An optimized quantization and inference library for running LLMs locally on modern consumer-class GPUs

A fast, helpful, and open-source document parser

Extract structured data from documents quickly and accurately.

A cross-platform AI agent orchestrator in C++ that turns any device — including your phone — into a self-contained, multi-agent automation platform

Local PaddleOCR WebUI for PaddleOCR-VL 1.6 and PP-OCRv6, with Docker deployment, on-demand model switching, task history, and aligned visual OCR results.

国产加速卡-海光DCU实战（大模型训练、微调、推理 等）

[ECCV'24] Kalman-Inspired Feature Propagation for Video Face Super-Resolution

EchoSR, an efficient context-harnessing framework that unifies fine-grained detail enhancement and global structure fusion through hierarchical and overlapping context modeling strategies for light…

OmniDocs📄 - One stop visual document processing framework

Ray-powered accelerator for MinerU, turning PDF → Markdown into a scalable, cluster-ready data infrastructure. 基于 Ray 的 MinerU 加速层，将 PDF → Markdown 构建为可扩展、面向集群的数据基础设施。

LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows

INF Tech's open-source MLLMs for SOTA visual-language understanding and advanced document intelligence.

"Deep Learning Crash Course" is a comprehensive and up-to-date guide that takes you from simple neural networks all the way to cutting-edge deep learning architectures-no advanced math and programm…

Emotion-LLaMA: Multimodal Emotion Recognition and Reasoning with Instruction Tuning

Open source audio annotation tool for humans

A machine learning-based video super resolution and frame interpolation framework. Est. Hack the Valley II, 2018.

Model compression toolkit engineered for enhanced usability, comprehensiveness, and efficiency.

DeepSeek-native AI coding agent for your terminal. Engineered around prefix-cache stability — leave it running.

Self-hosted spend firewall and gateway for LLM ( OpenAI / Anthropic / Gemini ). Hard per-user & per-project budget caps that block runaway costs before the API call, plus cost-per-customer tracking…

Cut LLM costs by up to 80% and unlock sub-millisecond responses with intelligent semantic caching.A drop-in, provider-agnostic LLM proxy written in Go with sub-millisecond response

llama.cpp fork optimized for NVIDIA DGX Spark / GB10 (Blackwell, SM 12.1) — TurboQuant weights + KV, NVFP4, DFlash MTP

Fused TBQ4 Flash Attention + MTP + Shared Tensors for llama.cpp — 82+ tok/s with lossless 4.25 bpv KV cache at 200K context on RTX 4090

Run GGUF models easily with a KoboldAI UI. One File. Zero Install.

A Deep Learning-based system that can detect whether any given image is a face-swap deep fake photo, to support the fight against misinformation by differentiating real photographs from edited ones.

Fake Face Photos by Photoshop Experts

Generate a diverse dataset of 100 face-swapped images using the Inswapper model for training robust face-swap detection classifiers. 🖼️🔍

🔥Hierarchical Fine-Grained Image Forgery Detection and Localization (CVPR23 + IJCV24)