The Hong Kong Polytechnic University

university

Verified

https://www.polyu.edu.hk/

Activity Feed Request to join this org

AI & ML interests

None defined yet.

Recent Activity

adamdad submitted a paper about 17 hours ago

SAE Interventions are Unreliable: Post-Intervention Recovery of Suppressed Behavior

adamdad submitted a paper 3 days ago

BadWorld: Adversarial Attacks on World Models

wyy-code submitted a paper 18 days ago

Not All Disagreement Is Learnable: Token Teachability in On-Policy Distillation

View all activity

Papers

SAE Interventions are Unreliable: Post-Intervention Recovery of Suppressed Behavior

BadWorld: Adversarial Attacks on World Models

View all Papers

PolyUHK 's papers 17

Submitted by

Xingyi Yang

SAE Interventions are Unreliable: Post-Intervention Recovery of Suppressed Behavior

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Xingyi Yang

BadWorld: Adversarial Attacks on World Models

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Yuanyi Wang

Access Sets Matter: Budgeting Expert Reads for Scalable Weight-Space Model Merging

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Yuanyi Wang

Not All Disagreement Is Learnable: Token Teachability in On-Policy Distillation

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Wenjun Wang

E-PMQ: Expert-Guided Post-Merge Quantization with Merged-Weight Anchoring

PolyUHK

The Hong Kong Polytechnic University

Submitted by

yanggangu

FeatCal: Feature Calibration for Post-Merging Models

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Yuanyi Wang

Model Merging Scaling Laws in Large Language Models

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Yuanyi Wang

Geometry Conflict: Explaining and Controlling Forgetting in LLM Continual Post-Training

PolyUHK

The Hong Kong Polytechnic University

Submitted by

SeanLee

ShadowPEFT: Shadow Network for Parameter-Efficient Fine-Tuning

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Anthony

Multiscale Switch for Semi-Supervised and Contrastive Learning in Medical Ultrasound Image Segmentation

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Maojun SUN

DARE: Aligning LLM Agents with the R Statistical Ecosystem via Distribution-Aware Retrieval

PolyUHK

The Hong Kong Polytechnic University

Submitted by

tomsawyer

CL4SE: A Context Learning Benchmark For Software Engineering Tasks

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Jinrui Zhang

Pretraining A Large Language Model using Distributed GPUs: A Memory-Efficient Decentralized Paradigm

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Maojun SUN

DSAEval: Evaluating Data Science Agents on a Wide Range of Real-World Data Science Problems

PolyUHK

The Hong Kong Polytechnic University

Submitted by

Xingyi Yang

DeContext as Defense: Safe Image Editing in Diffusion Transformers

PolyUHK

The Hong Kong Polytechnic University

Submitted by

yangxiao

SCALE: Selective Resource Allocation for Overcoming Performance Bottlenecks in Mathematical Test-time Scaling

PolyUHK

The Hong Kong Polytechnic University

Syllable based DNN-HMM Cantonese Speech to Text System

PolyUHK

The Hong Kong Polytechnic University

1