AI-Driven PPT Generation Tool.This project is based on the Nano Banana Pro API. It can automatically generate well-structured and neatly formatted PPTs according to users' ideas, outlines, or documents (like PDF, Markdown), and supports adjusting content through conversations.

Memory System for AI Programming Tools.This project is a memory system specifically designed for AI programming tools. It employs Git and JSON as persistent storage, offering long-term and structured memory for AI programming agents to tackle issues such as context loss when dealing with long-cycle and complex programming tasks.

Official Python Client for HF.This project is the officially open-source Python client of the Hugging Face platform, providing functions like downloading, uploading, and managing models, datasets, and Spaces

from huggingface_hub import hf_hub_download

hf_hub_download(
    repo_id="deepseek-ai/DeepSeek-V3.2",
    filename="config.json"
)

Illustrating the Principles of Large Model Technology.This project contains over 100 schematic diagrams of large model technology principles, systematically introducing large models and reinforcement learning, covering LLM/VLM large model architectures, training algorithms (RL, RLHF, GRPO, DPO, SFT, CoT), performance optimization, and RAG, etc.

Automated Evaluation Framework for RAG Applications.This is an open-source framework specifically designed for evaluating and optimizing RAG applications, which provides objective and quantifiable evaluation results and supports the automatic generation of test datasets.

import asyncio
from ragas.metrics.collections import AspectCritic
from ragas.llms import llm_factory

# Setup your LLM
llm = llm_factory("gpt-4o")

# Create a metric
metric = AspectCritic(
    name="summary_accuracy",
    definition="Verify if the summary is accurate and captures key information.",
    llm=llm
)

# Evaluate
test_data = {
    "user_input": "summarise given text\nThe company reported an 8% rise in Q3 2024, driven by strong performance in the Asian market. Sales in this region have significantly contributed to the overall growth. Analysts attribute this success to strategic marketing and product localization. The positive trend in the Asian market is expected to continue into the next quarter.",
    "response": "The company experienced an 8% increase in Q3 2024, largely due to effective marketing strategies and product adaptation, with expectations of continued growth in the coming quarter.",
}

score = await metric.ascore(
    user_input=test_data["user_input"],
    response=test_data["response"]
)
print(f"Score: {score.value}")
print(f"Reason: {score.reason}")

Out-of-the-box Intelligent Agent Memory Engine.This is a Python project specifically designed to provide memory functionality for AI agents. It integrates technologies such as graph databases, knowledge graphs, and vector databases. With just 5 lines of code, it can easily provide persistent and multimodal memory for AI agents, supporting the connection and retrieval of past conversations, documents, image and audio transcripts, etc.

import cognee
import asyncio


async def main():
    # Add text to cognee
    await cognee.add("Natural language processing (NLP) is an interdisciplinary subfield of computer science and information retrieval.")

    # Generate the knowledge graph
    await cognee.cognify()

    # Query the knowledge graph
    results = await cognee.search("Tell me about NLP")

    # Display the results
    for result in results:
        print(result)


if __name__ == '__main__':
    asyncio.run(main())

Control Your Phone with Natural Language.This is a mobile automation framework based on LLM Agents. It allows you to control Android devices or emulators through natural language. It supports mainstream large models such as DeepSeek, OpenAI, and Gemini. When using it, you need to install DroidRun Portal on your phone to collect UI information and execute operation commands. Then, the information is passed to the DroidRun framework on the computer through ADB. The framework interacts with the LLM and gives execution instructions.

Handy AI Desktop Assistant.This is a context-aware desktop AI assistant that can automatically obtain and understand the content on the current screen without the need for screenshotting, copying content or switching applications. Just one click can summon the AI to perform tasks such as answering questions, translation and providing answers to doubts.

Local-first AI Note and Meeting Assistant.This is a locally runnable AI intelligent note and meeting recording application. By integrating Ollama, it can complete voice transcription to summary generation locally. It supports functions such as meeting recording, real-time transcription, note organization and intelligent summary.

Let AI Draw Architecture Diagrams for You.This is a Web application built based on Next.js that integrates AI and draw.io chart drawing capabilities. Now you can directly generate, edit, and optimize flowcharts and architecture diagrams through conversations. It supports flowing effect connections, screenshot replication, and historical version functions.