Transformer-based large language models (LLMs) exhibit limitations such as generating unsafe responses, unreliable reasoning, etc. Existing inference intervention approaches attempt to mitigate these issues by finetuning additional models to produce calibration signals (such as rewards) that guide the LLM’s decoding process. However, this solution introduces substantial time and space overhead due to the separate models required. This work proposes Non-disruptive parameters insertion (Otter), inserting extra parameters into the transformer architecture to predict calibration signals along with the original LLM output. Otter offers state-of-the-art performance on multiple demanding tasks while saving up to 86.5% extra space and 98.5% extra time. Furthermore, Otter seamlessly integrates with existing inference engines, requiring only a one-line code change, and the original model response remains accessible after the parameter insertion.

In this paper, we begin by illustrating that, when presented with a query, Large Language Models (LLMs) capable of providing accurate responses tend to exhibit a more uniform probability distribution compared to their less proficient counterparts. Building upon this observation, we introduce a novel self-assessment criterion termed ProbDiff for evaluating the performance of diverse LLMs. This method eliminates the need for training an additional evaluation model or relying on external proprietary models such as GPT-4 as a judger. Instead, it solely relies on the LLMs under evaluation to compute the probability discrepancy between the original response generation and its revised versions. A higher discrepancy in two LLMs for the same query suggests a relatively weaker ability. We discover that ProbDiff yields comparable results to mainstream GPT-4-based evaluations on various scenarios including NLG tasks like translation and summarization, as well as LLM evaluation benchmarks such as AlignBench, MT-Bench, and AlpacaEval, across LLMs of different sizes.

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.

Information extraction (IE) encounters challenges due to the variety of schemas and objectives that differ across tasks. Recent advancements hint at the potential for universal approaches to model such tasks, referred to as Universal Information Extraction (UIE). While handling diverse tasks in one model, their generalization is limited since they are actually learning task-specific knowledge.In this study, we introduce an innovative paradigm known as TRUE-UIE, wherein all IE tasks are aligned to learn the same goals: extracting mention spans and two universal relations named NEXT and IS. During the decoding process, the NEXT relation is utilized to group related elements, while the IS relation, in conjunction with structured language prompts, undertakes the role of type recognition. Additionally, we consider the sequential dependency of tokens during span extraction, an aspect often overlooked in prevalent models.Our empirical experiments indicate that TRUE-UIE achieves state-of-the-art performance on established benchmarks encompassing 16 datasets, spanning 7 diverse IE tasks. Further evaluations reveal that our approach effectively share knowledge between different IE tasks, showcasing significant transferability in zero-shot and few-shot scenarios.

Considerable efforts have been invested in augmenting the role-playing proficiency of open-source large language models (LLMs) by emulating proprietary counterparts. Nevertheless, we posit that LLMs inherently harbor role-play capabilities, owing to the extensive knowledge of characters and potential dialogues ingrained in their vast training corpora. Thus, we introduce Ditto, the first self-alignment method for role-play, which encourages an instruction-following LLM to simulate role-play dialogues as a variant of reading comprehension, and creates a role-play training set comprising 4000 characters, surpassing the scale of currently available datasets by tenfold regarding the number of roles. Subsequently, we fine-tune the LLM using this self-generated dataset to augment its role-playing capabilities. Upon evaluating our meticulously constructed role-play benchmark and the roleplay subset of MT-Bench, Ditto, in various parameter scales, consistently maintains a consistent role identity and provides accurate role-specific knowledge in multi-turn role-play conversations, outperforming all open-source role-play baselines. Furthermore, we present the first cross-supervision role-play experiment, revealing that the role-play styles can be easily acquired, while the intrinsic capabilities of LLMs confine the knowledge within role-play.

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.

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.

Lifelong learning (LL) is an important ability for NLP models to learn new tasks continuously. Architecture-based approaches are reported to be effective implementations for LL models. However, it is non-trivial to extend previous approaches to domain incremental LL scenarios since they either require access to task identities in the testing phase or cannot handle samples from unseen tasks. In this paper, we propose Diana: a dynamic architecture-based lifelong learning model that tries to learn a sequence of tasks with a prompt-enhanced language model. Four types of hierarchically organized prompts are used in Diana to capture knowledge from different granularities. Specifically, we dedicate task-level prompts to capture task-specific knowledge to retain high LL performances and maintain instance-level prompts to learn knowledge shared across input samples to improve the model’s generalization performance. Moreover, we dedicate separate prompts to explicitly model unseen tasks and introduce a set of prompt key vectors to facilitate knowledge sharing between tasks. Extensive experiments demonstrate that Diana outperforms state-of-the-art LL models, especially in handling unseen tasks.

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.

Chinese pre-trained language models usually exploit contextual character information to learn representations, while ignoring the linguistics knowledge, e.g., word and sentence information. Hence, we propose a task-free enhancement module termed as Heterogeneous Linguistics Graph (HLG) to enhance Chinese pre-trained language models by integrating linguistics knowledge. Specifically, we construct a hierarchical heterogeneous graph to model the characteristics linguistics structure of Chinese language, and conduct a graph-based method to summarize and concretize information on different granularities of Chinese linguistics hierarchies. Experimental results demonstrate our model has the ability to improve the performance of vanilla BERT, BERTwwm and ERNIE 1.0 on 6 natural language processing tasks with 10 benchmark datasets. Further, the detailed experimental analyses have proven that this kind of modelization achieves more improvements compared with previous strong baseline MWA. Meanwhile, our model introduces far fewer parameters (about half of MWA) and the training/inference speed is about 7x faster than MWA.

Multi-intent detection and slot filling joint model attracts more and more attention since it can handle multi-intent utterances, which is closer to complex real-world scenarios. Most existing joint models rely entirely on the training procedure to obtain the implicit correlation between intents and slots. However, they ignore the fact that leveraging the rich global knowledge in the corpus can determine the intuitive and explicit correlation between intents and slots. In this paper, we aim to make full use of the statistical co-occurrence frequency between intents and slots as prior knowledge to enhance joint multiple intent detection and slot filling. To be specific, an intent-slot co-occurrence graph is constructed based on the entire training corpus to globally discover correlation between intents and slots. Based on the global intent-slot co-occurrence, we propose a novel graph neural network to model the interaction between the two subtasks. Experimental results on two public multi-intent datasets demonstrate that our approach outperforms the state-of-the-art models.

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.

Lifelong learning aims to accumulate knowledge and alleviate catastrophic forgetting when learning tasks sequentially. However, existing lifelong language learning methods only focus on the supervised learning setting. Unlabeled data, which can be easily accessed in real-world scenarios, are underexplored. In this paper, we explore a novel setting, semi-supervised lifelong language learning (SSLL), where a model learns sequentially arriving language tasks with both labeled and unlabeled data. We propose an unlabeled data enhanced lifelong learner to explore SSLL. Specially, we dedicate task-specific modules to alleviate catastrophic forgetting and design two modules to exploit unlabeled data: (1) a virtual supervision enhanced task solver is constructed on a teacher-student framework to mine the underlying knowledge from unlabeled data; and (2) a backward augmented learner is built to encourage knowledge transfer from newly arrived unlabeled data to previous tasks. Experimental results on various language tasks demonstrate our model’s effectiveness and superiority over competitive baselines under the new setting SSLL.

Named entity recognition (NER) remains challenging when entity mentions can be discontinuous. Existing methods break the recognition process into several sequential steps. In training, they predict conditioned on the golden intermediate results, while at inference relying on the model output of the previous steps, which introduces exposure bias. To solve this problem, we first construct a segment graph for each sentence, in which each node denotes a segment (a continuous entity on its own, or a part of discontinuous entities), and an edge links two nodes that belong to the same entity. The nodes and edges can be generated respectively in one stage with a grid tagging scheme and learned jointly using a novel architecture named Mac. Then discontinuous NER can be reformulated as a non-parametric process of discovering maximal cliques in the graph and concatenating the spans in each clique. Experiments on three benchmarks show that our method outperforms the state-of-the-art (SOTA) results, with up to 3.5 percentage points improvement on F1, and achieves 5x speedup over the SOTA model.

In this paper, we introduce FITAnnotator, a generic web-based tool for efficient text annotation. Benefiting from the fully modular architecture design, FITAnnotator provides a systematic solution for the annotation of a variety of natural language processing tasks, including classification, sequence tagging and semantic role annotation, regardless of the language. Three kinds of interfaces are developed to annotate instances, evaluate annotation quality and manage the annotation task for annotators, reviewers and managers, respectively. FITAnnotator also gives intelligent annotations by introducing task-specific assistant to support and guide the annotators based on active learning and incremental learning strategies. This assistant is able to effectively update from the annotator feedbacks and easily handle the incremental labeling scenarios.

Distantly supervised named entity recognition (DS-NER) efficiently reduces labor costs but meanwhile intrinsically suffers from the label noise due to the strong assumption of distant supervision. Typically, the wrongly labeled instances comprise numbers of incomplete and inaccurate annotations, while most prior denoising works are only concerned with one kind of noise and fail to fully explore useful information in the training set. To address this issue, we propose a robust learning paradigm named Self-Collaborative Denoising Learning (SCDL), which jointly trains two teacher-student networks in a mutually-beneficial manner to iteratively perform noisy label refinery. Each network is designed to exploit reliable labels via self denoising, and two networks communicate with each other to explore unreliable annotations by collaborative denoising. Extensive experimental results on five real-world datasets demonstrate that SCDL is superior to state-of-the-art DS-NER denoising methods.

Open Information Extraction (OpenIE) aims to discover textual facts from a given sentence. In essence, the facts contained in plain text are unordered. However, the popular OpenIE systems usually output facts sequentially in the way of predicting the next fact conditioned on the previous decoded ones, which enforce an unnecessary order on the facts and involve the error accumulation between autoregressive steps. To break this bottleneck, we propose MacroIE, a novel non-autoregressive framework for OpenIE. MacroIE firstly constructs a fact graph based on the table filling scheme, in which each node denotes a fact element, and an edge links two nodes that belong to the same fact. Then OpenIE can be reformulated as a non-parametric process of finding maximal cliques from the graph. It directly outputs the final set of facts in one go, thus getting rid of the burden of predicting fact order, as well as the error propagation between facts. Experiments conducted on two benchmark datasets show that our proposed model significantly outperforms current state-of-the-art methods, beats the previous systems by as much as 5.7 absolute gain in F1 score.

Distantly supervised named entity recognition (DS-NER) efficiently reduces labor costs but meanwhile intrinsically suffers from the label noise due to the strong assumption of distant supervision. Typically, the wrongly labeled instances comprise numbers of incomplete and inaccurate annotations, while most prior denoising works are only concerned with one kind of noise and fail to fully explore useful information in the training set. To address this issue, we propose a robust learning paradigm named Self-Collaborative Denoising Learning (SCDL), which jointly trains two teacher-student networks in a mutually-beneficial manner to iteratively perform noisy label refinery. Each network is designed to exploit reliable labels via self denoising, and two networks communicate with each other to explore unreliable annotations by collaborative denoising. Extensive experimental results on five real-world datasets demonstrate that SCDL is superior to state-of-the-art DS-NER denoising methods.

Extracting entities and relations from unstructured text has attracted increasing attention in recent years but remains challenging, due to the intrinsic difficulty in identifying overlapping relations with shared entities. Prior works show that joint learning can result in a noticeable performance gain. However, they usually involve sequential interrelated steps and suffer from the problem of exposure bias. At training time, they predict with the ground truth conditions while at inference it has to make extraction from scratch. This discrepancy leads to error accumulation. To mitigate the issue, we propose in this paper a one-stage joint extraction model, namely, TPLinker, which is capable of discovering overlapping relations sharing one or both entities while being immune from the exposure bias. TPLinker formulates joint extraction as a token pair linking problem and introduces a novel handshaking tagging scheme that aligns the boundary tokens of entity pairs under each relation type. Experiment results show that TPLinker performs significantly better on overlapping and multiple relation extraction, and achieves state-of-the-art performance on two public datasets.

Document-level relation extraction (RE) poses new challenges over its sentence-level counterpart since it requires an adequate comprehension of the whole document and the multi-hop reasoning ability across multiple sentences to reach the final result. In this paper, we propose a novel graph-based model with Dual-tier Heterogeneous Graph (DHG) for document-level RE. In particular, DHG is composed of a structure modeling layer followed by a relation reasoning layer. The major advantage is that it is capable of not only capturing both the sequential and structural information of documents but also mixing them together to benefit for multi-hop reasoning and final decision-making. Furthermore, we employ Graph Neural Networks (GNNs) based message propagation strategy to accumulate information on DHG. Experimental results demonstrate that the proposed method achieves state-of-the-art performance on two widely used datasets, and further analyses suggest that all the modules in our model are indispensable for document-level RE.

Incorporating lexicons into character-level Chinese NER by lattices is proven effective to exploitrich word boundary information. Previous work has extended RNNs to consume lattice inputsand achieved great success. However, due to the DAG structure and the inherently unidirectionalsequential nature, this method precludes batched computation and sufficient semantic interaction. In this paper, we propose PLTE, an extension of transformer encoder that is tailored for ChineseNER, which models all the characters and matched lexical words in parallel with batch process-ing. PLTE augments self-attention with positional relation representations to incorporate latticestructure. It also introduces a porous mechanism to augment localness modeling and maintainthe strength of capturing the rich long-term dependencies. Experimental results show that PLTEperforms up to 11.4 times faster than state-of-the-art methods while realizing better performance. We also demonstrate that using BERT representations further substantially boosts the performanceand brings out the best in PLTE.

Dependency trees have been shown to be effective in capturing long-range relations between target entities. Nevertheless, how to selectively emphasize target-relevant information and remove irrelevant content from the tree is still an open problem. Existing approaches employing pre-defined rules to eliminate noise may not always yield optimal results due to the complexity and variability of natural language. In this paper, we present a novel architecture named Dynamically Pruned Graph Convolutional Network (DP-GCN), which learns to prune the dependency tree with rethinking in an end-to-end scheme. In each layer of DP-GCN, we employ a selection module to concentrate on nodes expressing the target relation by a set of binary gates, and then augment the pruned tree with a pruned semantic graph to ensure the connectivity. After that, we introduce a rethinking mechanism to guide and refine the pruning operation by feeding back the high-level learned features repeatedly. Extensive experimental results demonstrate that our model achieves impressive results compared to strong competitors.

Most Chinese pre-trained models take character as the basic unit and learn representation according to character’s external contexts, ignoring the semantics expressed in the word, which is the smallest meaningful utterance in Chinese. Hence, we propose a novel word-aligned attention to exploit explicit word information, which is complementary to various character-based Chinese pre-trained language models. Specifically, we devise a pooling mechanism to align the character-level attention to the word level and propose to alleviate the potential issue of segmentation error propagation by multi-source information fusion. As a result, word and character information are explicitly integrated at the fine-tuning procedure. Experimental results on five Chinese NLP benchmark tasks demonstrate that our method achieves significant improvements against BERT, ERNIE and BERT-wwm.

Event detection (ED), a key subtask of information extraction, aims to recognize instances of specific event types in text. Previous studies on the task have verified the effectiveness of integrating syntactic dependency into graph convolutional networks. However, these methods usually ignore dependency label information, which conveys rich and useful linguistic knowledge for ED. In this paper, we propose a novel architecture named Edge-Enhanced Graph Convolution Networks (EE-GCN), which simultaneously exploits syntactic structure and typed dependency label information to perform ED. Specifically, an edge-aware node update module is designed to generate expressive word representations by aggregating syntactically-connected words through specific dependency types. Furthermore, to fully explore clues hidden from dependency edges, a node-aware edge update module is introduced, which refines the relation representations with contextual information. These two modules are complementary to each other and work in a mutual promotion way. We conduct experiments on the widely used ACE2005 dataset and the results show significant improvement over competitive baseline methods.

More recently, Named Entity Recognition hasachieved great advances aided by pre-trainingapproaches such as BERT. However, currentpre-training techniques focus on building lan-guage modeling objectives to learn a gen-eral representation, ignoring the named entity-related knowledge. To this end, we proposea NER-specific pre-training framework to in-ject coarse-to-fine automatically mined entityknowledge into pre-trained models. Specifi-cally, we first warm-up the model via an en-tity span identification task by training it withWikipedia anchors, which can be deemed asgeneral-typed entities. Then we leverage thegazetteer-based distant supervision strategy totrain the model extract coarse-grained typedentities. Finally, we devise a self-supervisedauxiliary task to mine the fine-grained namedentity knowledge via clustering.Empiricalstudies on three public NER datasets demon-strate that our framework achieves significantimprovements against several pre-trained base-lines, establishing the new state-of-the-art per-formance on three benchmarks. Besides, weshow that our framework gains promising re-sults without using human-labeled trainingdata, demonstrating its effectiveness in label-few and low-resource scenarios.