Picture for Pinjia He

Pinjia He

Difficult Task Yes but Simple Task No: Unveiling the Laziness in Multimodal LLMs

Add code
Oct 15, 2024
Viaarxiv icon

Insight Over Sight? Exploring the Vision-Knowledge Conflicts in Multimodal LLMs

Add code
Oct 10, 2024
Figure 1 for Insight Over Sight? Exploring the Vision-Knowledge Conflicts in Multimodal LLMs
Figure 2 for Insight Over Sight? Exploring the Vision-Knowledge Conflicts in Multimodal LLMs
Figure 3 for Insight Over Sight? Exploring the Vision-Knowledge Conflicts in Multimodal LLMs
Figure 4 for Insight Over Sight? Exploring the Vision-Knowledge Conflicts in Multimodal LLMs
Viaarxiv icon

Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training

Add code
Jul 12, 2024
Viaarxiv icon

Aligning LLMs for FL-free Program Repair

Add code
Apr 13, 2024
Viaarxiv icon

A & B == B & A: Triggering Logical Reasoning Failures in Large Language Models

Add code
Jan 01, 2024
Viaarxiv icon

Retromorphic Testing: A New Approach to the Test Oracle Problem

Add code
Oct 10, 2023
Viaarxiv icon

An Image is Worth a Thousand Toxic Words: A Metamorphic Testing Framework for Content Moderation Software

Add code
Aug 18, 2023
Viaarxiv icon

Automated Testing and Improvement of Named Entity Recognition Systems

Add code
Aug 14, 2023
Figure 1 for Automated Testing and Improvement of Named Entity Recognition Systems
Figure 2 for Automated Testing and Improvement of Named Entity Recognition Systems
Figure 3 for Automated Testing and Improvement of Named Entity Recognition Systems
Figure 4 for Automated Testing and Improvement of Named Entity Recognition Systems
Viaarxiv icon

GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Add code
Aug 12, 2023
Viaarxiv icon

Validating Multimedia Content Moderation Software via Semantic Fusion

Add code
May 23, 2023
Viaarxiv icon