Jinzhe Liu

Jinzhe Liu, Xiangsheng Huang, Zhuo Chen et al.

Large Language Models (LLMs) encounter challenges with the unique syntax of specific domains, such as biomolecules. Existing fine-tuning or modality alignment techniques struggle to bridge the domain knowledge gap and understand complex molecular data, limiting LLMs' progress in specialized fields. To overcome these limitations, we propose an expandable and adaptable non-parametric knowledge injection framework named Domain-specific Retrieval-Augmented Knowledge (DRAK), aimed at enhancing reasoning capabilities in specific domains. Utilizing knowledge-aware prompts and gold label-induced reasoning, DRAK has developed profound expertise in the molecular domain and the capability to handle a broad spectrum of analysis tasks. We evaluated two distinct forms of DRAK variants, proving that DRAK exceeds previous benchmarks on six molecular tasks within the Mol-Instructions dataset. Extensive experiments have underscored DRAK's formidable performance and its potential to unlock molecular insights, offering a unified paradigm for LLMs to tackle knowledge-intensive tasks in specific domains. Our code will be available soon.

TianTian Dang, Chao Bi, Shufan Shen et al.

Despite the significant advancements in Large Vision-Language Models (LVLMs), their tendency to generate hallucinations undermines reliability and restricts broader practical deployment. Among the hallucination mitigation methods, feature steering emerges as a promising approach that reduces erroneous outputs in LVLMs without increasing inference costs. However, current methods apply uniform feature steering across all layers. This heuristic strategy ignores inter-layer differences, potentially disrupting layers unrelated to hallucinations and ultimately leading to performance degradation on general tasks. In this paper, we propose a plug-and-play framework called Locate-Then-Sparsify for Feature Steering (LTS-FS), which controls the steering intensity according to the hallucination relevance of each layer. We first construct a synthetic dataset comprising token-level and sentence-level hallucination cases. Based on this dataset, we introduce an attribution method based on causal interventions to quantify the hallucination relevance of each layer. With the attribution scores across layers, we propose a layerwise strategy that converts these scores into feature steering intensities for individual layers, enabling more precise adjustments specifically on hallucination-relevant layers. Extensive experiments across multiple LVLMs and benchmarks demonstrate that our LTS-FS framework effectively mitigates hallucination while preserving strong performance.

Jinzhe Liu, Junshu Sun, Shufan Shen et al.

Lifelong knowledge editing enables continuous, precise updates to outdated knowledge in large language models (LLMs) without computationally expensive full retraining. However, existing methods often accumulate errors throughout the editing process, causing a gradual decline in both editing accuracy and generalization. To tackle this problem, we propose Neuron-Specific Masked Knowledge Editing (NMKE), a novel fine-grained editing framework that combines neuron-level attribution with dynamic sparse masking. Leveraging neuron functional attribution, we identify two key types of knowledge neurons, with knowledge-general neurons activating consistently across prompts and knowledge-specific neurons activating to specific prompts. NMKE further introduces an entropy-guided dynamic sparse mask, locating relevant neurons to the target knowledge. This strategy enables precise neuron-level knowledge editing with fewer parameter modifications. Experimental results from thousands of sequential edits demonstrate that NMKE outperforms existing methods in maintaining high editing success rates and preserving model general capabilities in lifelong editing.