Publications by Megh Thakkar

Alumni member

Activity

Towards Efficient and Effective Preference Alignment for Large Language Models
by Megh Thakkar, with Sarath Chandar as supervisor.
Université de Montréal ⸺ 2024.
[thesis]

Too Big to Fool: Resisting Deception in Language Models
Mohammad Reza Samsami, Mats Leon Richter, Juan Rodriguez, Megh Thakkar, Sarath Chandar, and Maxime Gasse
In ArXiv, 2024.
#NLP
[arXiv]

How to Train Your LLM Web Agent: A Statistical Diagnosis
Dheeraj Vattikonda, Santhoshi Ravichandran, Emiliano Penaloza, Hadi Nekoei, Megh Thakkar, Thibault Le Sellier de Chezelles, Nicolas Gontier, Miguel Muñoz-Mármol, Sahar Omidi Shayegan, Stefania Raimondo, Xue Liu, Alexandre Drouin, Laurent Charlin, Alexandre Piché, Alexandre Lacoste, and Massimo Caccia
Conference on Neural Information Processing Systems (NeurIPS), 2025.
#NLP, #RL
[arXiv]
Combining Domain and Alignment Vectors Provides Better Knowledge-Safety Trade-offs in LLMs
Megh Thakkar, Quentin Fournier, Matthew Riemer, Pin-Yu Chen, Amal Zouaq, Payel Das, and Sarath Chandar
Annual Meeting of the Association for Computational Linguistics (ACL), 2025.
#NLP
[acl], [arXiv]
ChartGemma: Visual Instruction-tuning for Chart Reasoning in the Wild
Ahmed Masry*, Megh Thakkar*, Aayush Bajaj, Aaryaman Kartha, Enamul Hoque, and Shafiq Joty
International Conference on Computational Linguistics (COLING) Industry Track, 2025.
#NLP
[acl], [arXiv], [code]

WorkArena++: Towards Compositional Planning and Reasoning-based Common Knowledge Work Tasks
Leo Boisvert*, Megh Thakkar*, Maxime Gasse, Massimo Caccia, Thibault Le Sellier De Chezelles, Quentin Cappart, Nicolas Chapados, Alexandre Lacoste, and Alexandre Drouin
Conference on Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track, 2024.
#NLP
[neurips], [openreview], [arXiv], [code]
A deep-dive into the tradeoffs of preference alignment with PEFT
Megh Thakkar, Quentin Fournier, Matthew Riemer, Pin-Yu Chen, Amal Zouaq, Payel Das, and Sarath Chandar
Annual Meeting of the Association for Computational Linguistics (ACL), 2024.
#NLP
[acl], [arXiv]

Self-Influence Guided Data Reweighting for Language Model Pre-training
Megh Thakkar, Tolga Bolukbasi, Sriram Ganapathy, Shikhar Vashishth, Sarath Chandar, and Partha Talukdar
Conference on Empirical Methods in Natural Language Processing (EMNLP), 2023.
#NLP
[acl], [openreview], [arXiv]