Do Question Answering Modeling Improvements Hold Across Benchmarks?
Comments 31 pages, 13 figures; to appear at ACL 2023
作者
Percy Liang
Natural Language Processing
Beyond Positive Scaling: How Negation Impacts Scaling Trends of Language Models
Comments Published at ACL 2023 Findings
Backpack Language Models
Comments ACL 2023 Camera-Ready
Lexinvariant Language Models
PRODIGY: Enabling In-context Learning Over Graphs
Cheaply Evaluating Inference Efficiency Metrics for Autoregressive Transformer APIs
Whose Opinions Do Language Models Reflect?
Foundation Models and Fair Use
Improving Representational Continuity via Continued Pretraining
Language-Driven Representation Learning for Robotics
Comments 30 Pages, 15 Figures
Benchmarking Large Language Models for News Summarization
Demonstrate-Search-Predict: Composing retrieval and language models for knowledge-intensive NLP
"No, to the Right" -- Online Language Corrections for Robotic Manipulation via Shared Autonomy
Comments Accepted to the 18th ACM/IEEE International Conference on Human Robot Interaction (HRI), Marc 2023. First two authors contributed equally. 9 Pages, 7 Figures
QA-GNN: Reasoning with Language Models and Knowledge Graphs for Question Answering
Comments NAACL 2021. Code & data available at https://github.com/michiyasunaga/qagnn
Improving Self-Supervised Learning by Characterizing Idealized Representations
Comments Accepted at NeurIPS 2022
Melody transcription via generative pre-training
Comments Published as a conference paper at ISMIR 2022
Connect, Not Collapse: Explaining Contrastive Learning for Unsupervised Domain Adaptation
Comments ICML 2022 (Long Talk)
Picking on the Same Person: Does Algorithmic Monoculture lead to Outcome Homogenization?
Comments Published at NeurIPS 2022, presented at EAAMO 2022
How do Authors' Perceptions of their Papers Compare with Co-authors' Perceptions and Peer-review Decisions?
Large Language Models Can Be Strong Differentially Private Learners
Comments 31 pages; update ethics statement to clarify benefits and potential long-term harms
Truncation Sampling as Language Model Desmoothing
Comments Findings of EMNLP, + small fixes
Emergent Abilities of Large Language Models
Comments Transactions on Machine Learning Research (TMLR), 2022
Deep Bidirectional Language-Knowledge Graph Pretraining
Comments Published at NeurIPS 2022. Code, data, and trained models are available at https://github.com/michiyasunaga/dragon
Social Simulacra: Creating Populated Prototypes for Social Computing Systems
Comments This work will appear in the 35th Annual ACM Symposium on User Interface Software and Technology (UIST '22)
An Explanation of In-context Learning as Implicit Bayesian Inference
Comments ICLR 2022
Calibrated ensembles can mitigate accuracy tradeoffs under distribution shift
Comments Accepted to UAI 2022
Is a Caption Worth a Thousand Images? A Controlled Study for Representation Learning
On the Opportunities and Risks of Foundation Models
Comments Authored by the Center for Research on Foundation Models (CRFM) at the Stanford Institute for Human-Centered Artificial Intelligence (HAI). Report page with citation guidelines: https://crfm.stanford.edu/report.html
Insights into Pre-training via Simpler Synthetic Tasks
Comments 30 pages
