ServiceNow Research

Efficient Learning

Breadth-First Pipeline Parallelism
Breadth-First Pipeline Parallelism
We introduce Breadth-First Pipeline Parallelism, a novel training schedule which optimizes the combination of pipeline and data …
Conference on Machine Learning and Systems,  2023.
PDF Cite Code
Unsupervised Model-based Pre-training for Data-efficient Reinforcement Learning from Pixels
Unsupervised Model-based Pre-training for Data-efficient Reinforcement Learning from Pixels
Reinforcement learning (RL) aims at autonomously performing complex tasks. To this end, a reward signal is used to steer the learning …
Workshop at the International Conference on Machine Learning (ICML),  2022.
PDF Cite
Conditionally Adaptive Multi-Task Learning: Improving Transfer Learning in NLP Using Fewer Parameters & Less Data
Conditionally Adaptive Multi-Task Learning: Improving Transfer Learning in NLP Using Fewer Parameters & Less Data
Multi-Task Learning (MTL) networks have emerged as a promising method for transferring learned knowledge across different tasks. …
International Conference on Learning Representations (ICLR),  2021.
PDF Cite Code