Percy Liang

papers

5,191

total citations

papers (29)

Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Models

ICML 2024arXiv

258

citations

Deep Bidirectional Language-Knowledge Graph Pretraining

NEURIPS 2022arXiv

251

citations

Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training

ICLR 2024arXiv

241

citations

Holistic Evaluation of Text-to-Image Models

NEURIPS 2023arXiv

178

citations

MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation

ICML 2024arXiv

168

citations

Large Language Models as Analogical Reasoners

ICLR 2024arXiv

137

citations

Decentralized Training of Foundation Models in Heterogeneous Environments

NEURIPS 2022arXiv

126

citations

Picking on the Same Person: Does Algorithmic Monoculture lead to Outcome Homogenization?

NEURIPS 2022arXiv

116

citations

PRODIGY: Enabling In-context Learning Over Graphs

NEURIPS 2023arXiv

103

citations

Enabling certification of verification-agnostic networks via memory-efficient semidefinite programming

NEURIPS 2020arXiv

101

citations

On the Learnability of Watermarks for Language Models

ICLR 2024arXiv

citations

BioDiscoveryAgent: An AI Agent for Designing Genetic Perturbation Experiments

ICLR 2025arXiv

citations

Improving Self-Supervised Learning by Characterizing Idealized Representations

NEURIPS 2022arXiv

citations

Benchmarking and Improving Generator-Validator Consistency of Language Models

ICLR 2024arXiv

citations

Insights into Pre-training via Simpler Synthetic Tasks

NEURIPS 2022arXiv

citations

Model Equality Testing: Which Model is this API Serving?

ICLR 2025arXiv

citations

Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes

NEURIPS 2023arXiv

citations

Establishing Best Practices in Building Rigorous Agentic Benchmarks

NEURIPS 2025arXiv

citations

Reliable and Efficient Amortized Model-based Evaluation

ICML 2025arXiv

citations

Position: In-House Evaluation Is Not Enough. Towards Robust Third-Party Evaluation and Flaw Disclosure for General-Purpose AI

ICML 2025

citations

Lexinvariant Language Models

NEURIPS 2023arXiv

citations

Cheaply Estimating Inference Efficiency Metrics for Autoregressive Transformer Models

NEURIPS 2023

citations

Position: A Safe Harbor for AI Evaluation and Red Teaming

ICML 2024

citations

Position: On the Societal Impact of Open Foundation Models

ICML 2024

citations

Percy Liang

papers (29)

Diffusion-LM Improves Controllable Text Generation

AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback

What Can Transformers Learn In-Context? A Case Study of Simple Function Classes

DoReMi: Optimizing Data Mixtures Speeds Up Language Model Pretraining

Data Selection for Language Models via Importance Resampling

Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Models

Deep Bidirectional Language-Knowledge Graph Pretraining

Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training

Holistic Evaluation of Text-to-Image Models

MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation

Large Language Models as Analogical Reasoners

Decentralized Training of Foundation Models in Heterogeneous Environments

Picking on the Same Person: Does Algorithmic Monoculture lead to Outcome Homogenization?

PRODIGY: Enabling In-context Learning Over Graphs

Enabling certification of verification-agnostic networks via memory-efficient semidefinite programming

On the Learnability of Watermarks for Language Models

BioDiscoveryAgent: An AI Agent for Designing Genetic Perturbation Experiments

Improving Self-Supervised Learning by Characterizing Idealized Representations

Benchmarking and Improving Generator-Validator Consistency of Language Models

Insights into Pre-training via Simpler Synthetic Tasks

Model Equality Testing: Which Model is this API Serving?

Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes

Establishing Best Practices in Building Rigorous Agentic Benchmarks

Reliable and Efficient Amortized Model-based Evaluation

Position: In-House Evaluation Is Not Enough. Towards Robust Third-Party Evaluation and Flaw Disclosure for General-Purpose AI

Lexinvariant Language Models

Cheaply Estimating Inference Efficiency Metrics for Autoregressive Transformer Models

Position: A Safe Harbor for AI Evaluation and Red Teaming

Position: On the Societal Impact of Open Foundation Models

papers (29)

Diffusion-LM Improves Controllable Text Generation

AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback

What Can Transformers Learn In-Context? A Case Study of Simple Function Classes

DoReMi: Optimizing Data Mixtures Speeds Up Language Model Pretraining

Data Selection for Language Models via Importance Resampling

Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Models

Deep Bidirectional Language-Knowledge Graph Pretraining

Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training

Holistic Evaluation of Text-to-Image Models

MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation

Large Language Models as Analogical Reasoners

Decentralized Training of Foundation Models in Heterogeneous Environments

Picking on the Same Person: Does Algorithmic Monoculture lead to Outcome Homogenization?

PRODIGY: Enabling In-context Learning Over Graphs

Enabling certification of verification-agnostic networks via memory-efficient semidefinite programming

On the Learnability of Watermarks for Language Models

BioDiscoveryAgent: An AI Agent for Designing Genetic Perturbation Experiments

Improving Self-Supervised Learning by Characterizing Idealized Representations

Benchmarking and Improving Generator-Validator Consistency of Language Models

Insights into Pre-training via Simpler Synthetic Tasks

Model Equality Testing: Which Model is this API Serving?

Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes

Establishing Best Practices in Building Rigorous Agentic Benchmarks

Reliable and Efficient Amortized Model-based Evaluation

Position: In-House Evaluation Is Not Enough. Towards Robust Third-Party Evaluation and Flaw Disclosure for General-Purpose AI

Lexinvariant Language Models

Cheaply Estimating Inference Efficiency Metrics for Autoregressive Transformer Models

Position: A Safe Harbor for AI Evaluation and Red Teaming

Position: On the Societal Impact of Open Foundation Models