pith. sign in

Mark Dras

Identifiers

  • name variant Mark Dras 0.60 · backfill

Papers (19)

  1. AlignCultura: Towards Culturally Aligned Large Language Models? cs.CL · 2026 · author #2
  2. Over-Refusal and Representation Subspaces: A Mechanistic Analysis of Task-Conditioned Refusal in Aligned LLMs cs.CL · 2026 · author #2
  3. Fairness Evaluation and Inference Level Mitigation in LLMs cs.CL · 2025 · author #2
  4. We Think, Therefore We Align LLMs to Helpful, Harmless and Honest Before They Go Wrong cs.CL · 2025 · author #2
  5. SafeConstellations: Mitigating Over-Refusals in LLMs Through Task-Aware Representation Steering cs.CL · 2025 · author #3
  6. Defending against Backdoor Attacks via Module Switching cs.CR · 2025 · author #4
  7. Generalised Differential Privacy for Text Document Processing cs.CR · 2018 · author #2
  8. Senti-Attend: Image Captioning using Sentiment and Attention cs.CV · 2018 · author #2
  9. Automatic Recognition of Student Engagement using Deep Learning and Facial Expression cs.CV · 2018 · author #2
  10. Face-Cap: Image Captioning using Facial Expression Analysis cs.CV · 2018 · author #2
  11. Author Obfuscation Using Generalised Differential Privacy cs.CR · 2018 · author #2
  12. VnCoreNLP: A Vietnamese Natural Language Processing Toolkit cs.CL · 2018 · author #4
  13. From Word Segmentation to POS Tagging for Vietnamese cs.CL · 2017 · author #4
  14. A Fast and Accurate Vietnamese Word Segmenter cs.CL · 2017 · author #4
  15. A Novel Neural Network Model for Joint POS Tagging and Graph-based Dependency Parsing cs.CL · 2017 · author #2
  16. Native Language Identification using Stacked Generalization cs.CL · 2017 · author #2
  17. An empirical study for Vietnamese dependency parsing cs.CL · 2016 · author #2
  18. Modeling Language Change in Historical Corpora: The Case of Portuguese cs.CL · 2016 · author #3
  19. Death and Lightness: Using a Demographic Model to Find Support Verbs cmp-lg · 1996 · author #1

Mentions

  • 2509.22510 #2 · arxiv_oai · confidence 0.70 Mark Dras

Frequent Coauthors