What is DeepSpeed built with?
DeepSpeed is a Python repo. It uses AI through accelerate, sentencepiece, torch, transformers, triton, and 8 more, makes its content with docutils, validates its data with Pydantic, is tested with pytest-forked, pytest-xdist, and Pytest, and deploys with psutil. It has 247,033 written lines, with 395 test files.
Read on November 6, 2026, at commit f47dc3e. Read it as markdown
Frontend
Styles 1
- sphinx-rtd-theme
Backend
AI 13
- torch397 files
diffusers3 files
safetensors3 files
lm-eval1 file
torchvision1 file
accelerate
- transformers28 files
- triton11 files
- einops4 files
- comet-ml2 files
- liger-kernel1 file
- xgboost1 file
1 more
Other repos with a similar stack
Ordered by how much of its stack they share with DeepSpeed, then by stars.
- hiyougaLlamaFactory75.9k
Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)
9 of 38 in commonCompare →
Frequently asked questions
What language is DeepSpeed written in?
DeepSpeed is written in Python, with 247,033 written lines.
Does DeepSpeed use AI?
Yes. DeepSpeed uses AI through accelerate, sentencepiece, torch, transformers, triton, einops, diffusers, safetensors, comet-ml, liger-kernel, lm-eval, torchvision, and xgboost.
Does DeepSpeed have tests?
Yes. DeepSpeed has 395 test files and is tested with pytest-forked, pytest-xdist, and Pytest.
Which repos have a stack like DeepSpeed's?
The ones that share most of its stack:ray-project/rayvllm-project/vllmunslothai/unsloth
See inside your code, today.
Paste your repo and see it drawn, part by part: a medium one in under a minute.
Free, no account.
Is it your repo? Add its badge, keep it up to date or take it down →

