Aditya Ramesh

Hotspot Mitigations for the Masses

2019

In an IaaS cloud, the dynamic VM scheduler observes and mitigates resource hotspots to maintain performance in an oversubscribed environment. Most systems are focused on schedulers that fit very large infrastructures, which lead to workload-dependent …

The Benefits of Model-Based Generalization in Reinforcement Learning

2022 · arXiv (Cornell University)

Model-Based Reinforcement Learning (RL) is widely believed to have the potential to improve sample efficiency by allowing an agent to synthesize large amounts of imagined experience. Experience Replay (ER) can be considered a simple kind …

Language Models are Few-Shot Learners

2020 · arXiv (Cornell University)

Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still …