arXiv subjects

Hwaran Lee

Publications and source records attributed to Hwaran Lee.

35 records · Page 2Linked to original sources

arXiv2024

Prometheus: Inducing Fine-grained Evaluation Capability in Language Models

Recently, using a powerful proprietary Large Language Model (LLM) (e.g., GPT-4) as an evaluator for long-form responses has become the de facto standard. However, for practitioners with large-scale evaluation tasks and custom criteria in consideration (e.g., child-readability), using proprietary LLMs as an evaluator is unreliable due to the closed-source nature, uncontrolled versioning, and prohibitive costs. In this work, we propose Prometheus, a fully open-source LLM that is on par with GPT-4's evaluation capabilities when the appropriate reference materials (reference answer, score rubric) are accompanied. We first construct the Feedback Collection, a new dataset that consists of 1K fine-grained score rubrics, 20K instructions, and 100K responses and language feedback generated by GPT-4. Using the Feedback Collection, we train Prometheus, a 13B evaluator LLM that can assess any given long-form text based on customized score rubric provided by the user. Experimental results show that Prometheus scores a Pearson correlation of 0.897 with human evaluators when evaluating with 45 customized score rubrics, which is on par with GPT-4 (0.882), and greatly outperforms ChatGPT (0.392). Furthermore, measuring correlation with GPT-4 with 1222 customized score rubrics across four benchmarks (MT Bench, Vicuna Bench, Feedback Bench, Flask Eval) shows similar trends, bolstering Prometheus's capability as an evaluator LLM. Lastly, Prometheus achieves the highest accuracy on two human preference benchmarks (HHH Alignment & MT Bench Human Judgment) compared to open-sourced reward models explicitly trained on human preference datasets, highlighting its potential as an universal reward model. We open-source our code, dataset, and model at https://kaistai.github.io/prometheus/.

cs.CL↗

arXiv2024

Calibrating Large Language Models Using Their Generations Only

As large language models (LLMs) are increasingly deployed in user-facing applications, building trust and maintaining safety by accurately quantifying a model's confidence in its prediction becomes even more important. However, finding effective ways to calibrate LLMs - especially when the only interface to the models is their generated text - remains a challenge. We propose APRICOT (auxiliary prediction of confidence targets): A method to set confidence targets and train an additional model that predicts an LLM's confidence based on its textual input and output alone. This approach has several advantages: It is conceptually simple, does not require access to the target model beyond its output, does not interfere with the language generation, and has a multitude of potential usages, for instance by verbalizing the predicted confidence or adjusting the given answer based on the confidence. We show how our approach performs competitively in terms of calibration error for white-box and black-box LLMs on closed-book question-answering to detect incorrect LLM answers.

cs.CL↗

arXiv2024

KoBBQ: Korean Bias Benchmark for Question Answering

The Bias Benchmark for Question Answering (BBQ) is designed to evaluate social biases of language models (LMs), but it is not simple to adapt this benchmark to cultural contexts other than the US because social biases depend heavily on the cultural context. In this paper, we present KoBBQ, a Korean bias benchmark dataset, and we propose a general framework that addresses considerations for cultural adaptation of a dataset. Our framework includes partitioning the BBQ dataset into three classes--Simply-Transferred (can be used directly after cultural translation), Target-Modified (requires localization in target groups), and Sample-Removed (does not fit Korean culture)-- and adding four new categories of bias specific to Korean culture. We conduct a large-scale survey to collect and validate the social biases and the targets of the biases that reflect the stereotypes in Korean culture. The resulting KoBBQ dataset comprises 268 templates and 76,048 samples across 12 categories of social bias. We use KoBBQ to measure the accuracy and bias scores of several state-of-the-art multilingual LMs. The results clearly show differences in the bias of LMs as measured by KoBBQ and a machine-translated version of BBQ, demonstrating the need for and utility of a well-constructed, culturally-aware social bias benchmark.

cs.CL↗

arXiv2023

ProPILE: Probing Privacy Leakage in Large Language Models

The rapid advancement and widespread use of large language models (LLMs) have raised significant concerns regarding the potential leakage of personally identifiable information (PII). These models are often trained on vast quantities of web-collected data, which may inadvertently include sensitive personal data. This paper presents ProPILE, a novel probing tool designed to empower data subjects, or the owners of the PII, with awareness of potential PII leakage in LLM-based services. ProPILE lets data subjects formulate prompts based on their own PII to evaluate the level of privacy intrusion in LLMs. We demonstrate its application on the OPT-1.3B model trained on the publicly available Pile dataset. We show how hypothetical data subjects may assess the likelihood of their PII being included in the Pile dataset being revealed. ProPILE can also be leveraged by LLM service providers to effectively evaluate their own levels of PII leakage with more powerful prompts specifically tuned for their in-house models. This tool represents a pioneering step towards empowering the data subjects for their awareness and control over their own data on the web.

cs.CR↗

arXiv2023

ClaimDiff: Comparing and Contrasting Claims on Contentious Issues

With the growing importance of detecting misinformation, many studies have focused on verifying factual claims by retrieving evidence. However, canonical fact verification tasks do not apply to catching subtle differences in factually consistent claims, which might still bias the readers, especially on contentious political or economic issues. Our underlying assumption is that among the trusted sources, one's argument is not necessarily more true than the other, requiring comparison rather than verification. In this study, we propose ClaimDiff, a novel dataset that primarily focuses on comparing the nuance between claim pairs. In ClaimDiff, we provide 2,941 annotated claim pairs from 268 news articles. We observe that while humans are capable of detecting the nuances between claims, strong baselines struggle to detect them, showing over a 19% absolute gap with the humans. We hope this initial study could help readers to gain an unbiased grasp of contentious issues through machine-aided comparison.

cs.CL↗

arXiv2023

KoSBi: A Dataset for Mitigating Social Bias Risks Towards Safer Large Language Model Application

Large language models (LLMs) learn not only natural text generation abilities but also social biases against different demographic groups from real-world data. This poses a critical risk when deploying LLM-based applications. Existing research and resources are not readily applicable in South Korea due to the differences in language and culture, both of which significantly affect the biases and targeted demographic groups. This limitation requires localized social bias datasets to ensure the safe and effective deployment of LLMs. To this end, we present KO SB I, a new social bias dataset of 34k pairs of contexts and sentences in Korean covering 72 demographic groups in 15 categories. We find that through filtering-based moderation, social biases in generated content can be reduced by 16.47%p on average for HyperCLOVA (30B and 82B), and GPT-3.

cs.CL↗

arXiv2023

SQuARe: A Large-Scale Dataset of Sensitive Questions and Acceptable Responses Created Through Human-Machine Collaboration

The potential social harms that large language models pose, such as generating offensive content and reinforcing biases, are steeply rising. Existing works focus on coping with this concern while interacting with ill-intentioned users, such as those who explicitly make hate speech or elicit harmful responses. However, discussions on sensitive issues can become toxic even if the users are well-intentioned. For safer models in such scenarios, we present the Sensitive Questions and Acceptable Response (SQuARe) dataset, a large-scale Korean dataset of 49k sensitive questions with 42k acceptable and 46k non-acceptable responses. The dataset was constructed leveraging HyperCLOVA in a human-in-the-loop manner based on real news headlines. Experiments show that acceptable response generation significantly improves for HyperCLOVA and GPT-3, demonstrating the efficacy of this dataset.

cs.CL↗

arXiv2023

Query-Efficient Black-Box Red Teaming via Bayesian Optimization

The deployment of large-scale generative models is often restricted by their potential risk of causing harm to users in unpredictable ways. We focus on the problem of black-box red teaming, where a red team generates test cases and interacts with the victim model to discover a diverse set of failures with limited query access. Existing red teaming methods construct test cases based on human supervision or language model (LM) and query all test cases in a brute-force manner without incorporating any information from past evaluations, resulting in a prohibitively large number of queries. To this end, we propose Bayesian red teaming (BRT), novel query-efficient black-box red teaming methods based on Bayesian optimization, which iteratively identify diverse positive test cases leading to model failures by utilizing the pre-defined user input pool and the past evaluations. Experimental results on various user input pools demonstrate that our method consistently finds a significantly larger number of diverse positive test cases under the limited query budget than the baseline methods. The source code is available at https://github.com/snu-mllab/Bayesian-Red-Teaming.

cs.AI↗

arXiv2022

Critic-Guided Decoding for Controlled Text Generation

Steering language generation towards objectives or away from undesired content has been a long-standing goal in utilizing language models (LM). Recent work has demonstrated reinforcement learning and weighted decoding as effective approaches to achieve a higher level of language control and quality with pros and cons. In this work, we propose a novel critic decoding method for controlled language generation (CriticControl) that combines the strengths of reinforcement learning and weighted decoding. Specifically, we adopt the actor-critic framework to train an LM-steering critic from non-differentiable reward models. And similar to weighted decoding, our method freezes the language model and manipulates the output token distribution using called critic, improving training efficiency and stability. Evaluation of our method on three controlled generation tasks, namely topic control, sentiment control, and detoxification, shows that our approach generates more coherent and well-controlled texts than previous methods. In addition, CriticControl demonstrates superior generalization ability in zero-shot settings. Human evaluation studies also corroborate our findings.

cs.CL↗

arXiv2022

Masked Summarization to Generate Factually Inconsistent Summaries for Improved Factual Consistency Checking

Despite the recent advances in abstractive summarization systems, it is still difficult to determine whether a generated summary is factual consistent with the source text. To this end, the latest approach is to train a factual consistency classifier on factually consistent and inconsistent summaries. Luckily, the former is readily available as reference summaries in existing summarization datasets. However, generating the latter remains a challenge, as they need to be factually inconsistent, yet closely relevant to the source text to be effective. In this paper, we propose to generate factually inconsistent summaries using source texts and reference summaries with key information masked. Experiments on seven benchmark datasets demonstrate that factual consistency classifiers trained on summaries generated using our method generally outperform existing models and show a competitive correlation with human judgments. We also analyze the characteristics of the summaries generated using our method. We will release the pre-trained model and the code at https://github.com/hwanheelee1993/MFMA.

cs.CL↗

arXiv2022

Plug-and-Play Adaptation for Continuously-updated QA

Language models (LMs) have shown great potential as implicit knowledge bases (KBs). And for their practical use, knowledge in LMs need to be updated periodically. However, existing tasks to assess LMs' efficacy as KBs do not adequately consider multiple large-scale updates. To this end, we first propose a novel task--Continuously-updated QA (CuQA)--in which multiple large-scale updates are made to LMs, and the performance is measured with respect to the success in adding and updating knowledge while retaining existing knowledge. We then present LMs with plug-in modules that effectively handle the updates. Experiments conducted on zsRE QA and NQ datasets show that our method outperforms existing approaches. We find that our method is 4x more effective in terms of updates/forgets ratio, compared to a fine-tuning baseline.

cs.CL↗

arXiv2021

Reasoning Visual Dialog with Sparse Graph Learning and Knowledge Transfer

Visual dialog is a task of answering a sequence of questions grounded in an image using the previous dialog history as context. In this paper, we study how to address two fundamental challenges for this task: (1) reasoning over underlying semantic structures among dialog rounds and (2) identifying several appropriate answers to the given question. To address these challenges, we propose a Sparse Graph Learning (SGL) method to formulate visual dialog as a graph structure learning task. SGL infers inherently sparse dialog structures by incorporating binary and score edges and leveraging a new structural loss function. Next, we introduce a Knowledge Transfer (KT) method that extracts the answer predictions from the teacher model and uses them as pseudo labels. We propose KT to remedy the shortcomings of single ground-truth labels, which severely limit the ability of a model to obtain multiple reasonable answers. As a result, our proposed model significantly improves reasoning capability compared to baseline methods and outperforms the state-of-the-art approaches on the VisDial v1.0 dataset. The source code is available at https://github.com/gicheonkang/SGLKT-VisDial.

cs.CV↗

arXiv2021

SUMBT+LaRL: Effective Multi-domain End-to-end Neural Task-oriented Dialog System

The recent advent of neural approaches for developing each dialog component in task-oriented dialog systems has remarkably improved, yet optimizing the overall system performance remains a challenge. Besides, previous research on modeling complicated multi-domain goal-oriented dialogs in end-to-end fashion has been limited. In this paper, we present an effective multi-domain end-to-end trainable neural dialog system SUMBT+LaRL that incorporates two previous strong models and facilitates them to be fully differentiable. Specifically, the SUMBT+ estimates user-acts as well as dialog belief states, and the LaRL models latent system action spaces and generates responses given the estimated contexts. We emphasize that the training framework of three steps significantly and stably increase dialog success rates: separately pretraining the SUMBT+ and LaRL, fine-tuning the entire system, and then reinforcement learning of dialog policy. We also introduce new reward criteria of reinforcement learning for dialog policy training. Then, we discuss experimental results depending on the reward criteria and different dialog evaluation methods. Consequently, our model achieved the new state-of-the-art success rate of 85.4% on corpus-based evaluation, and a comparable success rate of 81.40% on simulator-based evaluation provided by the DSTC8 challenge. To our best knowledge, our work is the first comprehensive study of a modularized E2E multi-domain dialog system that learning from each component to the entire dialog policy for task success.

cs.CL↗

arXiv2019

SUMBT: Slot-Utterance Matching for Universal and Scalable Belief Tracking

In goal-oriented dialog systems, belief trackers estimate the probability distribution of slot-values at every dialog turn. Previous neural approaches have modeled domain- and slot-dependent belief trackers, and have difficulty in adding new slot-values, resulting in lack of flexibility of domain ontology configurations. In this paper, we propose a new approach to universal and scalable belief tracker, called slot-utterance matching belief tracker (SUMBT). The model learns the relations between domain-slot-types and slot-values appearing in utterances through attention mechanisms based on contextual semantic vectors. Furthermore, the model predicts slot-value labels in a non-parametric way. From our experiments on two dialog corpora, WOZ 2.0 and MultiWOZ, the proposed model showed performance improvement in comparison with slot-dependent methods and achieved the state-of-the-art joint accuracy.

cs.CL↗

arXiv2018

Unpaired Speech Enhancement by Acoustic and Adversarial Supervision for Speech Recognition

Many speech enhancement methods try to learn the relationship between noisy and clean speech, obtained using an acoustic room simulator. We point out several limitations of enhancement methods relying on clean speech targets; the goal of this work is proposing an alternative learning algorithm, called acoustic and adversarial supervision (AAS). AAS makes the enhanced output both maximizing the likelihood of transcription on the pre-trained acoustic model and having general characteristics of clean speech, which improve generalization on unseen noisy speeches. We employ the connectionist temporal classification and the unpaired conditional boundary equilibrium generative adversarial network as the loss function of AAS. AAS is tested on two datasets including additive noise without and with reverberation, Librispeech + DEMAND and CHiME-4. By visualizing the enhanced speech with different loss combinations, we demonstrate the role of each supervision. AAS achieves a lower word error rate than other state-of-the-art methods using the clean speech target in both datasets.

cs.CL↗

arXiv2016

Deep CNNs along the Time Axis with Intermap Pooling for Robustness to Spectral Variations

Convolutional neural networks (CNNs) with convolutional and pooling operations along the frequency axis have been proposed to attain invariance to frequency shifts of features. However, this is inappropriate with regard to the fact that acoustic features vary in frequency. In this paper, we contend that convolution along the time axis is more effective. We also propose the addition of an intermap pooling (IMP) layer to deep CNNs. In this layer, filters in each group extract common but spectrally variant features, then the layer pools the feature maps of each group. As a result, the proposed IMP CNN can achieve insensitivity to spectral variations characteristic of different speakers and utterances. The effectiveness of the IMP CNN architecture is demonstrated on several LVCSR tasks. Even without speaker adaptation techniques, the architecture achieved a WER of 12.7% on the SWB part of the Hub5'2000 evaluation test set, which is competitive with other state-of-the-art methods.

cs.CL↗

arXiv2016

Compositional Sentence Representation from Character within Large Context Text

This paper describes a Hierarchical Composition Recurrent Network (HCRN) consisting of a 3-level hierarchy of compositional models: character, word and sentence. This model is designed to overcome two problems of representing a sentence on the basis of a constituent word sequence. The first is a data-sparsity problem in word embedding, and the other is a no usage of inter-sentence dependency. In the HCRN, word representations are built from characters, thus resolving the data-sparsity problem, and inter-sentence dependency is embedded into sentence representation at the level of sentence composition. We adopt a hierarchy-wise learning scheme in order to alleviate the optimization difficulties of learning deep hierarchical recurrent network in end-to-end fashion. The HCRN was quantitatively and qualitatively evaluated on a dialogue act classification task. Especially, sentence representations with an inter-sentence dependency are able to capture both implicit and explicit semantics of sentence, significantly improving performance. In the end, the HCRN achieved state-of-the-art performance with a test error rate of 22.7% for dialogue act classification on the SWBD-DAMSL database.

cs.CL↗