Shiwei Liu

Affiliations

UT Austin

papers

493

total citations

papers (17)

Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High Sparsity

ICML 2024arXiv

152

citations

Sparse Training via Boosting Pruning Plasticity with Neuroregeneration

NEURIPS 2021arXiv

135

citations

The Emergence of Essential Sparsity in Large Pre-trained Models: The Weights that Matter

NEURIPS 2023arXiv

citations

Dynamic Sparse Network for Time Series Classification: Learning What to “See”

NEURIPS 2022arXiv

citations

Visual Prompting Upgrades Neural Network Sparsification: A Data-Model Perspective

AAAI 2025arXiv

citations

Mix-LN: Unleashing the Power of Deeper Layers by Combining Pre-LN and Post-LN

ICLR 2025arXiv

citations

Dynamic Sparsity Is Channel-Level Sparsity Learner

NEURIPS 2023arXiv

citations

Predicting mutational effects on protein-protein binding via a side-chain diffusion probabilistic model

NEURIPS 2023arXiv

citations

From Low Rank Gradient Subspace Stabilization to Low-Rank Weights: Observations, Theories, and Applications

ICML 2025arXiv

citations

Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More

ICML 2025arXiv

citations

Data Augmented Flatness-aware Gradient Projection for Continual Learning

ICCV 2023

citations

Junk DNA Hypothesis: Pruning Small Pre-Trained Weights $\textit{Irreversibly}$ and $\textit{Monotonically}$ Impairs ``Difficult" Downstream Tasks in LLMs

ICML 2024

citations

Shiwei Liu

Affiliations

papers (17)

Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High Sparsity

Sparse Training via Boosting Pruning Plasticity with Neuroregeneration

The Emergence of Essential Sparsity in Large Pre-trained Models: The Weights that Matter

Dynamic Sparse Network for Time Series Classification: Learning What to “See”

Visual Prompting Upgrades Neural Network Sparsification: A Data-Model Perspective

Mix-LN: Unleashing the Power of Deeper Layers by Combining Pre-LN and Post-LN

Dynamic Sparsity Is Channel-Level Sparsity Learner

Predicting mutational effects on protein-protein binding via a side-chain diffusion probabilistic model

From Low Rank Gradient Subspace Stabilization to Low-Rank Weights: Observations, Theories, and Applications

Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More

Data Augmented Flatness-aware Gradient Projection for Continual Learning

Junk DNA Hypothesis: Pruning Small Pre-Trained Weights $\textit{Irreversibly}$ and $\textit{Monotonically}$ Impairs ``Difficult" Downstream Tasks in LLMs

Advancing Dynamic Sparse Training by Exploring Optimization Opportunities

Don’t just prune by magnitude! Your mask topology is a secret weapon

Sparse Cocktail: Every Sparse Pattern Every Sparse Ratio All At Once

CaM: Cache Merging for Memory-efficient LLMs Inference

Towards Data-Agnostic Pruning At Initialization: What Makes a Good Sparse Mask?

papers (17)

Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High Sparsity

Sparse Training via Boosting Pruning Plasticity with Neuroregeneration

The Emergence of Essential Sparsity in Large Pre-trained Models: The Weights that Matter

Dynamic Sparse Network for Time Series Classification: Learning What to “See”

Visual Prompting Upgrades Neural Network Sparsification: A Data-Model Perspective

Mix-LN: Unleashing the Power of Deeper Layers by Combining Pre-LN and Post-LN

Dynamic Sparsity Is Channel-Level Sparsity Learner

Predicting mutational effects on protein-protein binding via a side-chain diffusion probabilistic model

From Low Rank Gradient Subspace Stabilization to Low-Rank Weights: Observations, Theories, and Applications

Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More

Data Augmented Flatness-aware Gradient Projection for Continual Learning

Junk DNA Hypothesis: Pruning Small Pre-Trained Weights $\textit{Irreversibly}$ and $\textit{Monotonically}$ Impairs ``Difficult" Downstream Tasks in LLMs

Advancing Dynamic Sparse Training by Exploring Optimization Opportunities

Don’t just prune by magnitude! Your mask topology is a secret weapon

Sparse Cocktail: Every Sparse Pattern Every Sparse Ratio All At Once

CaM: Cache Merging for Memory-efficient LLMs Inference

Towards Data-Agnostic Pruning At Initialization: What Makes a Good Sparse Mask?