Papers by Jiawen Zhang

12 papers
Towards Efficient NLP: A Standard Evaluation and A Strong Baseline (2022.naacl-main)

Copied to clipboard

Challenge: Rather than pursuing the reachless SOTA accuracy, researchers are focusing on model efficiency and usability.
Approach: They propose an evaluation and a public leaderboard for efficient NLP models that depicts the Pareto Frontier for various language understanding tasks.
Outcome: The proposed model outperforms or performs on par with SOTA compressed and early exiting models.
InstructSafety: A Unified Framework for Building Multidimensional and Explainable Safety Detector through Instruction Tuning (2023.findings-emnlp)

Copied to clipboard

Challenge: Existing safety detection systems have limitations in terms of their versatility and interpretability.
Approach: They introduce a safety detection framework that unifies 7 common sub-tasks into a uniform formulation and process 39 human-annotated datasets for instruction tuning.
Outcome: The proposed framework unifies 7 common sub-tasks into a uniform formulation and then runs on 39 human-annotated datasets to fine-tune it.
Beyond Explicit Refusals: Soft-Failure Attacks on Retrieval-Augmented Generation (2026.acl-long)

Copied to clipboard

Challenge: Existing jamming attacks on RAG systems typically induce explicit refusals or denial-of-service behaviors.
Approach: They propose a black-box attack framework that exploits safety-aligned behaviors of large language models to trigger soft failures.
Outcome: The proposed framework exploits safety-aligned behaviors of large language models to induce soft failures.
Token-level Inference-Time Alignment for Vision-Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Vision-Language Models (VLMs) often prioritize linguistic fluency over visual fidelity . despite widespread adoption, VLMs often exhibit a critical failure mode: hallucination .
Approach: They propose a framework for Token-level Inference-Time Alignment that steers the decoding process without updating the base model parameters.
Outcome: The proposed framework improves performance on 13 benchmarks across architectures . it boosts LLaVA-1.5-7B by 8.6% on MMVet and achieves a 74.0 MMStar score .
Constructing Highly Inductive Contexts for Dialogue Safety through Controllable Reverse Generation (2022.findings-emnlp)

Copied to clipboard

Challenge: Existing methods to detect toxic generation of pretrained language models rely on templates, data extraction, crowdsourcing workers or automatic generation.
Approach: They propose a method to construct adversarial contexts conditioned on a given response . they augment existing dataset BAD+ and construct a new dataset B AD+ .
Outcome: The proposed method can detect toxic or biased content in large pretrained language models.
Alleviating Exposure Bias via Multi-level Contrastive Learning and Deviation Simulation in Abstractive Summarization (2023.findings-acl)

Copied to clipboard

Challenge: Abstractive summarization systems have a severe mismatch between training and inference, i.e., exposure bias.
Approach: They propose a multi-level contrastive learning framework for abstractive summarization and a tailored sparse decoder self-attention pattern to bridge the gap between training and inference.
Outcome: The proposed framework outperforms the state-of-the-art models on two summarization datasets while adding relatively low overhead.
Multimodal Pragmatic Jailbreak on Text-to-image Models (2025.acl-long)

Copied to clipboard

Challenge: Existing jailbreaks for diffusion-based text-to-image models generate unsafe content . experimental results show that all tested models suffer from unsafe generation .
Approach: They propose a jailbreak that triggers diffusion-based text-to-image models to generate the image with visual text, resulting in unsafe content.
Outcome: The proposed model generates image with visual text, but the model is unsafe under such jailbreak.
TriPlay-RL: Tri-Role Self-Play Reinforcement Learning for LLM Safety Alignment (2026.acl-long)

Copied to clipboard

Challenge: Existing approaches to safety alignment of large language models rely on costly manual annotations or human review.
Approach: They propose a closed-loop reinforcement learning framework called TriPlay-RL that enables iterative collaboration among three roles with near-zero manual annotation.
Outcome: The proposed framework achieves 20%–50% improvement in adversarial effectiveness while preserving high output diversity while achieving 10%–30% gains in safety performance without degrading general reasoning capability.
Coarse-to-Fine: Hierarchical Multi-task Learning for Natural Language Understanding (2022.coling-1)

Copied to clipboard

Challenge: Existing methods to learn downstream tasks by stitches skill block lack rationality and interpretation.
Approach: They propose a hierarchical framework with a coarse-to-fine paradigm for generalized text representations from the large-scale corpus.
Outcome: The proposed model learns basic language properties from all tasks and boosts performance on relevant tasks.
SplitThenMerge: Token-Level Skill-Compositional Sparse Mixture-of-Experts for Complex Domain-Specific Tasks (2026.findings-acl)

Copied to clipboard

Challenge: Existing domain adaptation methods train heterogeneous skills together, making it difficult to reliably coordinate multiple skills when solving complex tasks.
Approach: They propose a framework that decomposes domain competence into atomic skills and composes them dynamically during generation.
Outcome: The proposed framework decomposes domain competence into atomic skills, trains them independently, and composes them dynamically during generation.
EthicMind: A Risk-Aware Framework for Ethical-Emotional Alignment in Multi-Turn Dialogue (2026.acl-long)

Copied to clipboard

Challenge: Existing dialogue models address empathy and ethical safety in isolation . Existing models fail to adapt their behavior as ethical risk and user emotion evolve .
Approach: They propose a risk-aware framework that integrates ethical-emotional alignment in dialogue as an explicit turn-level decision problem.
Outcome: The proposed framework achieves more consistent ethical guidance and emotional engagement than baselines in ethically complex interactions.
GECSum: Generative Evaluation-Driven Sequence Level Contrastive Learning for Abstractive Summarization (2024.lrec-main)

Copied to clipboard

Challenge: Abstractive summarization is a technique in natural language processing that involves generating a summary of a source document by creating new sentences and phrases.
Approach: They propose a sequence-level contrastive learning framework that leverages the semantic understanding capabilities of the abstractive model itself to evaluate summary in reference-based settings.
Outcome: The proposed framework outperforms the state-of-the-art in four summarization datasets.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations