A Flexible and Powerful Parameter Server for large-scale machine learning

A unified, downstream-grounded benchmark for LLM-driven training data construction, selection, and quality evaluation.

All-in-one intelligent assistant powered by LlamaIndex — RAG, GraphRAG, NL2SQL, Skills & Memory with multimodal support.

Dataflow-LoopAI is an intelligent system with self-optimization capabilities that automatically detects and evaluates generation deficiencies in LLMs within specific domains. Through dialog-based a…

DataFlow Knowledge Graph -- Knowledge graph data preparation with DataFlow style operators and pipelines

[ECCV 2026] A diffusion-based framework for document OCR that replaces autoregressive decoding with block-level parallel diffusion decoding.

A flexible framework for orchestrating deep learning models with Ray . It dynamically schedules and serves multiple models — from NLP (e.g., FastText) to CV (e.g., YOLO, SAM) — enabling scalable, d…

Automated system for LLM evaluation via agents. Doc as below:

Unified Codebase for Advanced World Models.

An Open Source implementation of Notebook LM.

The First Unified Agent Data Synthesis Framework for Custom Agentic Task with all-in-one envrionment

A Python package for interacting with the MinerU Vision-Language Model.

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。

Open-source implementation of AI-powered academic writing workspace inspired by OpenAI Prism, featuring LaTeX editing, PDF preview, and intelligent AI assistance

An NL2Pipeline Harness for building AI-ready data workflows.

Connect people, agents, and teams for the next era of human-AI collaboration.

Easy Data Preparation with latest LLMs-based Operators and Pipelines.

Dataflow-MM, multi-media operators for Dataflow. We aim to prepare data for Multimodal Large Language Models.

A high-performance distributed deep learning system targeting large-scale and automated distributed training.

Turn paper/text/topic into editable research figures, technical route diagrams, and presentation slides.

Self-evolving memory OS for LLM & AI Agents: ultra-persistent memory, hybrid-retrieval, and cross-task skill reuse, with 35.24% token savings

Data-centric LLM training with dynamic sample selection, domain mixture optimization, and example reweighting inside the LLaMA-Factory training loop.

AI Database for unified, scalable SQL + vector data management, search and analytics

Easy Data Preparation with latest LLMs-based Operators and Pipelines.

Official repository of RARE: Retrieval-Augmented Reasoning Modeling [KDD 2026 Research Track]

A comprehensive guide for beginners in the field of data management and artificial intelligence.