Long-context modeling capabilities of Large Language Models (LLMs) have garnered widespread attention, leading to the emergence of LLMs with ultra-context windows. Meanwhile, benchmarks for evaluating long-context language models are gradually catching up. However, existing benchmarks employ irrelevant noise texts to artificially extend the length of test cases, diverging from the real-world scenarios of long-context applications. To bridge this gap, we propose a novel long-context benchmark, Loong, aligning with realistic scenarios through extended multi-document question answering (QA). Unlike typical document QA, in Loong’s test cases, each document is relevant to the final answer, ignoring any document will lead to the failure of the answer. Furthermore, Loong introduces four types of tasks with a range of context lengths: Spotlight Locating, Comparison, Clustering, and Chain of Reasoning, to facilitate a more realistic and comprehensive evaluation of long-context understanding. Extensive experiments indicate that existing long-context language models still exhibit considerable potential for enhancement. Retrieval augmented generation (RAG) achieves poor performance, demonstrating that Loong can reliably assess the model’s long-context modeling capabilities.

The alignment problem in Large Language Models (LLMs) involves adapting them to the broad spectrum of human values. This requirement challenges existing alignment methods due to diversity of preferences and regulatory standards. This paper introduces a novel alignment paradigm, priority rule following, which defines rules as the primary control mechanism in each dialog, prioritizing them over user instructions. Our preliminary analysis reveals that even the advanced LLMs, such as GPT-4, exhibit shortcomings in understanding and prioritizing the rules. Therefore, we present PriorityDistill, a semi-automated approach for distilling priority following signals from LLM simulations to ensure robust rule integration and adherence. Our experiments show that this method not only effectively minimizes misalignments utilizing only one general rule but also adapts smoothly to various unseen rules, ensuring they are shielded from hijacking and that the model responds appropriately.

Large language models (LLMs) rely on safety alignment to avoid responding to malicious user inputs. Unfortunately, jailbreak can circumvent safety guardrails, resulting in LLMs generating harmful content and raising concerns about LLM safety. Due to language models with intensive parameters often regarded as black boxes, the mechanisms of alignment and jailbreak are challenging to elucidate. In this paper, we employ weak classifiers to explain LLM safety through the intermediate hidden states. We first confirm that LLMs learn ethical concepts during pre-training rather than alignment and can identify malicious and normal inputs in the early layers. Alignment actually associates the early concepts with emotion guesses in the middle layers and then refines them to the specific reject tokens for safe generations. Jailbreak disturbs the transformation of early unethical classification into negative emotions. We conduct experiments on models from 7B to 70B across various model families to prove our conclusion. Overall, our paper indicates the intrinsical mechanism of LLM safety and how jailbreaks circumvent safety guardrails, offering a new perspective on LLM safety and reducing concerns.

This study evaluates the proficiency of Large Language Models (LLMs) in accurately labeling clinical document excerpts. Our focus is on the assignment of potential or confirmed diagnoses and medical procedures to snippets of medical text sourced from unstructured clinical patient records. We explore how the performance of LLMs compare against human annotators in classifying these excerpts. Employing a few-shot, chain-of-thought prompting approach with the MIMIC-III dataset, Med-PaLM 2 showcases annotation accuracy comparable to human annotators, achieving a notable precision rate of approximately 92% relative to the gold standard labels established by human experts.

Alignment with human preference prevents large language models (LLMs) from generating misleading or toxic content while requiring high-cost human feedback. Assuming resources of human annotation are limited, there are two different ways of allocating considered: more diverse PROMPTS or more diverse RESPONSES to be labeled. Nonetheless, a straightforward comparison between their impact is absent. In this work, we first control the diversity of both sides according to the number of samples for fine-tuning, which can directly reflect their influence. We find that instead of numerous prompts, more responses but fewer prompts better trigger LLMs for human alignment. Additionally, the concept of diversity for prompts can be more complex than responses that are typically quantified by single digits. Consequently, a new formulation of prompt diversity is proposed, further implying a linear correlation with the final performance of LLMs after fine-tuning. We also leverage it on data augmentation and conduct experiments to show its effect on different algorithms.

Training large language models (LLMs) with open-domain instruction data has yielded remarkable success in aligning to end tasks and human preferences. Extensive research has highlighted the importance of the quality and diversity of instruction data. However, the impact of data complexity, as a crucial metric, remains relatively unexplored from three aspects: (1)where the sustainability of performance improvements with increasing complexity is uncertain; (2)whether the improvement brought by complexity merely comes from introducing more training tokens; and (3)where the potential benefits of incorporating instructions from easy to difficult are not yet fully understood. In this paper, we propose Tree-Instruct to systematically enhance the instruction complexity in a controllable manner. By adding a specified number of nodes to instructions’ semantic trees, this approach not only yields new instruction data from the modified tree but also allows us to control the difficulty level of modified instructions. Our preliminary experiments reveal the following insights: (1)Increasing complexity consistently leads to sustained performance improvements of LLMs. (2)Under the same token budget, a few complex instructions outperform diverse yet simple instructions. (3)Curriculum instruction tuning might not yield the anticipated results; focusing on increasing complexity appears to be the key.

Cross-lingual semantic textual relatedness task is an important research task that addresses challenges in cross-lingual communication and text understanding. It helps establish semantic connections between different languages, crucial for downstream tasks like machine translation, multilingual information retrieval, and cross-lingual text understanding.Based on extensive comparative experiments, we choose the XLM-R-base as our base model and use pre-trained sentence representations based on whitening to reduce anisotropy.Additionally, for the given training data, we design a delicate data filtering method to alleviate the curse of multilingualism. With our approach, we achieve a 2nd score in Spanish, a 3rd in Indonesian, and multiple entries in the top ten results in the competition’s track C. We further do a comprehensive analysis to inspire future research aimed at improving performance on cross-lingual tasks.

Universal Information Extraction (Universal IE) aims to solve different extraction tasks in a uniform text-to-structure generation manner. Such a generation procedure tends to struggle when there exist complex information structures to be extracted. Retrieving knowledge from external knowledge bases may help models to overcome this problem but it is impossible to construct a knowledge base suitable for various IE tasks. Inspired by the fact that large amount of knowledge are stored in the pretrained language models (PLM) and can be retrieved explicitly, in this paper, we propose MetaRetriever to retrieve task-specific knowledge from PLMs to enhance universal IE. As different IE tasks need different knowledge, we further propose a Meta-Pretraining Algorithm which allows MetaRetriever to quicktly achieve maximum task-specific retrieval performance when fine-tuning on downstream IE tasks. Experimental results show that MetaRetriever achieves the new state-of-the-art on 4 IE tasks, 12 datasets under fully-supervised, low-resource and few-shot scenarios.

When trying to answer complex questions, people often rely on multiple sources of information, such as visual, textual, and tabular data. Previous approaches to this problem have focused on designing input features or model structure in the multi-modal space, which is inflexible for cross-modal reasoning or data-efficient training. In this paper, we call for an alternative paradigm, which transforms the images and tables into unified language representations, so that we can simplify the task into a simpler textual QA problem that can be solved using three steps: retrieval, ranking, and generation, all within a language space. This idea takes advantage of the power of pre-trained language models and is implemented in a framework called Solar. Our experimental results show that Solar outperforms all existing methods by 10.6-32.3 pts on two datasets, MultimodalQA and MMCoQA, across ten different metrics. Additionally, Solar achieves the best performance on the WebQA leaderboard.

This paper addresses the problem of generating questions from a given context and an answer, specifically focusing on questions that require multi-hop reasoning across an extended context. Previous studies have suggested that key phrase selection is essential for question generation (QG), yet it is still challenging to connect such disjointed phrases into meaningful questions, particularly for long context. To mitigate this issue, we propose MultiFactor, a novel QG framework based on multi-level content planning. Specifically, MultiFactor includes two components: FA-Model, which simultaneously selects key phrases and generates full answers, and Q-Model which takes the generated full answer as an additional input to generate questions. Here, full answer generation is introduced to connect the short answer with the selected key phrases, thus forming an answer-aware summary to facilitate QG. Both FA-Model and Q-Model are formalized as simple-yet-effective Phrase-Enhanced Transformers, our joint model for phrase selection and text generation. Experimental results show that our method outperforms strong baselines on two popular QG datasets. Our code is available at https://github.com/zeaver/MultiFactor.

Given a textual passage and an answer, humans are able to ask questions with various expressions, but this ability is still challenging for most question generation (QG) systems. Existing solutions mainly focus on the internal knowledge within the given passage or the semantic word space for diverse content planning. These methods, however, have not considered the potential of external knowledge for expression diversity. To bridge this gap, we propose RAST, a framework for Retrieval-Augmented Style Transfer, where the objective is to utilize the style of diverse templates for question generation. For training RAST, we develop a novel Reinforcement Learning (RL) based approach that maximizes a weighted combination of diversity reward and consistency reward. Here, the consistency reward is computed by a Question-Answering (QA) model, whereas the diversity reward measures how much the final output mimics the retrieved template. Experimental results show that our method outperforms previous diversity-driven baselines on diversity while being comparable in terms of consistency scores. Our code is available at https://github.com/gouqi666/RAST.

Recent research has demonstrated that Large Language Models (LLMs) can enhance their capabilities by utilizing external tools. However, three pivotal questions remain unanswered: (1) How effective are current LLMs in utilizing tools? (2) How can we enhance LLMs’ ability to utilize tools? (3) What obstacles need to be overcome to leverage tools? To address these questions, we introduce API-Bank, a groundbreaking benchmark, specifically designed for tool-augmented LLMs. For the first question, we develop a runnable evaluation system consisting of 73 API tools. We annotate 314 tool-use dialogues with 753 API calls to assess the existing LLMs’ capabilities in planning, retrieving, and calling APIs. For the second question, we construct a comprehensive training set containing 1,888 tool-use dialogues from 2,138 APIs spanning 1,000 distinct domains. Using this dataset, we train Lynx, a tool-augmented LLM initialized from Alpaca. Experimental results demonstrate that GPT-3.5 exhibits improved tool utilization compared to GPT-3, while GPT-4 excels in planning. However, there is still significant potential for further improvement. Moreover, Lynx surpasses Alpaca’s tool utilization performance by more than 26 pts and approaches the effectiveness of GPT-3.5. Through error analysis, we highlight the key challenges for future research in this field to answer the third question.

The goal of document-grounded dialogue (DocGD) is to generate a response by anchoring the evidence in a supporting document in accordance with the dialogue context. This entails four causally interconnected variables. While task-specific pre-training has significantly enhanced performances on numerous downstream tasks, existing DocGD methods still rely on general pre-trained language models without a specifically tailored pre-training approach that explicitly captures the causal relationships. To address this, we present the first causally-complete dataset construction strategy for developing million-scale DocGD pre-training corpora. Additionally, we propose a causally-perturbed pre-training strategy to better capture causality by introducing perturbations on the variables and optimizing the overall causal effect. Experiments conducted on three benchmark datasets demonstrate that our causal pre-training yields substantial and consistent improvements in fully-supervised, low-resource, few-shot, and zero-shot settings.

In the area of data augmentation research, the main focus to date has been on the improvement of the generation models, while the examination and improvements to synthetic data evaluation methods remains less explored. In our work, we explore a number of sentence similarity measures in the context of data generation filtering, and evaluate their impact on the performance of the targeted Natural Language Understanding problem on the example of the intent classification and named entity recognition tasks. Our experiments on ATIS dataset show that the right choice of filtering technique can bring up to 33% in sentence accuracy improvement for targeted underrepresented intents.

We present an open-source and extensible knowledge extraction toolkit DeepKE, supporting complicated low-resource, document-level and multimodal scenarios in the knowledge base population. DeepKE implements various information extraction tasks, including named entity recognition, relation extraction and attribute extraction. With a unified framework, DeepKE allows developers and researchers to customize datasets and models to extract information from unstructured data according to their requirements. Specifically, DeepKE not only provides various functional modules and model implementation for different tasks and scenarios but also organizes all components by consistent frameworks to maintain sufficient modularity and extensibility. We release the source code at GitHub in https://github.com/zjunlp/DeepKE with Google Colab tutorials and comprehensive documents for beginners. Besides, we present an online system in http://deepke.openkg.cn/EN/re_doc_show.html for real-time extraction of various tasks, and a demo video.

Open Information Extraction (OpenIE) facilitates the open-domain discovery of textual facts. However, the prevailing solutions evaluate OpenIE models on in-domain test sets aside from the training corpus, which certainly violates the initial task principle of domain-independence. In this paper, we propose to advance OpenIE towards a more realistic scenario: generalizing over unseen target domains with different data distributions from the source training domains, termed Generalized OpenIE. For this purpose, we first introduce GLOBE, a large-scale human-annotated multi-domain OpenIE benchmark, to examine the robustness of recent OpenIE models to domain shifts, and the relative performance degradation of up to 70% implies the challenges of generalized OpenIE. Then, we propose DragonIE, which explores a minimalist expression of textual fact: directed acyclic graph, to improve the OpenIE generalization ability. Extensive experiments demonstrate that DragonIE beats the previous methods in both in-domain and out-of-domain settings by as much as 6.0% in F1 score absolutely, but there is still ample room for improvement.

This paper introduces Doc2Bot, a novel dataset for building machines that help users seek information via conversations. This is of particular interest for companies and organizations that own a large number of manuals or instruction books. Despite its potential, the nature of our task poses several challenges: (1) documents contain various structures that hinder the ability of machines to comprehend, and (2) user information needs are often underspecified. Compared to prior datasets that either focus on a single structural type or overlook the role of questioning to uncover user needs, the Doc2Bot dataset is developed to target such challenges systematically. Our dataset contains over 100,000 turns based on Chinese documents from five domains, larger than any prior document-grounded dialog dataset for information seeking. We propose three tasks in Doc2Bot: (1) dialog state tracking to track user intentions, (2) dialog policy learning to plan system actions and contents, and (3) response generation which generates responses based on the outputs of the dialog policy. Baseline methods based on the latest deep learning models are presented, indicating that our proposed tasks are challenging and worthy of further research.

Current supervised relational triple extraction approaches require huge amounts of labeled data and thus suffer from poor performance in few-shot settings. However, people can grasp new knowledge by learning a few instances. To this end, we take the first step to study the few-shot relational triple extraction, which has not been well understood. Unlike previous single-task few-shot problems, relational triple extraction is more challenging as the entities and relations have implicit correlations. In this paper, We propose a novel multi-prototype embedding network model to jointly extract the composition of relational triples, namely, entity pairs and corresponding relations. To be specific, we design a hybrid prototypical learning mechanism that bridges text and knowledge concerning both entities and relations. Thus, implicit correlations between entities and relations are injected. Additionally, we propose a prototype-aware regularization to learn more representative prototypes. Experimental results demonstrate that the proposed method can improve the performance of the few-shot triple extraction.

Natural language processing covers a wide variety of tasks with token-level or sentence-level understandings. In this paper, we provide a simple insight that most tasks can be represented in a single universal extraction format. We introduce a prototype model and provide an open-source and extensible toolkit called OpenUE for various extraction tasks. OpenUE allows developers to train custom models to extract information from the text and supports quick model validation for researchers. Besides, OpenUE provides various functional modules to maintain sufficient modularity and extensibility. Except for the toolkit, we also deploy an online demo with restful APIs to support real-time extraction without training and deploying. Additionally, the online system can extract information in various tasks, including relational triple extraction, slot & intent detection, event extraction, and so on. We release the source code, datasets, and pre-trained models to promote future researches in http://github.com/zjunlp/openue.