News

Our paper about LLM-based web agents has been accepted at ICLR 2025.
Our paper about a new and an effective way of training of LLM has been accepted at TMLR 2025.
Curious how to improve your RLHF by building stronger SFT with just model's own generated tokens? Then check out our recent work!
Want to build simple yet effective LLM-based web agents? Our recent work has an effective solution for that!
Our paper about skill learning in RL has been accepted at CoRL 2024.
Curious about how to learning more effective policies with skill in RL than low-level actions? Then check out our recent work EXTRACT.
Our paper about a new and principled way to deal with target network has been accepted at ICML 2024.
Our two papers about drift estimation and efficient adaptation have been accepted at ICLR 2024.
Happy to share that AWS large model, Titan, finally got released and glad to be the core contributor of this amazing and challenging effort!
Check out our RL chapters in Dive into Deep Learning (D2L) book!
Our three papers about TD-convergence, batch-RL, and optimization in RL have been accepted at NeurIPS 2023.
Our Flexible Model Aggregation for Quantile Regression has been accepted at JMLR 2023.
Our Task-Agnostic Continual Reinforcement Learning: Gaining Insights and Overcoming Challenges has been accepted at CoLLAs 2023.
Our papers about proximal term for RL and Emphatic RL got accepted at NeurIPS 2022.
Selected as an ICML 2022 Outstanding Reviewer.
Selected as an ICLR 2022 Highlighted Reviewer.
Our batch RL work has been accepted at NeurIPS 2021.
Selected as an ICLR 2021 Outstanding Reviewer.
We publish a blog post about our MQL paper on Amazon Science.
Our Accurate, and Simple Models for Tabular Data via Augmented Distillation has been accepted at NeurIPS 2020.
Joint work with Pratik Chaudhari , Jonas Mueller, and Alexander Smola in transformer based density estimator is out.
Joint work with Pratik Chaudhari ,Stefano Soatto, and Alexander Smola in meta reinforcement learning (i.e. Meta-Q-Learning) has been accepted at ICLR 2020 as an oral paper.
We released P3O code.
Joint work with Pratik Chaudhari and Alexander Smola in combining on-policy and off-policy updates has been accepted at UAI 2019.
Our personal question answering work is featured on Amazon Science blog.
Our recent work in unfolding a greedy algorithm into a computational graph is out.
Our recent work in personal question answering will appear at SLT 2018.
Our recent work in constrained convolutional-recurrent networks will appear at ICASSP 2018.
Our recent work in black-box optimization using Reinforcement learning has been accepted at NIPS 2017 ML4Auido workshop.
I will be a research intern at Microsoft Research, Redmond, WA.
I will be interning with the Cognition AI group working with Pushmeet Kohli and Rishabh Singh at Microsoft Research, Redmond, WA.
Our joint work Pushmeet Kohli, Sing Bing Kang, Abdel-rahman Mohamed, and Margaret Mitchell, about video caption generation is available on arXiv and code is here.
I will be interning with the Cognition AI group working with Pushmeet Kohli, Sing Bing Kang, Abdel-rahman Mohamed, and Margaret Mitchell at Microsoft Research, Redmond, WA.
Our joint work with Dr. M. Bansal and Dr. M. Walter, "Deep Attribute-based Zero-shot Learning with Layer-specific Regularizers", has been accepted at NIPS 2015, Transfer and Multi-Task Learning Workshop.
Our joint work with Dr. M. Bansal and Dr. M. Walter, in Transfer learning and Deep Learning will be available soon.
Visiting Toyota Technological Institute at Chicago (TTIC) during summer and working with Dr. Mohit Bansal and Dr. Matthew Walter

Google Sites

Report abuse