Papers by Jiachen Li
WeCheck: Strong Factual Consistency Checker via Weakly Supervised Learning (2023.acl-long)
Copied to clipboard
| Challenge: | Existing factual consistency metrics are often uncontrollably generating text that is factually inconsistent with inputs. |
| Approach: | They propose a weakly supervised framework that is directly trained on actual generated samples from language models with weakly annotated labels. |
| Outcome: | The proposed framework improves on the TRUE benchmark by 3.3% over existing methods with 435M parameters. |
UNIMO: Towards Unified-Modal Understanding and Generation via Cross-Modal Contrastive Learning (2021.acl-long)
Copied to clipboard
| Challenge: | Existing pre-training methods focus on single-modal tasks or multi-modal ones . large-scale pre- training has drawn much attention in both the community of Compute Vision (CV) and Natural Language Processing (NLP). |
| Approach: | They propose a UNIfied-MOdal pre-training architecture which can adapt to both single-modal and multi-modal understanding and generation tasks. |
| Outcome: | The proposed model can learn more generalizable representations with rich non-paired single-modal data. |
Context-aware Embedding for Targeted Aspect-based Sentiment Analysis (P19-1)
Copied to clipboard
| Challenge: | Existing methods do not specifically pre-train reasonable embeddings for targets and aspects in TABSA. |
| Approach: | They propose to refine the embeddings of targets and aspects using a sparse coefficient vector . this allows the embeds to be refined from highly correlative words instead of context-independent vectors . |
| Outcome: | Experiments show that the proposed method improves on two benchmark datasets. |
UNIMO-G: Unified Image Generation through Multimodal Conditional Diffusion (2024.acl-long)
Copied to clipboard
| Challenge: | Existing text-to-image diffusion models generate images from text prompts due to inherent brevity of textual descriptions . however, the ability to accurately synthesize images with intricate details, such as specific entities or scenes, is limited due to the inherent bribery of text descriptions. |
| Approach: | They propose a multimodal conditional diffusion framework that operates on multimodal prompts with interleaved textual and visual inputs. |
| Outcome: | The proposed framework excels in both text-to-image generation and zero-shot subject-driven synthesis. |
FRSUM: Towards Faithful Abstractive Summarization via Enhancing Factual Robustness (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing models of abstractive summarization are able to generate fluent and coherent summaries, but they still suffer from the unfaithful generation problem. |
| Approach: | They propose to improve the faithfulness of existing models by enhancing their factual robustness by using a novel training strategy, namely FRSUM, which teaches the model to defend against both explicit adversarial samples and implicit factual adversarials. |
| Outcome: | The proposed training strategy improves faithfulness of various models, such as T5, BART, and T5 . |
Adversarial Attacks on Parts of Speech: An Empirical Study in Text-to-Image Generation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Recent studies show that text-to-image models are vulnerable to adversarial perturbations . |
| Approach: | They investigate the impact of adversarial attacks on different POS tags within text prompts on T2I models. |
| Outcome: | The proposed model is vulnerable to adversarial perturbations with noun perturbations in text prompts. |
HEAL: An Empirical Study on Hallucinations in Embodied Agents Driven by Large Language Models (2025.findings-emnlp)
Copied to clipboard
Trishna Chakraborty, Udita Ghosh, Xiaopan Zhang, Fahim Faisal Niloy, Yue Dong, Jiachen Li, Amit Roy-Chowdhury, Chengyu Song
| Challenge: | Large language models are increasingly being adopted as the cognitive core of embodied agents. |
| Approach: | They propose a systematic study of hallucinations in large language models . they aim to understand to what extent hallucinos occur, what types trigger them . |
| Outcome: | The proposed model can induce hallucinations up to 40 higher than base prompts . the model fails to resolve scene-task inconsistencies, the study finds . |
CuMA: Aligning LLMs with Sparse Cultural Values via Demographic-Aware Mixture of Adapters (2026.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have a global audience, so alignment must extend to cultural resonance. |
| Approach: | They propose a framework that frames alignment as a conditional capacity separation problem. |
| Outcome: | The proposed framework outperforms both dense baselines and semantic-only MoEs on three large language models. |
Precisely the Point: Adversarial Augmentations for Faithful and Informative Text Generation (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing models for text generation are weak enough to handle perturbations in inputs, leading to degeneration in faithfulness and informativeness. |
| Approach: | They propose a framework for improving faithfulness and informativeness of Seq2Seq models by perturbing word representations and word swapping. |
| Outcome: | The proposed framework improves faithfulness and informativeness of Seq2Seq models under automatic and human evaluation settings. |
Leveraging Graph to Improve Abstractive Multi-Document Summarization (2020.acl-main)
Copied to clipboard
| Challenge: | Empirical results show that our model brings substantial improvements over several strong baselines. |
| Approach: | They propose a neural abstractive multi-document summarization model which captures cross-document relations and can guide the summary generation process. |
| Outcome: | The proposed model improves on the WikiSum and MultiNews datasets and can be easily combined with pre-trained language models. |
InstructEval: Instruction-Tuned Text Evaluator from Human Preference (2024.findings-acl)
Copied to clipboard
| Challenge: | InstructEval is a general text evaluator based on open-source Large Language Models (LLMs). |
| Approach: | They propose to build a general multi-aspect text evaluator based on open-source Large Language Models (LLMs) they use extensive open Human Preference Modeling datasets and a small set of multi-spect annotated data to overcome the shortage of annotation resources for multi-task evaluations. |
| Outcome: | The proposed model performs comparable or superior to commercial LLMs like ChatGPT or GPT-4 in terms of both overall and multi-aspect evaluation tasks. |
Empowering Backbone Models for Visual Text Generation with Input Granularity Control and Glyph-Aware Training (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing text-to-image models struggle to generate images with legible visual texts . current models lack support for Chinese texts, misspelling, and lack of diversity . |
| Approach: | They propose to empower backbone models to generate visual texts in Chinese and English . they propose to augment conventional training objective with glyph-aware training losses . |
| Outcome: | The proposed methods can generate visual texts in English and Chinese while maintaining image generation quality. |
SgSum:Transforming Multi-document Summarization into Sub-graph Selection (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing extractive multi-document summarization methods score each sentence individually and extract salient sentences one by one. |
| Approach: | They propose a novel framework for extractive multi-document summarization that selects a sub-graph as the summary instead of selecting salient sentences. |
| Outcome: | The proposed framework improves on existing methods on multi-document datasets and human evaluations show it produces more coherent and informative summaries. |
Jailbreaking Prompt Attack: A Controllable Adversarial Attack against Diffusion Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Text-to-image (T2I) models can be used to generate harmful content such as sexually explicit, unfaithful, and misleading or Not-Safe-for-Work (NSFW) images. |
| Approach: | They propose a more practical and universal attack that does not require the presence of a target model. |
| Outcome: | The proposed attack bypasses both text and image safety checkers while preserving high semantic alignment with the target prompt. |
VisualWebInstruct: Scaling up Multimodal Instruction Data through Web Search (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing vision-language models struggle with reasoning-focused tasks due to the lack of high-quality training data. |
| Approach: | They propose a new approach that leverages search engines to create a multimodal multimodal dataset . they use a set of 30,000 seed images to extract HTML data from 700K unique URLs . |
| Outcome: | The proposed model achieves the best known performance on MMMU-Pro (40.7), MathVerse (42.6), and DynaMath (55.7). |
More Samples or More Prompts? Exploring Effective Few-Shot In-Context Learning for LLMs with In-Context Sampling (2024.findings-naacl)
Copied to clipboard
Bingsheng Yao, Guiming Chen, Ruishi Zou, Yuxuan Lu, Jiachen Li, Shao Zhang, Yisi Sang, Sijia Liu, James Hendler, Dakuo Wang
| Challenge: | Existing studies on LLM prompting focus on selecting a better set of data samples inside one single prompt input, but why not design and leverage multiple ICL prompts together to further improve the LLM’s performance? |
| Approach: | They propose a low-resource LLM prompting technique to optimize the construction of multiple ICL prompt inputs to produce confident predictions. |
| Outcome: | The proposed technique can produce confident predictions by optimizing the construction of multiple ICL prompt inputs on four NLI datasets and one QA dataset. |
Language Model Adaption for Reinforcement Learning with Natural Language Action Space (2024.acl-long)
Copied to clipboard
| Challenge: | Previous research has focused on reducing the size of the natural language action space due to the combinatorial nature of the language. |
| Approach: | They propose mutual-information regularized policy optimization to reduce the action space by dynamically adjusting the prior provided by the pretrained model. |
| Outcome: | The proposed method improves monotonically on the mutual-information regularized RL objective. |
UNIMO-2: End-to-End Unified Vision-Language Grounded Learning (2022.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for vision-language pre-training can only learn from aligned image-caption data and rely heavily on expensive regional features. |
| Approach: | They propose an end-to-end unified-modal pre-training framework for joint learning . they propose to conduct grounded learning on both images and texts via a sharing grounded space . |
| Outcome: | The proposed model improves visual and visual semantic alignment on images and texts. |
BASS: Boosting Abstractive Summarization with Unified Semantic Graph (2021.acl-long)
Copied to clipboard
| Challenge: | Abstractive summarization for long-document or multi-document remains challenging for Seq2Seq as it does not analyze long-distance relations in text. |
| Approach: | They propose a framework for Boosting Abstractive Summarization based on a unified Semantic graph which aggregates co-referent phrases distributing across a long range of context and conveys rich relations between phrases. |
| Outcome: | The proposed framework improves document representation and summary generation process by leveraging the graph structure. |
Variational Autoregressive Decoder for Neural Response Generation (D18-1)
Copied to clipboard
| Challenge: | Existing variational Bayesian models generate responses from a single latent variable, which is not sufficient to model high variability in responses. |
| Approach: | They propose a conditional variable auto-encoder that sequentially introduces latent variables to condition the generation of each word in the response sequence. |
| Outcome: | Empirical results show that the proposed model improves on state-of-the-art models on Opensubtitle and Reddit datasets. |
CLLMate: A Multimodal Benchmark for Weather and Climate Events Forecasting (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing environmental forecasting research focuses on predicting numerical meteorological variables, neglecting the translation of these variables into actionable textual narratives of events and their consequences. |
| Approach: | They propose a task that leverages numerical meteorological raster data and textual event data to predict weather and climate events. |
| Outcome: | The proposed task leverages numerical meteorological raster data and textual event data to predict weather and climate events. |
Improve LLM-as-a-Judge Ability as a General Ability (2025.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies focus on generative judges, but only on their judge ability. |
| Approach: | They propose a method that leverages the generative and reasoning capabilities of large language models to evaluate LLM responses across diverse scenarios, providing accurate preference signals. |
| Outcome: | The proposed model performs on RewardBench with only 2% to 40% of the data required by other training frameworks. |
TC-Bench: Benchmarking Temporal Compositionality in Conditional Video Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing video generation models struggle to interpret compositional changes and synthesize components across different time steps. |
| Approach: | They propose a temporal compositionality benchmark that uses text prompts and ground truth videos to evaluate compositional changes in video. |
| Outcome: | The proposed benchmark can be used for text-to-video and image-to video generation. |
BPO: Staying Close to the Behavior LLM Creates Better Online LLM Alignment (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing offline DAP methods for aligning large language models with human preference are computationally expensive due to their two-stage training pipeline that consists of a reward modeling phase. |
| Approach: | They propose to align large language models to human desiderata from offline preference datasets by using an online approach. |
| Outcome: | The proposed approach improves performance across a wide range of tasks when training with the same amount of preference data. |