Rasool Fakoor. Personal Webpage

Rasool Fakoor

My goal is to build AI agents that operate reliably across diverse settings, adapt to changing environments, follow instructions, reason, plan, and continually learn with minimal or no feedback. This goal builds on my work in efficient learning, generative models, reinforcement learning, and embodiment.

News

Our paper on LLM-based web agents has been accepted to ICLR 2025.
Our work on enhancing alignment in large models has been accepted to TMLR 2025.
Our paper on learning skills from offline robot data has been accepted to CoRL 2024.
Thrilled to announce the release of AWS's RLHF model, Titan, and proud to be a core contributor to this ambitious project!
Our work on a principled approach to handling target networks in RL has been accepted to ICML 2024.
Excited to announce the release of the Reinforcement Learning chapters in the Dive into Deep Learning (D2L) book!
Our paper introducing a new method to handle distribution shift in ever-changing environments has been accepted to ICLR 2024.
Our paper on efficient fine-tuning for task-specific imitation learning accepted to ICLR 2024.

Publications (Highlighted / All )

Book Chapters

Dive into Deep Learning
Zhang, et. al.
Website

I co-authored all content on RL in Volume II.

This website's template is largely adapted from this.