Picture for Wenqiang Lei

Wenqiang Lei

Sichuan University

Legend: Leveraging Representation Engineering to Annotate Safety Margin for Preference Datasets

Add code
Jun 12, 2024
Viaarxiv icon

Dishonesty in Helpful and Harmless Alignment

Add code
Jun 04, 2024
Viaarxiv icon

ARAIDA: Analogical Reasoning-Augmented Interactive Data Annotation

Add code
May 20, 2024
Viaarxiv icon

Selective Annotation via Data Allocation: These Data Should Be Triaged to Experts for Annotation Rather Than the Model

Add code
May 20, 2024
Viaarxiv icon

STYLE: Improving Domain Transferability of Asking Clarification Questions in Large Language Model Powered Conversational Agents

Add code
May 20, 2024
Figure 1 for STYLE: Improving Domain Transferability of Asking Clarification Questions in Large Language Model Powered Conversational Agents
Figure 2 for STYLE: Improving Domain Transferability of Asking Clarification Questions in Large Language Model Powered Conversational Agents
Figure 3 for STYLE: Improving Domain Transferability of Asking Clarification Questions in Large Language Model Powered Conversational Agents
Figure 4 for STYLE: Improving Domain Transferability of Asking Clarification Questions in Large Language Model Powered Conversational Agents
Viaarxiv icon

CLAMBER: A Benchmark of Identifying and Clarifying Ambiguous Information Needs in Large Language Models

Add code
May 20, 2024
Figure 1 for CLAMBER: A Benchmark of Identifying and Clarifying Ambiguous Information Needs in Large Language Models
Figure 2 for CLAMBER: A Benchmark of Identifying and Clarifying Ambiguous Information Needs in Large Language Models
Figure 3 for CLAMBER: A Benchmark of Identifying and Clarifying Ambiguous Information Needs in Large Language Models
Figure 4 for CLAMBER: A Benchmark of Identifying and Clarifying Ambiguous Information Needs in Large Language Models
Viaarxiv icon

Co-Matching: Towards Human-Machine Collaborative Legal Case Matching

Add code
May 16, 2024
Viaarxiv icon

Towards Analyzing and Understanding the Limitations of DPO: A Theoretical Perspective

Add code
Apr 06, 2024
Figure 1 for Towards Analyzing and Understanding the Limitations of DPO: A Theoretical Perspective
Viaarxiv icon

Concept -- An Evaluation Protocol on Conversation Recommender Systems with System-centric and User-centric Factors

Add code
Apr 06, 2024
Viaarxiv icon

Strength Lies in Differences! Towards Effective Non-collaborative Dialogues via Tailored Strategy Planning

Add code
Mar 11, 2024
Figure 1 for Strength Lies in Differences! Towards Effective Non-collaborative Dialogues via Tailored Strategy Planning
Figure 2 for Strength Lies in Differences! Towards Effective Non-collaborative Dialogues via Tailored Strategy Planning
Figure 3 for Strength Lies in Differences! Towards Effective Non-collaborative Dialogues via Tailored Strategy Planning
Figure 4 for Strength Lies in Differences! Towards Effective Non-collaborative Dialogues via Tailored Strategy Planning
Viaarxiv icon