Human preference alignment is essential to improve the interaction quality of large language models (LLMs). Existing alignment methods depend on manually annotated preference data to guide the LLM optimization directions. However, continuously updating LLMs for alignment raises a distribution gap between model-generated samples and human-annotated responses, hindering training effectiveness. To mitigate this issue, previous methods require additional preference annotation on newly generated samples to adapt to the shifted distribution, which consumes a large amount of annotation resources. Targeting more efficient human preference optimization, we propose an Adversarial Preference Optimization (APO) framework, in which the LLM and the reward model update alternatively via a min-max game. Through adversarial training, the reward model can adapt to the shifted generation distribution of the LLM without any additional annotation. With comprehensive experiments, we find the proposed adversarial training framework further enhances existing alignment baselines in terms of LLM helpfulness and harmlessness. The code is at https://github.com/Linear95/APO.

Aligning large language models (LLMs) with human preferences has been recognized as the key to improving LLMs’ interaction quality. However, in this pluralistic world, human preferences can be diversified due to annotators’ different tastes, which hinders the effectiveness of LLM alignment methods. This paper presents the first quantitative analysis of the experimental scaling law for reward models with varying sizes, from 1.3 billion to 7 billion parameters, trained with human feedback exhibiting diverse preferences. Our analysis reveals that the impact of diversified human preferences depends on both model size and data size. Larger models with sufficient capacity mitigate the negative effects of diverse preferences, while smaller models struggle to accommodate them. To mitigate the impact of diverse preferences, we introduce a new metric, Expected Calibration Error (ECE), to evaluate RMs and show their obvious positive correlation with the alignment performance of LLMs. Furthermore, we propose a Multi-Objective Reward learning method (MORE) to enhance the calibration performance of RMs on shared preferences. Through experiments on four models and five human preference datasets, we find the calibration error can be adopted as a key metric for evaluating RMs and MORE can obtain superior alignment performance.

The rapid advancement of large language models (LLMs) has led to a new era marked by the development of autonomous applications in real-world scenarios, which drives innovation in creating advanced web agents. Existing web agents typically only handle one input modality and are evaluated only in simplified web simulators or static web snapshots, greatly limiting their applicability in real-world scenarios. To bridge this gap, we introduce WebVoyager, an innovative Large Multimodal Model (LMM) powered web agent that can complete user instructions end-to-end by interacting with real-world websites. Moreover, we establish a new benchmark by compiling real-world tasks from 15 popular websites and introduce an automatic evaluation protocol leveraging multimodal understanding abilities of GPT-4V to evaluate open-ended web agents. We show that WebVoyager achieves a 59.1% task success rate on our benchmark, significantly surpassing the performance of both GPT-4 (All Tools) and the WebVoyager (text-only) setups, underscoring the exceptional capability of WebVoyager. The proposed automatic evaluation metric achieves 85.3% agreement with human judgment, indicating its effectiveness in providing reliable and accurate assessments of web agents.

Although dominant in natural language processing, transformer-based models still struggle with long-sequence processing, due to the computational costs of their self-attention operations, which increase exponentially as the length of the input sequence grows. To address this challenge, we propose a **Sim**ple framework to enhance the long-content processing of off-the-shelf pre-trained transformers via three steps: **C**hunk, **A**lign, and **S**elect (SimCAS). More specifically, we first divide each long-sequence input into a batch of chunks, then align the inter-chunk information during the encoding steps, and finally, select the most representative hidden states from the encoder for the decoding process. With our SimCAS, the computation and memory costs can be reduced to linear complexity. In experiments, we demonstrate the effectiveness of the proposed method on various real-world long-text summarization and reading comprehension tasks, in which SimCAS significantly outperforms prior long-sequence processing baselines. The code is at [https://github.com/xjw-nlp/SimCAS](https://github.com/xjw-nlp/SimCAS).

With increasing concerns about data privacy, there is an increasing necessity of fine-tuning pre-trained language models (PLMs) for adapting to downstream tasks located in end-user devices or local clients without transmitting data to the central server. This urgent necessity therefore calls the research of investigating federated learning (FL) for PLMs. However, large PLMs bring the curse of prohibitive communication overhead and local model adaptation costs for the FL system. To this end, we investigate the parameter-efficient tuning (PETuning) of PLMs and develop a corresponding federated benchmark for four representative PETuning methods, dubbed FedPETuning. Specifically, FedPETuning provides the first holistic empirical study of representative PLMs tuning methods in FL, covering privacy attacks, performance comparisons, and resource-constrained analysis. Intensive experimental results have indicated that FedPETuning can efficiently defend against privacy attacks and maintains acceptable performance with reducing heavy resource consumption. The open-source code and data are available at https://github.com/SMILELab-FL/FedPETuning.

While GPT has become the de-facto method for text generation tasks, its application to pinyin input method remains unexplored. In this work, we make the first exploration to leverage Chinese GPT for pinyin input method. We find that a frozen GPT achieves state-of-the-art performance on perfect pinyin. However, the performance drops dramatically when the input includes abbreviated pinyin.A reason is that an abbreviated pinyin can be mapped to many perfect pinyin, which links to even larger number of Chinese characters. We mitigate this issue with two strategies,including enriching the context with pinyin and optimizing the training process to help distinguish homophones. To further facilitate the evaluation of pinyin input method, we create a dataset consisting of 270K instances from fifteen domains. Results show that our approach improves the performance on abbreviated pinyin across all domains. Model analysis demonstrates that both strategiescontribute to the performance boost.

Whole word masking (WWM), which masks all subwords corresponding to a word at once, makes a better English BERT model. For the Chinese language, however, there is no subword because each token is an atomic character. The meaning of a word in Chinese is different in that a word is a compositional unit consisting of multiple characters. Such difference motivates us to investigate whether WWM leads to better context understanding ability for Chinese BERT. To achieve this, we introduce two probing tasks related to grammatical error correction and ask pretrained models to revise or insert tokens in a masked language modeling manner. We construct a dataset including labels for 19,075 tokens in 10,448 sentences. We train three Chinese BERT models with standard character-level masking (CLM), WWM, and a combination of CLM and WWM, respectively. Our major findings are as follows: First, when one character needs to be inserted or replaced, the model trained with CLM performs the best. Second, when more than one character needs to be handled, WWM is the key to better performance. Finally, when being fine-tuned on sentence-level downstream tasks, models trained with different masking strategies perform comparably.

Aspect category detection (ACD) aims to automatically identify user-concerned aspects from online reviews, which is of great value for evaluating the fine-grained performance of a product. The most recent solutions tackle this problem via weakly supervised methods, achieving remarkable improvement over unsupervised methods. However, a closer look at these methods reveals that the required human efforts are nontrivial and can sometimes be hard to obtain. In this study, we explore the possibility of minimizing human guidance while improving detection performance, with a deep clustering method that relies merely on the category name of each aspect and a pretrained language model (LM). The LM, combined with prompt techniques, is employed as a knowledge base to automatically generate constraints for clustering, as well as to provide a representation space to perform the clustering. Our method (1) extracts extensive keywords to expand our understanding of each aspect, (2) automatically generates instance-level and concept-level constraints for clustering, and (3) trains the clustering model with the above constraints. We demonstrate the capability of the proposed framework through extensive experiments on nine benchmark datasets. Our model not only performs noticeably better than existing unsupervised approaches but also considerably surpasses weakly supervised methods that require more human efforts.

In this paper, we develop Sindhi subjective lexicon using a merger of existing English resources: NRC lexicon, list of opinion words, SentiWordNet, Sindhi-English bilingual dictionary, and collection of Sindhi modifiers. The positive or negative sentiment score is assigned to each Sindhi opinion word. Afterwards, we determine the coverage of the proposed lexicon with subjectivity analysis. Moreover, we crawl multi-domain tweet corpus of news, sports, and finance. The crawled corpus is annotated by experienced annotators using the Doccano text annotation tool. The sentiment annotated corpus is evaluated by employing support vector machine (SVM), recurrent neural network (RNN) variants, and convolutional neural network (CNN).

Incorporating knowledge bases (KB) into end-to-end task-oriented dialogue systems is challenging, since it requires to properly represent the entity of KB, which is associated with its KB context and dialogue context. The existing works represent the entity with only perceiving a part of its KB context, which can lead to the less effective representation due to the information loss, and adversely favor KB reasoning and response generation. To tackle this issue, we explore to fully contextualize the entity representation by dynamically perceiving all the relevant entities and dialogue history. To achieve this, we propose a COntext-aware Memory Enhanced Transformer framework (COMET), which treats the KB as a sequence and leverages a novel Memory Mask to enforce the entity to only focus on its relevant entities and dialogue history, while avoiding the distraction from the irrelevant entities. Through extensive experiments, we show that our COMET framework can achieve superior performance over the state of the arts.