mit-han-lab

[ICLR 2024] Efficient Streaming Language Models with Attention Sinks

[MLSys 2024 Best Paper Award] AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration

Efficient vision foundation models for high-resolution generation and perception.

[ICCV 2019] TSM: Temporal Shift Module for Efficient Video Understanding

[ICML 2023] SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models

A PyTorch-based framework for Quantum Classical Simulation, Quantum Machine Learning, Quantum Neural Networks, Parameterized Quantum Circuits with support for easy deployments on real quantum computers.

[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.

[ICLR 2019] ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware

[NeurIPS 2020] Differentiable Augmentation for Data-Efficient GAN Training

[CVPR 2020] GAN Compression: Efficient Architectures for Interactive Conditional GANs

StreamingVLM: Real-Time Understanding for Infinite Video Streams

TinyChatEngine: On-Device LLM Inference Library

[NeurIPS 2020] MCUNet: Tiny Deep Learning on IoT Devices; [NeurIPS 2021] MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning; [NeurIPS 2022] MCUNetV3: On-Device Training Under 256KB Memory

[MLSys'25] QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving; [MLSys'25] LServe: Efficient Long-sequence LLM Serving with Unified Sparse Attention

[CVPR 2024 Highlight] DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Models

[IJCV] FastComposer: Tuning-Free Multi-Subject Image Generation with Localized Attention

[NeurIPS 2020] MCUNet: Tiny Deep Learning on IoT Devices; [NeurIPS 2021] MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning

HART: Efficient Visual Generation with Hybrid Autoregressive Transformer

[NeurIPS 2025] Radial Attention: O(nlogn) Sparse Attention with Energy Decay for Long Video Generation

A sparse attention kernel supporting mix sparse patterns

[ICLR 2025] DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads

On-Device Training Under 256KB Memory [NeurIPS'22]