both data bring, following an RLAIF (reinforced learning from AI feedback) fashion. To evaluate and benchmark the models, we propose a comprehensive evaluation scheme (including automatic and manual metrics). Experimental results demonstrate that HuatuoGPT achieves state-of-the-art results in performing medical consultation among open-source LLMs in GPT-4 evaluation, human evaluation, and medical benchmark datasets. It is worth noting that by using additional real-world data and RLAIF, the distilled language model (i.e., HuatuoGPT) outperforms its teacher model ChatGPT in most cases. Our code, data, and models are publicly available at \url{https://github.com/FreedomIntelligence/HuatuoGPT}. The online demo is available at \url{https://www.HuatuoGPT.cn/}.
Cream: Visually-Situated Natural Language Understanding with Contrastive Reading Model and Frozen Large Language Models
Authors: Geewook Kim, Hodong Lee, Daehee Kim, Haeji Jung, Sanghee Park, Yoonsik Kim, Sangdoo Yun, Taeho Kil, Bado Lee, Seunghyun Park
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Advances in Large Language Models (LLMs) have inspired a surge of research exploring their expansion into the visual domain. While recent models exhibit promise in generating abstract captions for images and conducting natural conversations, their performance on text-rich images leaves room for improvement. In this paper, we propose the Contrastive Reading Model (Cream), a novel neural architecture designed to enhance the language-image understanding capability of LLMs by capturing intricate details typically overlooked by existing methods. Cream integrates vision and auxiliary encoders, complemented by a contrastive feature alignment technique, resulting in a more effective understanding of textual information within document images. Our approach, thus, seeks to bridge the gap between vision and language understanding, paving the way for more sophisticated Document Intelligence Assistants. Rigorous evaluations across diverse tasks, such as visual question answering on document images, demonstrate the efficacy of Cream as a state-of-the-art model in the field of visual document understanding. We provide our codebase and newly-generated datasets at https://github.com/naver-ai/cream
CSTS: Conditional Semantic Textual Similarity
Authors: Ameet Deshpande, Carlos E. Jimenez, Howard Chen, Vishvak Murahari, Victoria Graf, Tanmay Rajpurohit, Ashwin Kalyan, Danqi Chen, Karthik Narasimhan
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Semantic textual similarity (STS) has been a cornerstone task in NLP that measures the degree of similarity between a pair of sentences, with applications in information retrieval, question answering, and embedding methods. However, it is an inherently ambiguous task, with the sentence similarity depending on the specific aspect of interest. We resolve this ambiguity by proposing a novel task called conditional STS (C-STS) which measures similarity conditioned on an aspect elucidated in natural language (hereon, condition). As an example, the similarity between the sentences "The NBA player shoots a three-pointer." and "A man throws a tennis ball into the air to serve." is higher for the condition "The motion of the ball." (both upward) and lower for "The size of the ball." (one large and one small). C-STS's advantages are two-fold: (1) it reduces the subjectivity and ambiguity of STS, and (2) enables fine-grained similarity evaluation using diverse conditions. C-STS contains almost 20,000 instances from diverse domains and we evaluate several state-of-the-art models to demonstrate that even the most performant fine-tuning and in-context learning models (GPT-4, Flan, SimCSE) find it challenging, with Spearman correlation scores of <50. We encourage the community to evaluate their models on C-STS to provide a more holistic view of semantic similarity and natural language understanding.
On Degrees of Freedom in Defining and Testing Natural Language Understanding
Authors: Saku Sugawara, Shun Tsugita
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Natural language understanding (NLU) studies often exaggerate or underestimate the capabilities of systems, thereby limiting the reproducibility of their findings. These erroneous evaluations can be attributed to the difficulty of defining and testing NLU adequately. In this position paper, we reconsider this challenge by identifying two types of researcher degrees of freedom. We revisit Turing's original interpretation of the Turing test and indicate that an NLU test does not provide an operational definition; it merely provides inductive evidence that the test subject understands the language sufficiently well to meet stakeholder objectives. In other words, stakeholders are free to arbitrarily define NLU through their objectives. To use the test results as inductive evidence, stakeholders must carefully assess if the interpretation of test scores is valid or not. However, designing and using NLU tests involve other degrees of freedom, such as specifying target skills and defining evaluation metrics. As a result, achieving consensus among stakeholders becomes difficult. To resolve this issue, we propose a validity argument, which is a framework comprising a series of validation criteria across test components. By demonstrating that current practices in NLU studies can be associated with those criteria and organizing them into a comprehensive checklist, we prove that the validity argument can serve as a coherent guideline for designing credible test sets and facilitating scientific communication.
SciReviewGen: A Large-scale Dataset for Automatic Literature Review Generation
Abstract
Automatic literature review generation is one of the most challenging tasks in natural language processing. Although large language models have tackled literature review generation, the absence of large-scale datasets has been a stumbling block to the progress. We release SciReviewGen, consisting of over 10,000 literature reviews and 690,000 papers cited in the reviews. Based on the dataset, we evaluate recent transformer-based summarization models on the literature review generation task, including Fusion-in-Decoder extended for literature review generation. Human evaluation results show that some machine-generated summaries are comparable to human-written reviews, while revealing the challenges of automatic literature review generation such as hallucinations and a lack of detailed information. Our dataset and code are available at https://github.com/tetsu9923/SciReviewGen.
Boosting Cross-lingual Transferability in Multilingual Models via In-Context Learning
Authors: Sunkyoung Kim, Dayeon Ki, Yireun Kim, Jinsik Lee
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Existing cross-lingual transfer (CLT) prompting methods are only concerned with monolingual demonstration examples in the source language. In this paper, we propose In-CLT, a novel cross-lingual transfer prompting method that leverages both source and target languages to construct the demonstration examples. We conduct comprehensive evaluations on multilingual benchmarks, focusing on question answering tasks. Experiment results show that In-CLT prompt not only improves multilingual models' cross-lingual transferability, but also demonstrates remarkable unseen language generalization ability. In-CLT prompting, in particular, improves model performance by 10 to 20\% points on average when compared to prior cross-lingual transfer approaches. We also observe the surprising performance gain on the other multilingual benchmarks, especially in reasoning tasks. Furthermore, we investigate the relationship between lexical similarity and pre-training corpora in terms of the cross-lingual transfer gap.
Revisiting Parallel Context Windows: A Frustratingly Simple Alternative and Chain-of-Thought Deterioration
Abstract
We identify two crucial limitations in the evaluation of recent parallel-integrated method Parallel Context Windows (PCW), which extends the maximum context lengths of language models, e.g., 2048 for LLaMA, by harnessing window-wise attention and positional embedding techniques. We first show that a simple yet strong baseline, weighted sum ensemble, is missing for the in-context few-shot classification. Moreover, on more challenging Chain-of-Thought (CoT) reasoning (e.g., HotpotQA), PCW would present unexpected deterioration regarding question miscomprehension and false inference. Based on our findings, we suggest that the existing PCW design may not guarantee sufficient improvement and practicality in handling lengthy documents in real-world applications. More community efforts on enabling language models' long context understanding ability should be paid.
EvEval: A Comprehensive Evaluation of Event Semantics for Large Language Models
Abstract
Events serve as fundamental units of occurrence within various contexts. The processing of event semantics in textual information forms the basis of numerous natural language processing (NLP) applications. Recent studies have begun leveraging large language models (LLMs) to address event semantic processing. However, the extent that LLMs can effectively tackle these challenges remains uncertain. Furthermore, the lack of a comprehensive evaluation framework for event semantic processing poses a significant challenge in evaluating these capabilities. In this paper, we propose an overarching framework for event semantic processing, encompassing understanding, reasoning, and prediction, along with their fine-grained aspects. To comprehensively evaluate the event semantic processing abilities of models, we introduce a novel benchmark called EVEVAL. We collect 8 datasets that cover all aspects of event semantic processing. Extensive experiments are conducted on EVEVAL, leading to several noteworthy findings based on the obtained results.
Learning Answer Generation using Supervision from Automatic Question Answering Evaluators
Abstract
Recent studies show that sentence-level extractive QA, i.e., based on Answer Sentence Selection (AS2), is outperformed by Generation-based QA (GenQA) models, which generate answers using the top-k answer sentences ranked by AS2 models (a la retrieval-augmented generation style). In this paper, we propose a novel training paradigm for GenQA using supervision from automatic QA evaluation models (GAVA). Specifically, we propose three strategies to transfer knowledge from these QA evaluation models to a GenQA model: (i) augmenting training data with answers generated by the GenQA model and labelled by GAVA (either statically, before training, or (ii) dynamically, at every training epoch); and (iii) using the GAVA score for weighting the generator loss during the learning of the GenQA model. We evaluate our proposed methods on two academic and one industrial dataset, obtaining a significant improvement in answering accuracy over the previous state of the art.
Unit-based Speech-to-Speech Translation Without Parallel Data
Authors: Anuj Diwan, Anirudh Srinivasan, David Harwath, Eunsol Choi
Subjects: Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
Abstract
We propose an unsupervised speech-to-speech translation (S2ST) system that does not rely on parallel data between the source and target languages. Our approach maps source and target language speech signals into automatically discovered, discrete units and reformulates the problem as unsupervised unit-to-unit machine translation. We develop a three-step training procedure that involves (a) pre-training an unit-based encoder-decoder language model with a denoising objective (b) training it with word-by-word translated utterance pairs created by aligning monolingual text embedding spaces and (c) running unsupervised backtranslation bootstrapping off of the initial translation model. Our approach avoids mapping the speech signal into text and uses speech-to-unit and unit-to-speech models instead of automatic speech recognition and text to speech models. We evaluate our model on synthetic-speaker Europarl-ST English-German and German-English evaluation sets, finding that unit-based translation is feasible under this constrained scenario, achieving 9.29 ASR-BLEU in German to English and 8.07 in English to German.
Keyword: text generation
Enhancing Generation through Summarization Duality and Explicit Outline Control
Language Model Self-improvement by Reinforcement Learning Contemplation
Meta-review Generation with Checklist-guided Iterative Introspection
In-Context Demonstration Selection with Cross Entropy Difference
Faithful Low-Resource Data-to-Text Generation through Cycle Training
Trade-Offs Between Fairness and Privacy in Language Modeling
Large Language Models are Effective Table-to-Text Generators, Evaluators, and Feedback Providers
MuLER: Detailed and Scalable Reference-based Evaluation
Active Learning for Natural Language Generation
Inference-Time Policy Adapters (IPA): Tailoring Extreme-Scale LMs without Fine-tuning
Peek Across: Improving Multi-Document Modeling via Cross-Document Question-Answering
Keyword: machine translation
Language Model Self-improvement by Reinforcement Learning Contemplation
CMOT: Cross-modal Mixup via Optimal Transport for Speech Translation
Diffusion Models in NLP: A Survey
Leveraging GPT-4 for Automatic Translation Post-Editing
Dolphin: A Challenging and Diverse Benchmark for Arabic NLG
MuLER: Detailed and Scalable Reference-based Evaluation
Not All Metrics Are Guilty: Improving NLG Evaluation with LLM Paraphrasing
Unit-based Speech-to-Speech Translation Without Parallel Data
Keyword: non-autoregressive
NAIL: Lexical Retrieval Indices with Efficient Non-Autoregressive Decoders
Keyword: abstractive summarization
Improving Factuality of Abstractive Summarization without Sacrificing Summary Quality
Keyword: factual
Exploring Contrast Consistency of Open-Domain Question Answering Systems on Minimally Edited Questions
LLMs as Factual Reasoners: Insights from Existing Benchmarks and Beyond
Interpretable Automatic Fine-grained Inconsistency Detection in Text Summarization
Self-Checker: Plug-and-Play Modules for Fact-Checking with Large Language Models
Enabling Large Language Models to Generate Text with Citations
Mixture of Prompt Experts for Generalizable and Interpretable Question Answering
Analyzing Influential Factors in Human Preference Judgments via GPT-4
The student becomes the master: Matching GPT3 on Scientific Factual Error Correction
Trusting Your Evidence: Hallucinate Less with Context-aware Decoding
Measuring the Knowledge Acquisition-Utilization Gap in Pretrained Language Models
Large Language Models as Counterfactual Generator: Strengths and Weaknesses
Privacy Implications of Retrieval-Based Language Models
Getting Sick After Seeing a Doctor? Diagnosing and Mitigating Knowledge Conflicts in Event Temporal Reasoning
Improving Factuality of Abstractive Summarization without Sacrificing Summary Quality
RefGPT: Reference -> Truthful & Customized Dialogues Generation by GPTs and for GPTs
Understanding Arithmetic Reasoning in Language Models using Causal Mediation Analysis
Meta-Learning Online Adaptation of Language Models
ASPER: Answer Set Programming Enhanced Neural Network Models for Joint Entity-Relation Extraction
Keyword: knowledge distillation
Large Language Model Distillation Doesn't Need a Teacher
CoLaDa: A Collaborative Label Denoising Framework for Cross-lingual Named Entity Recognition
How to Distill your BERT: An Empirical Study on the Impact of Weight Initialisation and Distillation Objectives
Keyword: Hallucination
Sources of Hallucination by Large Language Models on Inference Tasks
Enabling Large Language Models to Generate Text with Citations
Trusting Your Evidence: Hallucinate Less with Context-aware Decoding
SummIt: Iterative Text Summarization via ChatGPT
PURR: Efficiently Editing Language Model Hallucinations by Denoising Language Model Corruptions
Getting Sick After Seeing a Doctor? Diagnosing and Mitigating Knowledge Conflicts in Event Temporal Reasoning
RefGPT: Reference -> Truthful & Customized Dialogues Generation by GPTs and for GPTs
Lawyer LLaMA Technical Report
SciReviewGen: A Large-scale Dataset for Automatic Literature Review Generation
Enhancing Retrieval-Augmented Large Language Models with Iterative Retrieval-Generation Synergy
Gorilla: Large Language Model Connected with Massive APIs
Keyword: evaluation
Is Information Extraction Solved by ChatGPT? An Analysis of Performance, Evaluation Criteria, Robustness and Errors
On Robustness of Finetuned Transformer-based NLP Models
Pre-training Language Models for Comparative Reasoning
Dancing Between Success and Failure: Edit-level Simplification Evaluation using SALSA
Towards Massively Multi-domain Multilingual Readability Assessment
CGCE: A Chinese Generative Chat Evaluation Benchmark for General and Financial Domains
Language Model Self-improvement by Reinforcement Learning Contemplation
Sociocultural Norm Similarities and Differences via Situational Alignment and Explainable Textual Entailment
Self-Polish: Enhance Reasoning in Large Language Models via Problem Refinement
How to Choose How to Choose Your Chatbot: A Massively Multi-System MultiReference Data Set for Dialog Metric Evaluation
LLMs as Factual Reasoners: Insights from Existing Benchmarks and Beyond
Interpretable Automatic Fine-grained Inconsistency Detection in Text Summarization
Unraveling ChatGPT: A Critical Analysis of AI-Generated Goal-Oriented Dialogues and Annotations
Evaluating OpenAI's Whisper ASR for Punctuation Prediction and Topic Modeling of life histories of the Museum of the Person
Contextualized Topic Coherence Metrics
OpenPI2.0: An Improved Dataset for Entity Tracking in Texts
This Land is {Your, My} Land: Evaluating Geopolitical Biases in Language Models
KNN-LM Does Not Improve Open-ended Text Generation
Enabling Large Language Models to Generate Text with Citations
Evaluate What You Can't Evaluate: Unassessable Generated Responses Quality
ExpertPrompting: Instructing Large Language Models to be Distinguished Experts
Analyzing Influential Factors in Human Preference Judgments via GPT-4
Gender Biases in Automatic Evaluation Metrics: A Case Study on Image Captioning
I Spy a Metaphor: Large Language Models and Diffusion Models Co-Create Visual Metaphors
SenteCon: Leveraging Lexicons to Learn Human-Interpretable Language Representations
Mastering the ABCDs of Complex Questions: Answer-Based Claim Decomposition for Fine-grained Self-Evaluation
Don't Take This Out of Context! On the Need for Contextual Models and Evaluations for Stylistic Rewriting
Human-Centered Metrics for Dialog System Evaluation
Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Models
Using Natural Language Explanations to Rescale Human Judgments
ChatGPT and Simple Linguistic Inferences: Blind Spots and Blinds
Faithful Low-Resource Data-to-Text Generation through Cycle Training
MQuAKE: Assessing Knowledge Editing in Language Models via Multi-Hop Questions
SummIt: Iterative Text Summarization via ChatGPT
ComSL: A Composite Speech-Language Model for End-to-End Speech-to-Text Translation
Exploring Sentiment Analysis Techniques in Natural Language Processing: A Comprehensive Review
BUFFET: Benchmarking Large Language Models for Few-shot Cross-lingual Transfer
From Words to Wires: Generating Functioning Electronic Devices from Natural Language Descriptions
ByteSized32: A Corpus and Challenge Task for Generating Task-Specific World Models Expressed as Text Games
Evaluating NLG Evaluation Metrics: A Measurement Theory Perspective
Text encoders are performance bottlenecks in contrastive vision-language models
A Fair and In-Depth Evaluation of Existing End-to-End Entity Linking Systems
Do LLMs Understand Social Knowledge? Evaluating the Sociability of Large Language Models with SocKET Benchmark
How Predictable Are Large Language Model Capabilities? A Case Study on BIG-bench
Editing Commonsense Knowledge in GPT
Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback
GPTAraEval: A Comprehensive Evaluation of ChatGPT on Arabic NLP
Large Language Models are Effective Table-to-Text Generators, Evaluators, and Feedback Providers
Dolphin: A Challenging and Diverse Benchmark for Arabic NLG
MuLER: Detailed and Scalable Reference-based Evaluation
An Examination of the Robustness of Reference-Free Image Captioning Evaluation Metrics
Sentiment Analysis in the Era of Large Language Models: A Reality Check
Bactrian-X : A Multilingual Replicable Instruction-Following Model with Low-Rank Adaptation
An Efficient Multilingual Language Model Compression through Vocabulary Trimming
ImageNetVC: Zero-Shot Visual Commonsense Evaluation on 1000 ImageNet Categories
How to Distill your BERT: An Empirical Study on the Impact of Weight Initialisation and Distillation Objectives
Self-ICL: Zero-Shot In-Context Learning with Self-Generated Demonstrations
Is GPT-4 a Good Data Analyst?
Is Summary Useful or Not? An Extrinsic Human Evaluation of Text Summaries on Downstream Tasks
SETI: Systematicity Evaluation of Textual Inference
Ranger: A Toolkit for Effect-Size Based Multi-Task Evaluation
Not All Metrics Are Guilty: Improving NLG Evaluation with LLM Paraphrasing
Have LLMs Advanced Enough? A Challenging Problem Solving Benchmark For Large Language Models
HuatuoGPT, towards Taming Language Model to Be a Doctor
Cream: Visually-Situated Natural Language Understanding with Contrastive Reading Model and Frozen Large Language Models
CSTS: Conditional Semantic Textual Similarity
On Degrees of Freedom in Defining and Testing Natural Language Understanding
SciReviewGen: A Large-scale Dataset for Automatic Literature Review Generation
Boosting Cross-lingual Transferability in Multilingual Models via In-Context Learning
Revisiting Parallel Context Windows: A Frustratingly Simple Alternative and Chain-of-Thought Deterioration
EvEval: A Comprehensive Evaluation of Event Semantics for Large Language Models
Learning Answer Generation using Supervision from Automatic Question Answering Evaluators
Unit-based Speech-to-Speech Translation Without Parallel Data