publications

publications by categories in reversed chronological order.

2025

  1. ARRNeurIPS LLM Eval
    RefusalBench: Generative Evaluation of Selective Refusal in Grounded Language Models
    Aashiq Muhamed, Leonardo F. R. Ribeiro, Markus Dreyer, and 2 more authors
    In ACL Rolling Review (in submission), 2025
  2. NeurIPS MechInterp Spotlight
    Position: Mechanistic Interpretability Should Prioritize Feature Consistency in SAEs
    Xiangchen Song*, Aashiq Muhamed*, Yujia Zheng, and 5 more authors
    In Mechanistic Interpretability Workshop at NeurIPS, 2025
  3. NeurIPS MechInterp Spotlight
    Towards a Mechanistic Understanding of Robustness in Finetuned Reasoning Models
    Aashiq Muhamed, Xuandong Zhao, Mona T. Diab, and 2 more authors
    In Mechanistic Interpretability Workshop at NeurIPS, 2025
  4. COLMICML Actionable InterpICML R2FM
    SAEs can improve unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMs
    Aashiq Muhamed, Jacopo Bonato, Mona Diab, and 1 more author
    In Conference on Language Modeling (COLM), 2025
  5. NAACL
    CoRAG: Collaborative Retrieval-Augmented Generation
    Aashiq Muhamed, Mona T. Diab, and Virginia Smith
    In 2025 Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL), 2025
  6. NAACL FindingsNeurIPS ATTRIBNAACL TrustNLP Oral
    Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
    Aashiq Muhamed, Mona Diab, and Virginia Smith
    In Findings of the Association for Computational Linguistics: NAACL 2025, 2025
  7. TMLR
    NeurIPS 2023 Competition: Privacy Preserving Federated Learning Document VQA
    Marlon Tobaben, Mohamed Ali Souibgui, Rubèn Tito, and 24 more authors
    In Transactions on Machine Learning Research, 2025
  8. ICML Actionable InterpICML R2FM
    The Geometry of Forgetting: Analyzing Machine Unlearning through Local Learning Coefficients
    Aashiq Muhamed, and Virginia Smith
    In Actionable Interpretability Workshop at the International Conference on Machine Learning (ICML), 2025
  9. SIGIR LLM4Eval
    CCRS: A Zero-Shot LLM-as-a-Judge Framework for Comprehensive RAG Evaluation
    Aashiq Muhamed
    In SIGIR 2025: The Third Workshop on Large Language Models (LLMs) for Evaluation in Information Retrieval, 2025

2024

  1. EMNLPICML ES-FoMo
    Grass: Compute Efficient Low-Memory LLM Training with Structured Sparse Gradients
    Aashiq Muhamed, Oscar Li, David Woodruff, and 2 more authors
    In The 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024
  2. CVPR
    Adversarial Text to Continuous Image Generation
    Kilichbek Haydarov, Aashiq Muhamed, Jovana Lazarevic, and 4 more authors
    In The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024, 2024
  3. ICLR Private ML
    Fed Up with Complexity: Simplifying Many-Task Federated Learning with NTKFedAvg
    Aashiq Muhamed, Meher Mankikar, and Virginia Smith
    In Privacy Regulation and Protection in Machine Learning, Workshop at ICLR, 2024
  4. ICLR Private ML
    Cache Me If You Can: The Case For Retrieval Augmentation in Federated Learning
    Aashiq Muhamed, Pratiksha Thaker, Mona T. Diab, and 1 more author
    In Privacy Regulation and Protection in Machine Learning, Workshop at ICLR, 2024
  5. NeurIPS ENLSP
    Inducing Elasticity in Foundation Models: Post-Training Techniques for Adaptable Inference
    Aashiq Muhamed, Jiarui Liu, Mona Diab, and 1 more author
    In The 4th Workshop on Efficient Natural Language and Speech Processing, 2024
  6. EMNLP CustomNLP4UEACL MOOMIN Oral
    Less is Fed More: Sparsity Reduces Feature Distortion in Federated Learning
    Aashiq Muhamed*, Harshita Diddee*, and Abhinav Rao*
    In Workshop on Customizable NLP (CustomNLP4U) at EMNLP 2024, 2024

2023

  1. PAKDD
    Web-scale semantic product search with large language models
    Aashiq Muhamed*, Sriram Srinivasan*, Choon-Hui Teo, and 4 more authors
    In Pacific-Asia Conference on Knowledge Discovery and Data Mining, 2023
  2. ACL Oral
    ReAugKD: Retrieval-Augmented Knowledge Distillation for Pre-trained Language Models
    Jianyi Zhang, Aashiq Muhamed, Aditya Anantharaman, and 7 more authors
    In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics ACL, 2023
  3. ICML ES-FoMo
    Reverse Distillation: Training Billion Parameter Models For CTR Prediction
    Aditya Anantharam*, Aashiq Muhamed*, Hemant Pugaliya, and 6 more authors
    In ICML Workshop on Efficient Systems for Foundation Models (ES-FoMo), 2023
  4. Preprint
    An In-depth Look at Gemini’s Language Abilities
    Syeda Nahida Akter*, Zichun Yu*, Aashiq Muhamed*, and 6 more authors
    arXiv preprint arXiv: 2312.11444, 2023
  5. KDD
    Tutorial on Training Large-scale Foundation Models on Emerging AI Chips
    Aashiq Muhamed, Christian Bock, Rahul Solanki, and 3 more authors
    In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2023

2022

  1. WWW
    DCAF-BERT: A Distilled Cachable Adaptable Factorized Model For Improved Ads CTR Prediction
    Aashiq Muhamed, Jaspreet Singh, Shuai Zheng, and 8 more authors
    In WWW ’22: Proceedings of the Web Conference, 2022
  2. Preprint
    HyperCGAN: Text-to-Image Synthesis with HyperNet-Modulated Conditional Generative Adversarial Networks
    Kilichbek Haydarov, Aashiq Muhamed, Jovana Lazarevic, and 2 more authors
    https://openreview.net/forum?id=z-5BjnU3-OQ, 2022

2021

  1. AAAI OralNeurIPS ML4CDISMIR NLP4MusA Oral
    Symbolic Music Generation with Transformer-GANs
    Aashiq Muhamed, Liang Li, Xingjian Shi, and 6 more authors
    In Proceedings of the AAAI Conference on Artificial Intelligence, 2021
  2. EMBC
    Gated Transformer for Decoding Human Brain EEG Signals
    Yunzhe Tao, Tao Sun, Aashiq Muhamed, and 6 more authors
    43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2021
  3. NeurIPS ENLSP Oral
    CTR-BERT: Cost-effective knowledge distillation for billion-parameter teacher models
    Aashiq Muhamed, Iman Keivanloo, Sujan Perera, and 6 more authors
    In NeurIPS Workshop on Efficient Natural Language and Speech Processing (ENLSP), 2021

2020

  1. NeurIPS ML4CD
    Transformer-GAN: Symbolic music generation using a learned loss
    Aashiq Muhamed, Liang Li, Xingjian Shi, and 6 more authors
    In 4th NeurIPS Workshop on Machine Learning for Creativity and Design, 2020
  2. Patent
    3D convolutional neural networks for television advertisement detection. US Patent 10,706,286
    Aashiq Muhamed, Susmita Ghose, and Dawnis Chow
    Jul 2020
    US Patent 10,706,286
  3. Patent
  4. Amazon CVC
    Designing event representations for symbolic music
    Aashiq Muhamed, Liang Li, and Xingjian Shi
    In Amazon Computer Vision Conference (CVC), Jun 2020