activity
20242026
collaborators

7 papers

cs.CV2026

Evaluating the Diagnostic Robustness of Vision-Language Models Under Visual and Textual Perturbations

Ali Khoramfar, Mohammad Javad Dousti, Alireza Mohamadian +1

Standard accuracy metrics for VLMs often mask significant reliability failures in sensitive domains. In this work, we utilize a histopathology-validated brain MRI dataset to system…

cs.LG2026

MDP-GRPO: Stabilized Group Relative Policy Optimization for Multi-Constraint Instruction Following

Mohammad Mahdi Salmani-Zarchi, Zahra Rahimi, Heshaam Faili +1

Reinforcement learning with verifiable rewards is ideal for multi-constraint instruction following, yet standard group-relative policy optimization (GRPO) becomes unstable under di…

cs.CL2026

DeepQuestion: Systematic Generation of Real-World Challenges for Evaluating LLMs Performance

Ali Khoramfar, Ali Ramezani, Mohammad Mahdi Mohajeri +3

While Large Language Models (LLMs) achieve near-human performance on standard benchmarks, their capabilities often fail to generalize to complex, real-world problems. To bridge thi…

cs.CL2025

Dynamic Jointly Batch Selection for Data Efficient Machine Translation Fine-Tuning

Mohammad Amin Ghanizadeh, Mohammad Javad Dousti

Data quality and its effective selection are fundamental to improving the performance of machine translation models, serving as cornerstones for achieving robust and reliable trans…

cs.CL2025

: Data-Driven LoRA Initialization for Low Resource Tasks

Javad SeraJ, Mohammad Mahdi Mohajeri, Mohammad Javad Dousti

Tuning large language models is essential for optimizing their performance across diverse applications, particularly in scenarios with limited data availability. Tuning large langu…

cs.CL2025

Towards Data-Efficient Language Models: A Child-Inspired Approach to Language Learning

Mohammad Amin Ghanizadeh, Mohammad Javad Dousti

In this work, we explain our approach employed in the BabyLM Challenge, which uses various methods of training language models (LMs) with significantly less data compared to tradit…