Picture for Zhaoyi Li

Zhaoyi Li

A Frustratingly Simple Yet Highly Effective Attack Baseline: Over 90% Success Rate Against the Strong Black-box Models of GPT-4.5/4o/o1

Add code
Mar 13, 2025
Viaarxiv icon

Learning to Substitute Components for Compositional Generalization

Add code
Feb 28, 2025
Viaarxiv icon

General Time-series Model for Universal Knowledge Representation of Multivariate Time-Series data

Add code
Feb 05, 2025
Figure 1 for General Time-series Model for Universal Knowledge Representation of Multivariate Time-Series data
Figure 2 for General Time-series Model for Universal Knowledge Representation of Multivariate Time-Series data
Figure 3 for General Time-series Model for Universal Knowledge Representation of Multivariate Time-Series data
Figure 4 for General Time-series Model for Universal Knowledge Representation of Multivariate Time-Series data
Viaarxiv icon

Dataset Distillation via Committee Voting

Add code
Jan 13, 2025
Viaarxiv icon

ICODE: Modeling Dynamical Systems with Extrinsic Input Information

Add code
Nov 21, 2024
Figure 1 for ICODE: Modeling Dynamical Systems with Extrinsic Input Information
Figure 2 for ICODE: Modeling Dynamical Systems with Extrinsic Input Information
Figure 3 for ICODE: Modeling Dynamical Systems with Extrinsic Input Information
Figure 4 for ICODE: Modeling Dynamical Systems with Extrinsic Input Information
Viaarxiv icon

Mitigating the Language Mismatch and Repetition Issues in LLM-based Machine Translation via Model Editing

Add code
Oct 09, 2024
Figure 1 for Mitigating the Language Mismatch and Repetition Issues in LLM-based Machine Translation via Model Editing
Figure 2 for Mitigating the Language Mismatch and Repetition Issues in LLM-based Machine Translation via Model Editing
Figure 3 for Mitigating the Language Mismatch and Repetition Issues in LLM-based Machine Translation via Model Editing
Figure 4 for Mitigating the Language Mismatch and Repetition Issues in LLM-based Machine Translation via Model Editing
Viaarxiv icon

Interpretable Catastrophic Forgetting of Large Language Model Fine-tuning via Instruction Vector

Add code
Jun 18, 2024
Figure 1 for Interpretable Catastrophic Forgetting of Large Language Model Fine-tuning via Instruction Vector
Figure 2 for Interpretable Catastrophic Forgetting of Large Language Model Fine-tuning via Instruction Vector
Figure 3 for Interpretable Catastrophic Forgetting of Large Language Model Fine-tuning via Instruction Vector
Figure 4 for Interpretable Catastrophic Forgetting of Large Language Model Fine-tuning via Instruction Vector
Viaarxiv icon

Benchmarking and Improving Compositional Generalization of Multi-aspect Controllable Text Generation

Add code
Apr 05, 2024
Viaarxiv icon

Understanding and Patching Compositional Reasoning in LLMs

Add code
Feb 22, 2024
Viaarxiv icon

Invariant Representation Learning via Decoupling Style and Spurious Features

Add code
Dec 11, 2023
Viaarxiv icon