Zechun Liu

papers

2,039

total citations

papers (14)

Single Path One-Shot Neural Architecture Search with Uniform Sampling

ECCV 2020arXiv

1,003

citations

ReActNet: Towards Precise Binary Neural Network with Generalized Activation Functions

ECCV 2020arXiv

404

citations

MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases

ICML 2024arXiv

195

citations

Nonuniform-to-Uniform Quantization: Towards Accurate Quantization via Generalized Straight-Through Estimation

CVPR 2022arXiv

146

citations

Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space

CVPR 2022arXiv

citations

S2-BNN: Bridging the Gap Between Self-Supervised Real and 1-Bit Neural Networks via Guided Distribution Calibration

CVPR 2021arXiv

citations

Efficient Track Anything

ICCV 2025arXiv

citations

Data-Free Neural Architecture Search via Recursive Label Calibration

ECCV 2022arXiv

citations

PARQ: Piecewise-Affine Regularized Quantization

ICML 2025arXiv

citations

RDD: Retrieval-Based Demonstration Decomposer for Planner Alignment in Long-Horizon Tasks

NEURIPS 2025arXiv

citations

Param$\Delta$ for Direct Mixing: Post-Train Large Language Model At Zero Cost

ICLR 2025

citations

Zechun Liu

papers (14)

Single Path One-Shot Neural Architecture Search with Uniform Sampling

ReActNet: Towards Precise Binary Neural Network with Generalized Activation Functions

MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases

Nonuniform-to-Uniform Quantization: Towards Accurate Quantization via Generalized Straight-Through Estimation

Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space

BiT: Robustly Binarized Multi-distilled Transformer

Binarizing MobileNet via Evolution-Based Searching

Sliced Recursive Transformer

S2-BNN: Bridging the Gap Between Self-Supervised Real and 1-Bit Neural Networks via Guided Distribution Calibration

Efficient Track Anything

Data-Free Neural Architecture Search via Recursive Label Calibration

PARQ: Piecewise-Affine Regularized Quantization

RDD: Retrieval-Based Demonstration Decomposer for Planner Alignment in Long-Horizon Tasks

Param$\Delta$ for Direct Mixing: Post-Train Large Language Model At Zero Cost

papers (14)

Single Path One-Shot Neural Architecture Search with Uniform Sampling

ReActNet: Towards Precise Binary Neural Network with Generalized Activation Functions

MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases

Nonuniform-to-Uniform Quantization: Towards Accurate Quantization via Generalized Straight-Through Estimation

Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space

BiT: Robustly Binarized Multi-distilled Transformer

Binarizing MobileNet via Evolution-Based Searching

Sliced Recursive Transformer

S2-BNN: Bridging the Gap Between Self-Supervised Real and 1-Bit Neural Networks via Guided Distribution Calibration

Efficient Track Anything

Data-Free Neural Architecture Search via Recursive Label Calibration

PARQ: Piecewise-Affine Regularized Quantization

RDD: Retrieval-Based Demonstration Decomposer for Planner Alignment in Long-Horizon Tasks

Param$\Delta$ for Direct Mixing: Post-Train Large Language Model At Zero Cost