Mihir Prabhudesai

Projects

• Latent Thought Flows with Text Compression

  Mihir Prabhudesai*, Zhengyang Geng*

  Blog Post

  webpage
• Solving Physics Olympiad via Reinforcement Learning on Physics Simulators

  Mihir Prabhudesai*, Aryan Satpathy*, Yangmin Li, Zheyang Qin, Nikash Bhardwaj,
  Amir Zadeh, Chuan Li, Katerina Fragkiadaki, Deepak Pathak

  ICML 2026

  webpage | arxiv | code
• 

  Can LLMs Lie? Investigation beyond Hallucination

  Haoran Huan, Mihir Prabhudesai, Mengning Wu, Shantanu Jaiswal, Deepak Pathak

  arXiv

  webpage | arxiv | code
• 

  Iterative Refinement Improves Compositional Image Generation

  Shantanu Jaiswal, Mihir Prabhudesai, Nikash Bhardwaj, Zheyang Qin, Amir Zadeh, Chuan Li, Katerina Fragkiadaki, Deepak Pathak

  NeurIPS 2026 Evaluations and Datasets Track

  webpage | arxiv | code
• 

  Diffusion Beats Autoregressive in Data-Constrained Settings

  Mihir Prabhudesai*, Mengning Wu*, Amir Zadeh, Katerina Fragkiadaki, Deepak Pathak

  NeurIPS 2025

  webpage | arxiv | code
• 

  Self-Questioning Language Models

  Lili Chen, Mihir Prabhudesai, Katerina Fragkiadaki, Hao Liu, Deepak Pathak

  MALGAI Workshop, ICLR 2026

  webpage | arxiv | code
• 

  Maximizing Confidence Alone Improves Reasoning

  Mihir Prabhudesai*, Lili Chen*, Alex Ippoliti*, Katerina Fragkiadaki, Hao Liu, Deepak Pathak

  TMLR 2026

  webpage | arxiv | code
• Unified Multimodal Discrete Diffusion

  Alexander Swerdlow*, Mihir Prabhudesai*, Siddharth Gandhi, Deepak Pathak, Katerina Fragkiadaki

  arXiv

  webpage | arxiv | code
• Video Diffusion Alignment via Reward Gradient

  Mihir Prabhudesai*, Russell Mendonca*, Zheyang Qin*, Katerina Fragkiadaki, Deepak Pathak

  arXiv

  webpage | arxiv | code
• Aligning Text-to-Image Diffusion Models with Reward Backpropagation

  Mihir Prabhudesai, Anirudh Goyal, Deepak Pathak, Katerina Fragkiadaki

  arXiv

  webpage | arxiv | code
• 

  Diffusion-TTA: Test-time Adaptation of Discriminative Models via Generative Feedback

  Mihir Prabhudesai*, Tsung-Wei Ke*, Alexander Cong Li, Deepak Pathak, Katerina Fragkiadaki

  NeurIPS 2023

  webpage | arxiv | code
• 

  Your diffusion model is secretly a zero-shot classifier

  Alexander C. Li, Mihir Prabhudesai, Shivam Duggal, Ellis Brown, Deepak Pathak

  ICCV 2023

  webpage | arxiv | code
• 

  Test-time Adaptation with Slot-Centric Models

  Mihir Prabhudesai, Anirudh Goyal, Sujoy Paul, Sjoerd Steenkiste, Mehdi Sajjadi, Gaurav Aggarwal, Thomas Kipf, Deepak Pathak, Katerina Fragkiadaki

  ICML 2023

  webpage | arxiv | code
• 

  CoCoNets: Continuous contrastive 3D scene representations

  Shamit Lal*, Mihir Prabhudesai*, Ishita Mediratta, Adam W Harley, Katerina Fragkiadaki

  CVPR 2021

  webpage | arxiv | code
• 

  Disentangling 3d prototypical networks for few-shot concept learning

  Mihir Prabhudesai, Shamit Lal*, Darshan Patil*, Hsiao-Yu Tung, Adam W Harley, Katerina Fragkiadaki

  ICLR 2021

  webpage | arxiv | code
• 

  3d-oes: Viewpoint-invariant object-factorized environment simulators

  Hsiao-Yu Fish Tung*, Zhou Xian*, Mihir Prabhudesai, Shamit Lal, Katerina Fragkiadaki

  CoRL 2020

  webpage | arxiv
• 

  Embodied language grounding with implicit 3d visual feature representations

  Mihir Prabhudesai, Hsiao-Yu Fish Tung, Syed Ashar Javed, Maximilian Sieb, Adam W Harley, Katerina Fragkiadaki

  CVPR 2020

  webpage | arxiv