Picture for Liyuan Liu

Liyuan Liu

DHRNet: A Dual-Path Hierarchical Relation Network for Multi-Person Pose Estimation

Add code
Apr 27, 2024
Viaarxiv icon

Learning a Decision Tree Algorithm with Transformers

Add code
Feb 06, 2024
Viaarxiv icon

Tell Your Model Where to Attend: Post-hoc Attention Steering for LLMs

Add code
Nov 03, 2023
Viaarxiv icon

Fast-ELECTRA for Efficient Pre-training

Add code
Oct 11, 2023
Figure 1 for Fast-ELECTRA for Efficient Pre-training
Figure 2 for Fast-ELECTRA for Efficient Pre-training
Figure 3 for Fast-ELECTRA for Efficient Pre-training
Figure 4 for Fast-ELECTRA for Efficient Pre-training
Viaarxiv icon

Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs

Add code
Oct 07, 2023
Figure 1 for Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs
Figure 2 for Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs
Figure 3 for Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs
Figure 4 for Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs
Viaarxiv icon

Sparse Backpropagation for MoE Training

Add code
Oct 01, 2023
Viaarxiv icon

Bridging Discrete and Backpropagation: Straight-Through and Beyond

Add code
Apr 17, 2023
Figure 1 for Bridging Discrete and Backpropagation: Straight-Through and Beyond
Figure 2 for Bridging Discrete and Backpropagation: Straight-Through and Beyond
Figure 3 for Bridging Discrete and Backpropagation: Straight-Through and Beyond
Figure 4 for Bridging Discrete and Backpropagation: Straight-Through and Beyond
Viaarxiv icon

SoTeacher: A Student-oriented Teacher Network Training Framework for Knowledge Distillation

Add code
Jun 14, 2022
Figure 1 for SoTeacher: A Student-oriented Teacher Network Training Framework for Knowledge Distillation
Figure 2 for SoTeacher: A Student-oriented Teacher Network Training Framework for Knowledge Distillation
Figure 3 for SoTeacher: A Student-oriented Teacher Network Training Framework for Knowledge Distillation
Figure 4 for SoTeacher: A Student-oriented Teacher Network Training Framework for Knowledge Distillation
Viaarxiv icon

PILED: An Identify-and-Localize Framework for Few-Shot Event Detection

Add code
Feb 15, 2022
Figure 1 for PILED: An Identify-and-Localize Framework for Few-Shot Event Detection
Figure 2 for PILED: An Identify-and-Localize Framework for Few-Shot Event Detection
Figure 3 for PILED: An Identify-and-Localize Framework for Few-Shot Event Detection
Figure 4 for PILED: An Identify-and-Localize Framework for Few-Shot Event Detection
Viaarxiv icon

Double Descent in Adversarial Training: An Implicit Label Noise Perspective

Add code
Oct 07, 2021
Figure 1 for Double Descent in Adversarial Training: An Implicit Label Noise Perspective
Figure 2 for Double Descent in Adversarial Training: An Implicit Label Noise Perspective
Figure 3 for Double Descent in Adversarial Training: An Implicit Label Noise Perspective
Figure 4 for Double Descent in Adversarial Training: An Implicit Label Noise Perspective
Viaarxiv icon