metadata

title: README
emoji: 🦀
colorFrom: green
colorTo: red
sdk: static
pinned: false
license: apache-2.0

Introduction

ModelScope is built upon the notion of “Model-as-a-Service” (MaaS). It seeks to bring together most advanced machine learning models from the AI community, and streamlines the process of leveraging AI models in real-world applications. We are joining hands with HuggingFace to make AI more accessible for everyone.

To use models in ModelScope, you can pip install modelscope or refer to our github repo https://github.com/modelscope/modelscope

Models and Online Accessibility

Tens of thousands of models are made publicly available on ModelScope, covering the latest development in areas such as NLP, CV, Audio, Multi-modality, and AI for Science, etc. Many of these models represent the SOTA in their specific fields, and made their open-sourced debut on ModelScope. Users can visit ModelScope(modelscope.cn) and experience first-hand how these models perform via online experience, with just a few clicks. Immediate developer-experience is also possible through the ModelScope Notebook, which is backed by ready-to-use CPU/GPU development environment in the cloud - only one click away on ModelScope.

Note: Most models on ModelScope are public and can be downloaded without account registration on (ModelScope website), please refer to instructions for model download, for dowloading models with api provided by modelscope library or git.

Open source repositories

ModelScope has open-sourced numerous libraries available for deep learning, here lists some of them:

The main modelscope library: The main repo of ModelScope, with the capatilities of model inference, model training and model structures.
The SWIFT library: The LLM&MLLM training repo of ModelScope, developers can use this repo to train 500+ LLMs and MLLMs, covering the training stage of PT, SFT, RLFT, RLVR(GRPO).
The DiffSynth library: The SD training repo of ModelScope, developers can train and infer SD models with this repo.
ModelScope-Agent: A single agent framework with basic capatilities of code interpeter, RAG and tool calling.
AgentScope: A multi-agent framework not only the basic capatilities of common agent framework, but also workflow editing and multi-player games.
Funasr:A Fundamental End-to-End Speech Recognition Toolkit and Open Source SOTA Pretrained Models, Supporting Speech Recognition, Voice Activity Detection, Text Post-processing etc.