The PebblesDB write-optimized key-value store (SOSP 17)

nanobind: tiny and efficient C++/Python bindings

GLake: optimizing GPU memory management and IO transmission.

Tile primitives for speedy kernels

Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.

PyTorch bindings for CUTLASS grouped GEMM.

xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism

Patch convolution to avoid large GPU memory usage of Conv2D

Code at the speed of thought – Zed is a high-performance, multiplayer code editor from the creators of Atom and Tree-sitter.

GLM-4 series: Open Multilingual Multimodal Chat LMs | 开源多语言多模态对话模型

Triton-based implementation of Sparse Mixture of Experts.

Autoregressive Model Beats Diffusion: 🦙 Llama for Scalable Image Generation

A multi-voice TTS system trained with an emphasis on quality

Zero-Shot Speech Editing and Text-to-Speech in the Wild

Instant voice cloning by MIT and MyShell. Audio foundation model.

MARS5 speech model (TTS) from CAMB.AI

Fixes mojibake and other glitches in Unicode text, after the fact.

Scalable, Low-latency and Hybrid-enabled Vector Search in Postgres. Revolutionize Vector Search, not Database.

Inference and training library for high-quality TTS models.

A generative speech model for daily dialogue.

A framework for serving and evaluating LLM routers - save LLM costs without compromising quality

Understand Human Behavior to Align True Needs

SWE-bench: Can Language Models Resolve Real-world Github Issues?

A massively parallel, optimal functional runtime in Rust

The official repository for the paper Multilingual Mathematical Autoformalization