Examining the Inter-Consistency of Large Language Models: An In-depth Analysis via Debate

Large Language Models (LLMs) have demonstrated human-like intelligence and are widely used in various applications. However, LLMs still exhibit various kinds of inconsistency problems. Existing works mainly focus on the inconsistency issues within a single LLM, while we investigate the inter-consistency among multiple LLMs, which is critical for collaborating to solve a complex task. To examine whether LLMs can collaborate to ultimately achieve a consensus for the shared goal and whether LLMs easily change their viewpoints, we introduce a Formal Debate framework (FORD) With FORD, we conduct a three-stage debate aligned with real-world scenarios: fair debate, mismatched debate, and roundtable debate. Through extensive experiments on the commonsense reasoning task, LLMs not only become more inter-consistent but also achieve higher performance. Moreover, we observe that stronger LLMs tend to dominate the debates by adhering to their perspectives, while weaker ones are more likely to change viewpoints. Additionally, we highlight the importance of a competent judge, such as GPT-4, to draw more proper conclusions. Our work contributes to understanding the inter-consistency among LLMs and lays the foundation for the development of future collaboration methods.

Is ChatGPT a Good Causal Reasoner? A Comprehensive Evaluation

Causal reasoning ability is crucial for numerous NLP applications. Despite the impressive emerging ability of ChatGPT in various NLP tasks, it is unclear how well ChatGPT performs in causal reasoning. In this paper, we conduct the first comprehensive evaluation of the ChatGPT's causal reasoning capabilities. Experiments show that ChatGPT is not a good causal reasoner, but a good causal interpreter. Besides, ChatGPT has a serious hallucination on causal reasoning, possibly due to the reporting biases between causal and non-causal relationships in natural language, as well as ChatGPT's upgrading processes, such as RLHF. The In-Context Learning (ICL) and Chain-of-Though (COT) techniques can further exacerbate such causal hallucination. Additionally, the causal reasoning ability of ChatGPT is sensitive to the words used to express the causal concept in prompts, and close-ended prompts perform better than open-ended prompts. For events in sentences, ChatGPT excels at capturing explicit causality rather than implicit causality, and performs better in sentences with lower event density and smaller lexical distance between events.

NoisywikiHow: A Benchmark for Learning with Real-world Noisy Labels in Natural Language Processing

Large-scale datasets in the real world inevitably involve label noise. Deep models can gradually overfit noisy labels and thus degrade model generalization. To mitigate the effects of label noise, learning with noisy labels (LNL) methods are designed to achieve better generalization performance. Due to the lack of suitable datasets, previous studies have frequently employed synthetic label noise to mimic real-world label noise. However, synthetic noise is not instance-dependent, making this approximation not always effective in practice. Recent research has proposed benchmarks for learning with real-world noisy labels. However, the noise sources within may be single or fuzzy, making benchmarks different from data with heterogeneous label noises in the real world. To tackle these issues, we contribute NoisywikiHow, the largest NLP benchmark built with minimal supervision. Specifically, inspired by human cognition, we explicitly construct multiple sources of label noise to imitate human errors throughout the annotation, replicating real-world noise, whose corruption is affected by both ground-truth labels and instances. Moreover, we provide a variety of noise levels to support controlled experiments on noisy data, enabling us to evaluate LNL methods systematically and comprehensively. After that, we conduct extensive multi-dimensional experiments on a broad range of LNL methods, obtaining new and intriguing findings.

CSED: A Chinese Semantic Error Diagnosis Corpus

Recently, much Chinese text error correction work has focused on Chinese Spelling Check (CSC) and Chinese Grammatical Error Diagnosis (CGED). In contrast, little attention has been paid to the complicated problem of Chinese Semantic Error Diagnosis (CSED), which lacks relevant datasets. The study of semantic errors is important because they are very common and may lead to syntactic irregularities or even problems of comprehension. To investigate this, we build the CSED corpus, which includes two datasets. The one is for the CSED-Recognition (CSED-R) task. The other is for the CSED-Correction (CSED-C) task. Our annotation guarantees high-quality data through quality assurance mechanisms. Our experiments show that powerful pre-trained models perform poorly on this corpus. We also find that the CSED task is challenging, as evidenced by the fact that even humans receive a low score. This paper proposes syntax-aware models to specifically adapt to the CSED task. The experimental results show that the introduction of the syntax-aware approach is meaningful.

Learning Robust Deep Equilibrium Models

Deep equilibrium (DEQ) models have emerged as a promising class of implicit layer models in deep learning, which abandon traditional depth by solving for the fixed points of a single nonlinear layer. Despite their success, the stability of the fixed points for these models remains poorly understood. Recently, Lyapunov theory has been applied to Neural ODEs, another type of implicit layer model, to confer adversarial robustness. By considering DEQ models as nonlinear dynamic systems, we propose a robust DEQ model named LyaDEQ with guaranteed provable stability via Lyapunov theory. The crux of our method is ensuring the fixed points of the DEQ models are Lyapunov stable, which enables the LyaDEQ models to resist minor initial perturbations. To avoid poor adversarial defense due to Lyapunov-stable fixed points being located near each other, we add an orthogonal fully connected layer after the Lyapunov stability module to separate different fixed points. We evaluate LyaDEQ models on several widely used datasets under well-known adversarial attacks, and experimental results demonstrate significant improvement in robustness. Furthermore, we show that the LyaDEQ model can be combined with other defense methods, such as adversarial training, to achieve even better adversarial robustness.

Learning Robust Visual-Semantic Embedding for Generalizable Person Re-identification

Generalizable person re-identification (Re-ID) is a very hot research topic in machine learning and computer vision, which plays a significant role in realistic scenarios due to its various applications in public security and video surveillance. However, previous methods mainly focus on the visual representation learning, while neglect to explore the potential of semantic features during training, which easily leads to poor generalization capability when adapted to the new domain. In this paper, we propose a Multi-Modal Equivalent Transformer called MMET for more robust visual-semantic embedding learning on visual, textual and visual-textual tasks respectively. To further enhance the robust feature learning in the context of transformer, a dynamic masking mechanism called Masked Multimodal Modeling strategy (MMM) is introduced to mask both the image patches and the text tokens, which can jointly works on multimodal or unimodal data and significantly boost the performance of generalizable person Re-ID. Extensive experiments on benchmark datasets demonstrate the competitive performance of our method over previous approaches. We hope this method could advance the research towards visual-semantic representation learning. Our source code is also publicly available at https://github.com/JeremyXSC/MMET.

HuaTuo: Tuning LLaMA Model with Chinese Medical Knowledge

Large Language Models (LLMs), such as the LLaMA model, have demonstrated their effectiveness in various general-domain natural language processing (NLP) tasks. Nevertheless, LLMs have not yet performed optimally in biomedical domain tasks due to the need for medical expertise in the responses. In response to this challenge, we propose HuaTuo, a LLaMA-based model that has been supervised-fine-tuned with generated QA (Question-Answer) instances. The experimental results demonstrate that HuaTuo generates responses that possess more reliable medical knowledge. Our proposed HuaTuo model is accessible at https://github.com/SCIR-HI/Huatuo-Llama-Med-Chinese.

Monte Carlo Linear Clustering with Single-Point Supervision is Enough for Infrared Small Target Detection

Single-frame infrared small target (SIRST) detection aims at separating small targets from clutter backgrounds on infrared images. Recently, deep learning based methods have achieved promising performance on SIRST detection, but at the cost of a large amount of training data with expensive pixel-level annotations. To reduce the annotation burden, we propose the first method to achieve SIRST detection with single-point supervision. The core idea of this work is to recover the per-pixel mask of each target from the given single point label by using clustering approaches, which looks simple but is indeed challenging since targets are always insalient and accompanied with background clutters. To handle this issue, we introduce randomness to the clustering process by adding noise to the input images, and then obtain much more reliable pseudo masks by averaging the clustered results. Thanks to this "Monte Carlo" clustering approach, our method can accurately recover pseudo masks and thus turn arbitrary fully supervised SIRST detection networks into weakly supervised ones with only single point annotation. Experiments on four datasets demonstrate that our method can be applied to existing SIRST detection networks to achieve comparable performance with their fully supervised counterparts, which reveals that single-point supervision is strong enough for SIRST detection. Our code will be available at: https://github.com/YeRen123455/SIRST-Single-Point-Supervision.

Spatiotemporally Discriminative Video-Language Pre-Training with Text Grounding

Most of existing video-language pre-training methods focus on instance-level alignment between video clips and captions via global contrastive learning but neglect rich fine-grained local information, which is of importance to downstream tasks requiring temporal localization and semantic reasoning. In this work, we propose a simple yet effective video-language pre-training framework, namely G-ViLM, to learn discriminative spatiotemporal features. Two novel designs involving spatiotemporal grounding and temporal grouping promote learning local region-noun alignment and temporal-aware features simultaneously. Specifically, spatiotemporal grounding aggregates semantically similar video tokens and aligns them with noun phrases extracted from the caption to promote local region-noun correspondences. Moreover, temporal grouping leverages cut-and-paste to manually create temporal scene changes and then learns distinguishable features from different scenes. Comprehensive evaluations demonstrate that G-ViLM performs favorably against existing approaches on four representative downstream tasks, covering text-video retrieval, video question answering, video action recognition and temporal action localization. G-ViLM performs competitively on all evaluated tasks and in particular achieves R@10 of 65.1 on zero-shot MSR-VTT retrieval, over 9% higher than the state-of-the-art method.

Steering Prototype with Prompt-tuning for Rehearsal-free Continual Learning

Prototype, as a representation of class embeddings, has been explored to reduce memory footprint or mitigate forgetting for continual learning scenarios. However, prototype-based methods still suffer from abrupt performance deterioration due to semantic drift and prototype interference. In this study, we propose Contrastive Prototypical Prompt (CPP) and show that task-specific prompt-tuning, when optimized over a contrastive learning objective, can effectively address both obstacles and significantly improve the potency of prototypes. Our experiments demonstrate that CPP excels in four challenging class-incremental learning benchmarks, resulting in 4% to 6% absolute improvements over state-of-the-art methods. Moreover, CPP does not require a rehearsal buffer and it largely bridges the performance gap between continual learning and offline joint-learning, showcasing a promising design scheme for continual learning systems under a Transformer architecture.