Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks57 from public data
- Parameter-efficient fine-tuning of large-scale pre-trained language models1,311
Different methods of fine-tuning with only a small number of parameters are compared on a large set of natural language processing tasks and demonstrates that large-scale models could be effectively stimulated by the optimization of a few parameters.
- KEPLER: A Unified Model for Knowledge Embedding and Pre-trained Language Representation883
A unified model for Knowledge Embedding and Pre-trained LanguagERepresentation (KEPLER), which can not only better integrate factual knowledge into PLMs but also produce effective text-enhanced KE with the strong PLMs is proposed.
- Delta Tuning: A Comprehensive Study of Parameter Efficient Methods for Pre-trained Language Models250
The theoretical principles underlying the effectiveness of delta tuning are discussed and frameworks to interpret delta tuning from the perspective of optimization and optimal control are proposed, where results on over 100 NLP tasks demonstrate a comprehensive performance comparison of different approaches.
- MAVEN: A Massive General Domain Event Detection Dataset232
A MAssive eVENt detection dataset is presented, which contains 4,480 Wikipedia documents, 117,200 event mention instances, and 207 event types and shows that existing ED methods cannot achieve promising results as on the small datasets, which suggests ED in real world remains a challenging task and requires further research efforts.
- Adversarial Training for Weakly Supervised Event Detection157
The experiments show that the candidate selection and adversarial training can cooperate together to obtain more diverse and accurate training data for ED, and significantly outperform the state-of-the-art methods in various weakly supervised scenarios.
- CPM: A large-scale generative Chinese Pre-trained language model135
CPM, with 2.6 billion parameters and 100GB Chinese training data, is the largest Chinese pre-trained language model, which could facilitate several downstream Chinese NLP tasks, such as conversation, essay generation, cloze test, and language understanding.
- CLEVE: Contrastive Pre-training for Event Extraction133
CLEVE is proposed, a contrastive pre-training framework for EE to better learn event knowledge from large unsupervised data and their semantic structures (e.g. AMR) obtained with automatic parsers.
- On Transferability of Prompt Tuning for Natural Language Processing122
It is found that the overlapping rate of activated neurons strongly reflects the transferability, which suggests how the prompts stimulate PLMs is essential, and that prompt transfer is promising for improving PT.
- MAVEN-ERE: A Unified Large-scale Dataset for Event Coreference, Temporal, Causal, and Subevent Relation Extraction116
A unified large-scale human-annotated ERE dataset MAVEN-ERE, which contains 103,193 event coreference chains, 1,216,217 temporal relations, 57,992 causal relations, and 15,841 subevent relations, larger than existing datasets of all the ERE tasks by at least an order of magnitude.
- HMEAE: Hierarchical Modular Event Argument Extraction102
A Hierarchical Modular Event Argument Extraction model is proposed, to provide effective inductive bias from the concept hierarchy of event argument roles and significantly outperform the state-of-the-art baselines.
- KoLA: Carefully Benchmarking World Knowledge of Large Language Models96
A Knowledge-oriented LLM Assessment benchmark (KoLA), in which a contrastive system is adopted for better numerical comparability across tasks and models and a unique self-contrast metric for automatically evaluating knowledge-creating ability is adopted.
- Train No Evil: Selective Masking for Task-Guided Pre-Training71
Experimental results on two sentiment analysis tasks show that the proposed selective masking task-guided pre-training method can achieve comparable or even better performance with less than 50\% overall computation cost, which indicates the method is both effective and efficient.
- When does In-context Learning Fall Short and Why? A Study on Specification-Heavy Tasks57
It is found that ICL falls short of handling specification-heavy tasks, which are tasks with complicated and extensive task specifications, requiring several hours for ordinary humans to master, such as traditional information extraction tasks.
- AGENTIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios49
This paper introduces AgentIF, the first benchmark for systematically evaluating LLM instruction following ability in agentic scenarios, and uses AgentIF to systematically evaluate existing advanced LLMs.
- GOAL: A Challenging Knowledge-grounded Video Captioning Benchmark for Real-time Soccer Commentary Generation46
Based on soccer game videos and synchronized commentary data, GOAL, a benchmark of over 8.9k soccer video clips, 22k sentences, and 42k knowledge triples is presented for proposing a challenging new task setting as Knowledge-grounded Video Captioning (KGVC).
- Adversarial Multi-lingual Neural Relation Extraction41
This work proposes an adversarial multi-lingual neural relation extraction (AMNRE) model, which builds both consistent and individual representations for each sentence to consider the consistency and diversity among languages.
- Toward Generalizable Evaluation in the LLM Era: A Survey Beyond Benchmarks39
This survey probes the core challenges that the rise of LLMs poses for evaluation, and identifies and analyzes two pivotal transitions: from task-specific to capability-based evaluation and from manual to automated evaluation.
- 27
- StoryWriter: A Multi-Agent Framework for Long Story Generation26
This paper presents StoryWriter, a modular and open-source multi-agent framework for controllable and scalable long story generation, and significantly outperforms existing story generation baselines in both story quality and length.
- 24
- 19
- 17
- 15
- 15
- Exploring Universal Intrinsic Task Subspace for Few-Shot Learning via Prompt Tuning11
Evidence is empirically found indicating that the adaptations of PLMs to various few-shot tasks can be reparameterized as optimizing only a few free parameters in a unified low-dimensional intrinsic task subspace, which may help to understand why PLMs could easily adapt to various NLP tasks with small-scale data.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-13. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.