FlashInfer: Kernel Library for LLM Serving

SparseDrive: End-to-End Autonomous Driving via Sparse Scene Representation

Quantized Attention that achieves speedups of 2.1x and 2.7x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.

This project analyzes Tennis players in a video to measure their speed, ball shot speed and number of shots. This project will detect players and the tennis ball using YOLO and also utilizes CNNs t…

SGLang is a fast serving framework for large language models and vision language models.

Qwen2-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud.

PLUTO: Push the Limit of Imitation Learning-based Planning for Autonomous Driving

Brand new TTS solution

Official code implementation of General OCR Theory: Towards OCR-2.0 via a Unified End-to-end Model

[ICLR2024 spotlight] OmniQuant is a simple and powerful quantization technique for LLMs.

The repository provides code for running inference with the Meta Segment Anything Model 2 (SAM 2), links for downloading the trained model checkpoints, and example notebooks that show how to use th…

Instant voice cloning by MIT and MyShell.

Material for gpu-mode lectures

LLM101n: Let's build a Storyteller

A generative speech model for daily dialogue.

Tile primitives for speedy kernels

1 min voice data can also be used to train a good TTS model! (few shot voice cloning)

本人的科研经验

A minimal GPU design in Verilog to learn how GPUs work from the ground up

[ECCV 2024] Fully Sparse 3D Occupancy Prediction & RayIoU Evaluation Metric

Multiple Object Tracking as ID Prediction

[CVPR 2023] VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking

[NeurIPS 2022 Spotlight] VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training

The suite of modeling video with Mamba

Code release for ActionFormer (ECCV 2022)

PySlowFast: video understanding codebase from FAIR for reproducing state-of-the-art video models.

Unofficial implementation of "TTNet: Real-time temporal and spatial video analysis of table tennis" (CVPR 2020)

RTMPose series (RTMPose, DWPose, RTMO, RTMW) without mmcv, mmpose, mmdet etc.

Hiera: A fast, powerful, and simple hierarchical vision transformer.