Source: https://seerepo.com/r/modelscope/ms-swift/built
Markdown source: https://seerepo.com/r/modelscope/ms-swift/built.md
Description: ms-swift is a Python repo built with FastAPI, Ray, AIOHTTP, and uvicorn. It uses AI through openai, sentencepiece, tiktoken, transformers, peft, and 11 more.

# What is ms-swift built with?

ms-swift is a Python repo built with FastAPI, Ray, AIOHTTP, and uvicorn. It uses AI through openai, sentencepiece, tiktoken, transformers, peft, and 11 more, makes its content with docutils, myst-parser, recommonmark, sphinx-markdown-tables, and pillow, is tested with expecttest and Pytest, deploys with swanlab, and uses oss2. It has 165,674 written lines, with 249 test files.

Read on November 7, 2026, at commit 47b437a.

Repository: https://github.com/modelscope/ms-swift

## Frontend

### Styles

- **sphinx-rtd-theme**

### Interface

- **[matplotlib](https://seerepo.com/uses/matplotlib)**: 2 files

## Backend

### The base

- **[FastAPI](https://seerepo.com/uses/fastapi)**: 2 files

### Data

- **oss2**

### AI

- **[openai](https://seerepo.com/uses/openai)**
- **[sentencepiece](https://seerepo.com/uses/sentencepiece)**
- **[tiktoken](https://seerepo.com/uses/tiktoken)**
- **[transformers](https://seerepo.com/uses/transformers)**: 160 files
- **[peft](https://seerepo.com/uses/peft)**: 26 files
- **[ModelScope](https://seerepo.com/uses/modelscope)**: 21 files
- **[accelerate](https://seerepo.com/uses/accelerate)**: 19 files
- **trl**: 16 files
- **[datasets](https://seerepo.com/uses/datasets)**: 15 files
- **[safetensors](https://seerepo.com/uses/safetensors)**: 7 files
- **[torchvision](https://seerepo.com/uses/torchvision)**: 3 files
- **rouge**: 2 files
- **[torchaudio](https://seerepo.com/uses/torchaudio)**: 2 files
- **[einops](https://seerepo.com/uses/einops)**: 1 file
- **[nltk](https://seerepo.com/uses/nltk)**: 1 file
- **[scikit-learn](https://seerepo.com/uses/scikit-learn)**: 1 file

## Support

### The base

- **Ray**: 13 files
- **[AIOHTTP](https://seerepo.com/uses/aiohttp)**: 4 files
- **[uvicorn](https://seerepo.com/uses/uvicorn)**: 2 files

### Content

- **[docutils](https://seerepo.com/uses/docutils)**
- **[myst-parser](https://seerepo.com/uses/myst-parser)**
- **recommonmark**
- **sphinx-markdown-tables**
- **[pillow](https://seerepo.com/uses/pillow)**: 24 files

### Testing

- **expecttest**
- **Pytest**: 40 files

### Deploy

- **swanlab**: 8 files

### Tools

- **flake8**
- **isort**
- **pre-commit**
- **Sphinx**
- **sphinx-copybutton**
- **sphinxcontrib-mermaid**
- **yapf**
- **tensorboard**: 1 file

### Other

- **addict**
- **charset-normalizer**
- **decorator**
- **importlib-metadata**
- **simplejson**
- **sortedcontainers**
- **zstandard**
- **NumPy**: 40 files
- **tqdm**: 13 files
- **requests**: 7 files
- **pandas**: 5 files
- **pyyaml**: 2 files
- **json-repair**: 1 file
- **SciPy**: 1 file

## Other repos with a similar stack

- [ray-project/ray](https://seerepo.com/r/ray-project/ray/built): FastAPI, Ray, AIOHTTP, uvicorn, matplotlib, sentencepiece, tiktoken, transformers, peft, accelerate, datasets, safetensors, torchvision, torchaudio, einops, nltk, scikit-learn, docutils, myst-parser, pillow · [Compare](https://seerepo.com/compare/modelscope/ms-swift/vs/ray-project/ray)
- [hpcaitech/ColossalAI](https://seerepo.com/r/hpcaitech/ColossalAI/built): FastAPI, Ray, uvicorn, matplotlib, OpenAI, sentencepiece, tiktoken, transformers, peft, ModelScope, accelerate, datasets, safetensors, torchvision, torchaudio, einops, scikit-learn, pillow · [Compare](https://seerepo.com/compare/hpcaitech/ColossalAI/vs/modelscope/ms-swift)
- [OpenBMB/MiniCPM-V](https://seerepo.com/r/OpenBMB/MiniCPM-V/built): FastAPI, AIOHTTP, uvicorn, matplotlib, OpenAI, sentencepiece, tiktoken, transformers, peft, accelerate, torchvision, torchaudio, einops, nltk, scikit-learn, docutils, myst-parser, pillow · [Compare](https://seerepo.com/compare/modelscope/ms-swift/vs/OpenBMB/MiniCPM-V)
- [hiyouga/LlamaFactory](https://seerepo.com/r/hiyouga/LlamaFactory/built): FastAPI, uvicorn, matplotlib, sentencepiece, tiktoken, transformers, peft, ModelScope, accelerate, trl, datasets, safetensors, torchvision, torchaudio, einops, nltk, swanlab · [Compare](https://seerepo.com/compare/hiyouga/LlamaFactory/vs/modelscope/ms-swift)
- [mudler/LocalAI](https://seerepo.com/r/mudler/LocalAI/built): FastAPI, uvicorn, matplotlib, OpenAI, sentencepiece, tiktoken, transformers, peft, ModelScope, accelerate, trl, datasets, safetensors, torchvision, torchaudio, einops, pillow · [Compare](https://seerepo.com/compare/modelscope/ms-swift/vs/mudler/LocalAI)
- [bojieli/ai-agent-book](https://seerepo.com/r/bojieli/ai-agent-book/built): FastAPI, AIOHTTP, uvicorn, matplotlib, OpenAI, sentencepiece, tiktoken, transformers, peft, accelerate, trl, datasets, torchaudio, nltk, scikit-learn, pillow · [Compare](https://seerepo.com/compare/bojieli/ai-agent-book/vs/modelscope/ms-swift)

## Frequently asked questions

### What language is ms-swift written in?

ms-swift is written in Python, with 165,674 written lines.

### What framework does ms-swift use?

ms-swift is built with FastAPI, Ray, AIOHTTP, and uvicorn.

### Does ms-swift use AI?

Yes. ms-swift uses AI through openai, sentencepiece, tiktoken, transformers, peft, ModelScope, accelerate, trl, datasets, safetensors, torchvision, rouge, torchaudio, einops, nltk, and scikit-learn.

### What services does ms-swift use?

ms-swift uses oss2.

### Does ms-swift have tests?

Yes. ms-swift has 249 test files and is tested with expecttest and Pytest.

### Which repos have a stack like ms-swift's?

The ones that share most of its stack: ray-project/ray, hpcaitech/ColossalAI, and OpenBMB/MiniCPM-V.

## What is your repo built with?

Any public repo, with no account: https://seerepo.com/r/owner/repo/built, with the owner and name it has on GitHub.