Eric P. Xing

EvoLM: In Search of Lost Language Model Training Dynamics

What is Your Data Worth to GPT? LLM-Scale Data Valuation with Influence Functions

CITER: Collaborative Inference for Efficient Large Language Model Decoding with Token-Level Routing

Faster Video Diffusion with Trainable Sparse Attention

Dimensional Collapse in VQVAEs: Evidence and Remedies

Pruning Spurious Subgraphs for Graph Out-of-Distribution Generalization

Follow My Instruction and Spill the Beans: Scalable Data Extraction from Retrieval-Augmented Generation Systems

MegaMath: Pushing the Limits of Open Math Corpora

Understanding the Skill Gap in Recurrent Language Models: The Role of the Gather-and-Aggregate Mechanism

Pruning Spurious Subgraphs for Graph Out-of-Distribtuion Generalization

Synthesizing Privacy-Preserving Text Data via Finetuning *without* Finetuning Billion-Scale LLMs

scPilot: Large Language Model Reasoning Toward Automated Single-Cell Analysis and Discovery

Data Mixing Optimization for Supervised Fine-Tuning of Large Language Models

Causal Representation Learning from Multimodal Biomedical Observations

How does controllability emerge in language models during pretraining?

Diversifying Spurious Subgraphs for Graph Out-of-Distribution Generalization

Scaling Long Context Training Data by Long-Distance Referrals

Large-Scale Training Data Attribution with Efficient Influence Functions

Learning Vision and Language Concepts for Controllable Image Generation

CITER: Collaborative Inference for Efficient Large Language Model Decoding with Token-Level Routing

ShareLoRA: Less Tuning, More Performance for LoRA Fine-tuning of LLMs

LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Dataset

LLM360: Towards Fully Transparent Open-Source LLMs

LQ-LoRA: Low-rank plus Quantized Matrix Decomposition for Efficient Language Model Finetuning

Fusing Models with Complementary Expertise

Crystal: Illuminating LLM Abilities on Language and Code

DISTFLASHATTN: Distributed Memory-efficient Attention for Long-context LLMs Training

Learning Discrete Concepts in Latent Hierarchical Models

Towards Understanding Extrapolation: a Causal Lens

PromptAgent: Strategic Planning with Language Models Enables Expert-level Prompt Optimization

Transformers to SSMs: Distilling Quadratic Knowledge to Subquadratic Models

One-for-All: Generalized LoRA for Parameter-Efficient Fine-tuning

Memoization-Aware Bayesian Optimization for AI Pipelines with Unknown Costs

LightSeq: Sequence Level Parallelism for Distributed Training of Long Context Transformers

Contextualized Policy Recovery: Modeling and Interpreting Medical Decisions with Adaptive Imitation Learning

Liteformer: Lightweight Evoformer for Protein Structure Prediction

Generation, Reconstruction, Representation All-in-One: A Joint Autoencoding Diffusion Model