111 citations · 282 across the 52 of their papers we have counts for
32 papers · 1 filter
Beyond Shallow Alignment: How Post-Training Methods Determine Refusal Circuits And Steering Robustness
Hoang Cuong Nguyen, Mark Dras, Usman Naseem
How do the methods used to train language models to refuse harmful requests shape how that refusal actually works inside the model? We compare three post-training methods - supervi…
AlignCultura: Towards Culturally Aligned Large Language Models?
Gautam Siddharth Kashyap, Mark Dras, Usman Naseem
Cultural alignment in Large Language Models (LLMs) is essential for producing contextually aware, respectful, and trustworthy outputs. Without it, models risk generating stereotype…
Over-Refusal and Representation Subspaces: A Mechanistic Analysis of Task-Conditioned Refusal in Aligned LLMs
Utsav Maskey, Mark Dras, Usman Naseem
Aligned language models that are trained to refuse harmful requests also exhibit over-refusal: they decline safe instructions that seemingly resemble harmful instructions. A natura…
When the Model Said 'No Comment', We Knew Helpfulness Was Dead, Honesty Was Alive, and Safety Was Terrified
Gautam Siddharth Kashyap, Mark Dras, Usman Naseem
Large Language Models (LLMs) need to be in accordance with human values-being helpful, harmless, and honest (HHH)-is important for safe deployment. Existing works use Supervised Fi…
CogMem: A Cognitive Memory Architecture for Sustained Multi-Turn Reasoning in Large Language Models
Yiran Zhang, Jincheng Hu, Mark Dras +1
Large language models (LLMs) excel at single-turn reasoning but often lose accuracy and coherence over extended, multi-turn interactions. Recent evaluations such as TurnBench highl…
Beyond the Black Box: Demystifying Multi-Turn LLM Reasoning with VISTA
Yiran Zhang, Mingyang Lin, Mark Dras +1
Recent research has increasingly focused on the reasoning capabilities of Large Language Models (LLMs) in multi-turn interactions, as these scenarios more closely mirror real-world…
SHIELD: Classifier-Guided Prompting for Robust and Safer LVLMs
Juan Ren, Mark Dras, Usman Naseem
Large Vision-Language Models (LVLMs) unlock powerful multimodal reasoning but also expand the attack surface, particularly through adversarial inputs that conceal harmful goals in…
Steering Over-refusals Towards Safety in Retrieval Augmented Generation
Utsav Maskey, Mark Dras, Usman Naseem
Safety alignment in large language models (LLMs) induces over-refusals -- where LLMs decline benign requests due to aggressive safety filters. We analyze this phenomenon in retriev…
Fairness Evaluation and Inference Level Mitigation in LLMs
Afrozah Nadeem, Mark Dras, Usman Naseem
Large language models often display undesirable behaviors embedded in their internal representations, undermining fairness, inconsistency drift, amplification of harmful content, a…
Steering Towards Fairness: Mitigating Political Bias in LLMs
Afrozah Nadeem, Mark Dras, Usman Naseem
Recent advancements in large language models (LLMs) have enabled their widespread use across diverse real-world applications. However, concerns remain about their tendency to encod…
Too Helpful, Too Harmless, Too Honest or Just Right?
Gautam Siddharth Kashyap, Mark Dras, Usman Naseem
Large Language Models (LLMs) exhibit strong performance across a wide range of NLP tasks, yet aligning their outputs with the principles of Helpfulness, Harmlessness, and Honesty (…
We Think, Therefore We Align LLMs to Helpful, Harmless and Honest Before They Go Wrong
Gautam Siddharth Kashyap, Mark Dras, Usman Naseem
Alignment of Large Language Models (LLMs) is the ability to satisfy desired objectives during generation, which is critical for trustworthy deployment. In practice, alignment is of…
SafeConstellations: Mitigating Over-Refusals in LLMs Through Task-Aware Representation Steering
Utsav Maskey, Sumit Yadav, Mark Dras +1
LLMs increasingly exhibit over-refusal behavior, where safety mechanisms cause models to reject benign instructions that seemingly resemble harmful content. This phenomenon diminis…
Should LLM Safety Be More Than Refusing Harmful Instructions?
Utsav Maskey, Mark Dras, Usman Naseem
This paper presents a systematic evaluation of Large Language Models' (LLMs) behavior on long-tail distributed (encrypted) texts and their safety implications. We introduce a two-d…
Framing Political Bias in Multilingual LLMs Across Pakistani Languages
Afrozah Nadeem, Mark Dras, Usman Naseem
Large Language Models (LLMs) increasingly shape public discourse, yet most evaluations of political and economic bias have focused on high-resource, Western languages and contexts.…
Seeing the Threat: Vulnerabilities in Vision-Language Models to Adversarial Attack
Juan Ren, Mark Dras, Usman Naseem
Large Vision-Language Models (LVLMs) have shown remarkable capabilities across a wide range of multimodal tasks. However, their integration of visual inputs introduces expanded att…
A Survey on Progress in LLM Alignment from the Perspective of Reward Design
Miaomiao Ji, Yanqiu Wu, Zhibin Wu +4
Reward design plays a pivotal role in aligning large language models (LLMs) with human values, serving as the bridge between feedback signals and model optimization. This survey pr…
Myanmar XNLI: Building a Dataset and Exploring Low-resource Approaches to Natural Language Inference with Myanmar
Aung Kyaw Htet, Mark Dras
Despite dramatic recent progress in NLP, it is still a major challenge to apply Large Language Models (LLM) to low-resource languages. This is made visible in benchmarks such as Cr…
VaxGuard: A Multi-Generator, Multi-Type, and Multi-Role Dataset for Detecting LLM-Generated Vaccine Misinformation
Syed Talal Ahmad, Haohui Lu, Sidong Liu +4
Recent advancements in Large Language Models (LLMs) have significantly improved text generation capabilities. However, they also present challenges, particularly in generating vacc…
VITAL: A New Dataset for Benchmarking Pluralistic Alignment in Healthcare
Anudeex Shetty, Amin Beheshti, Mark Dras +1
Alignment techniques have become central to ensuring that Large Language Models (LLMs) generate outputs consistent with human values. However, existing alignment paradigms often mo…
IDT: Dual-Task Adversarial Attacks for Privacy Protection
Pedro Faustini, Shakila Mahjabin Tonni, Annabelle McIver +2
Natural language processing (NLP) models may leak private information in different ways, including membership inference, reconstruction or attribute inference attacks. Sensitive in…
Here's a Free Lunch: Sanitizing Backdoored Models with Model Merge
Ansh Arora, Xuanli He, Maximilian Mozes +3
The democratization of pre-trained language models through open-source initiatives has rapidly advanced innovation and expanded access to cutting-edge technologies. However, this o…
Detecting Textual Adversarial Examples Based on Distributional Characteristics of Data Representations
Na Liu, Mark Dras, Wei Emma Zhang
Although deep neural networks have achieved state-of-the-art performance in various machine learning tasks, adversarial examples, constructed by adding small non-random perturbatio…
Neural Rule-Execution Tracking Machine For Transformer-Based Text Generation
Yufei Wang, Can Xu, Huang Hu +5
Sequence-to-Sequence (S2S) neural text generation models, especially the pre-trained ones (e.g., BART and T5), have exhibited compelling performance on various natural language gen…