MIT HAN Lab

All

56 repositories

llm-awq
Public
[MLSys 2024 Best Paper Award] AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration
Python
•
MIT License
•214•2.6k•139•7•Updated Dec 20, 2024Dec 20, 2024
nunchaku
Public
SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
flux lora quantization diffusion-models mlsys genai
Cuda
•
Apache License 2.0
•27•535•34•2•Updated Dec 9, 2024Dec 9, 2024
efficientvit
Public
Efficient vision foundation models for high-resolution generation and perception.
imagenet segmentation high-resolution vision-transformer efficientvit segment-anything deep-compression-autoencoder efficient-diffusion-model
Python
•
Apache License 2.0
•197•2.5k•98•0•Updated Dec 9, 2024Dec 9, 2024
distrifuser
Public
[CVPR 2024 Highlight] DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Models
acceleration parallelism generative-model diffusion-models generative-ai
Python
•
MIT License
•24•616•8•1•Updated Dec 2, 2024Dec 2, 2024
tinyengine
Public
[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
c microcontroller cpp pytorch codegenerator tinyml deep-learning quantization edge-computing neural-architecture-search
C
•
MIT License
•133•822•34•1•Updated Nov 27, 2024Nov 27, 2024
Quest
Public
[ICML 2024] Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference
Cuda
•21•220•4•0•Updated Nov 22, 2024Nov 22, 2024
torchquantum
Public
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.
machine-learning system deep-learning neural-network quantum pytorch quantum-computing quantum-machine-learning quantum-simulation ml-for-systems
Jupyter Notebook
•
MIT License
•208•1.4k•60•8•Updated Nov 19, 2024Nov 19, 2024
torchsparse
Public
[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.
acceleration pytorch
Cuda
•
MIT License
•143•1.2k•28•2•Updated Nov 12, 2024Nov 12, 2024
deepcompressor
Public
Model Compression Toolbox for Large Language Models and Diffusion Models
Python
•
Apache License 2.0
•22•274•27•1•Updated Nov 10, 2024Nov 10, 2024
qserve
Public
QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving
Python
•
Apache License 2.0
•28•466•28•4•Updated Nov 9, 2024Nov 9, 2024
tinychat-tutorial
Public
C++
•20•57•4•2•Updated Nov 5, 2024Nov 5, 2024
duo-attention
Public
DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
Python
•
MIT License
•21•408•5•0•Updated Oct 31, 2024Oct 31, 2024
vila-u
Public
VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation
Python
•
MIT License
•3•179•9•0•Updated Oct 24, 2024Oct 24, 2024
hart
Public
HART: Efficient Visual Generation with Hybrid Autoregressive Transformer
Python
•
MIT License
•19•385•9•2•Updated Oct 16, 2024Oct 16, 2024
Block-Sparse-Attention
Public
A sparse attention kernel supporting mix sparse patterns
C++
•
BSD 3-Clause "New" or "Revised" License
•2•82•5•0•Updated Oct 15, 2024Oct 15, 2024
data-efficient-gans
Public
[NeurIPS 2020] Differentiable Augmentation for Data-Efficient GAN Training
tensorflow generative-adversarial-network image-generation gans data-efficient neurips-2020 pytorch
Python
•
BSD 2-Clause "Simplified" License
•175•1.3k•28•0•Updated Sep 24, 2024Sep 24, 2024
proxylessnas
Public
[ICLR 2019] ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware
acceleration automl specialization efficient-model on-device-ai hardware-aware
C++
•
MIT License
•285•1.4k•0•2•Updated Aug 30, 2024Aug 30, 2024
spatten
Public
[HPCA'21] SpAtten: Efficient Sparse Attention Architecture with Cascade Token and Head Pruning
rtl attention hardware-acceleration spinalhdl llm-inference
Scala
•
MIT License
•9•78•1•0•Updated Aug 27, 2024Aug 27, 2024
fastcomposer
Public
[IJCV] FastComposer: Tuning-Free Multi-Subject Image Generation with Localized Attention
Python
•
MIT License
•38•675•16•0•Updated Aug 21, 2024Aug 21, 2024
bevfusion
Public archive
[ICRA'23] BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation
camera pytorch lidar object-detection sensor-fusion semantic-segmentation 3d-perception
Python
•
Apache License 2.0
•435•2.4k•0•0•Updated Jul 31, 2024Jul 31, 2024
hardware-aware-transformers
Public
[ACL'20] HAT: Hardware-Aware Transformers for Efficient Natural Language Processing
natural-language-processing machine-translation transformer specialization efficient-model hardware-aware
Python
•
Other
•53•331•3•0•Updated Jul 14, 2024Jul 14, 2024
smoothquant
Public
[ICML 2023] SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models
Python
•
MIT License
•151•1.3k•66•1•Updated Jul 12, 2024Jul 12, 2024
spvnas
Public archive
[ECCV 2020] Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
computer-vision deep-learning efficiency pytorch lidar architecture-search point-cloud 3d-deep-learning semantickitti
Python
•
MIT License
•110•588•0•0•Updated Jul 11, 2024Jul 11, 2024
lite-transformer
Public archive
[ICLR 2020] Lite Transformer with Long-Short Range Attention
nlp pytorch transformer
Python
•
Other
•81•601•0•0•Updated Jul 11, 2024Jul 11, 2024
temporal-shift-module
Public
[ICCV 2019] TSM: Temporal Shift Module for Efficient Video Understanding
acceleration low-latency video-understanding efficient-model temporal-modeling tsm nvidia-jetson-nano
Python
•
MIT License
•416•2.1k•94•6•Updated Jul 11, 2024Jul 11, 2024
streaming-llm
Public
[ICLR 2024] Efficient Streaming Language Models with Attention Sinks
Python
•
MIT License
•371•6.7k•43•2•Updated Jul 11, 2024Jul 11, 2024
TinyChatEngine
Public
TinyChatEngine: On-Device LLM Inference Library
c arm deep-learning cpp x86-64 quantization edge-computing cuda-programming on-device-ai large-language-models
C++
•
MIT License
•74•776•34•5•Updated Jul 4, 2024Jul 4, 2024
litepose
Public
[CVPR'22] Lite Pose: Efficient Architecture Design for 2D Human Pose Estimation
pose-estimation efficient-models litepose
Python
•
MIT License
•36•311•19•1•Updated Jun 5, 2024Jun 5, 2024
gan-compression
Public
[CVPR 2020] GAN Compression: Efficient Architectures for Interactive Conditional GANs
compression pytorch gans pix2pix cyclegan image-to-image-translation conditional-gans gaugan
Python
•
Other
•150•1.1k•3•6•Updated Jun 5, 2024Jun 5, 2024
patch_conv
Public
Patch convolution to avoid large GPU memory usage of Conv2D
Python
•
MIT License
•5•80•1•1•Updated May 26, 2024May 26, 2024