Stars
AGILE: Lightweight and Efficient Asynchronous GPU-SSD Integration (SC25)
Enhancing CUDA Intra-Streaming-Multiprocessor Parallelism for Large Language Models via Fine-Grained Task Graph
An Activation Offloading Framework to SSDs for Faster Large Language Model Training
Evaluation harness and norm-direction method for KV cache compression. Cross-model worst-case quality metrics.
DGX Spark / GB10 vLLM image for Gemma 4 31B Deckard Heretic Uncensored NVFP4 with z-lab DFlash speculative decoding.
Official Repository for ICML 2026 paper Coevolutionary Continuous Discrete Diffusion: Make Your Diffusion Language Model a Latent Reasoner
A Minimal and Elegant Framework for Real-Time Interactive World Models
VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo
Experimenting with a visual representation of Wikipedia
Open diffusion language model for code generation — releasing pretraining, evaluation, inference, and checkpoints.
Free and open-source chat SDK. Build fast, real-time apps and generative AI agents with a high-performance, customizable, cross-platform UI.
Modelence is a full-stack framework for building production web apps with a built-in database, authentication and monitoring. Modelence is opinionated and AI agent-first, which means it's optimized…
implementing minimal versions of joint-embedding predictive architecture (JEPA)
Zero-shot expressive voice cloning and speech generation. Generate anything from short clips to full-length audiobooks with realistic emotional delivery, pacing, and breath control. Clone any voice…
🔥 Search, scrape, and clean the web for AI agents.
This is an example flask backend to interface with a custom version of the Huggingface Chatui
am17an / llama.cpp
Forked from ggml-org/llama.cppLLM inference in C/C++
DeepSeek 4 Flash local inference engine for Metal and CUDA
API client for AUTOMATIC111/stable-diffusion-webui for nodejs/browser
Windsurf-to-OpenAI compatible API proxy
TTS voice notifications for Claude Code — hear when Claude finishes or needs your input
[ICLR 2026] TS-Attn: Temporal-wise Separable Attention for Multi-Event Video Generation
Official implementation of Paper "System-Aware 4-Bit KV-Cache Quantization for Real-World LLM Serving"