Important
This repo is archived. ML Intern has now moved to the HuggingChat repo, and you can see the code there.
To use ML Intern in HuggingChat, visit https://hf.co/chat/?mode=ml-intern
Archived — ML Intern is no longer maintained. Continue with HuggingChat.
AI summary: A terminal-based AI agent configured for machine learning operations using Hugging Face endpoints and local models.
PythonApache-2.0Created Oct 30, 2025Last push 29d ago+-3 stars this week+-28 this month
Widely adopted
10,779 stars
Permissive license
Apache-2.0
Continuous integration
Automated checks passing
Repeat trending
3 trending appearances
ML Intern provides a command-line interface that connects to various LLM providers (including Hugging Face routers, OpenAI, and local endpoints like Ollama) to execute machine learning tasks. It handles operations via an internal agentic loop that processes tool calls and manages conversation context up to 170k tokens, automatically compacting history when needed. The runtime supports executing code directly on the local filesystem or securely inside Hugging Face Spaces using a sandbox tool runtime. Furthermore, it automatically uploads session traces to a private Hugging Face dataset for observability, while providing integrations with Slack for out-of-band notifications on required approvals or errors.
This tool is aimed at machine learning engineers and researchers who require an AI assistant deeply integrated with the Hugging Face ecosystem and local hardware. It is ideal for users who want to safely automate model testing and script execution with robust trace visibility.
Developers use ML Intern to run agents entirely offline by configuring local endpoint routing to tools like Ollama or LM Studio.
Machine learning engineers run training scripts safely inside dynamically provisioned Hugging Face Spaces using the sandbox tool runtime.
Teams share agent traces and execution histories by viewing auto-uploaded datasets via the Hugging Face Agent Trace Viewer.
Operators rely on the built-in Slack integration to approve destructive local actions or high-cost jobs asynchronously.
Getting started: ml-intern --model moonshotai/Kimi-K2.7-Code:novita "your prompt"
Important
This repo is archived. ML Intern has now moved to the HuggingChat repo, and you can see the code there.
To use ML Intern in HuggingChat, visit https://hf.co/chat/?mode=ml-intern
abidlabs committed 919ecbb · 29d ago
#368 by abidlabs · updated 1mo ago
abidlabs committed d9a882d · 1mo ago
#219 by Goodnight77 · updated 2mo ago
#43 by kiankyars · updated 2mo ago
#83 by JasonOA888 · updated 2mo ago
#98 by demirciberk · updated 2mo ago
#106 by ZakAnun · updated 2mo ago
#110 by triscacezar-droid · updated 2mo ago
#238 by Suvradippaul · updated 2mo ago
lewtun committed 3555bec · 2mo ago
lewtun committed b85312c · 2mo ago
| Date | List | Rank | Stars gained |
|---|---|---|---|
| Apr 25, 2026 | daily | #11 | +220 |
| Apr 24, 2026 | daily | #4 | +267 |
| Apr 23, 2026 | daily | #5 | +164 |
A collective list of free APIs
500.5K stars · Python
Learn how to design large-scale systems. Prep for the system design interview. Includes Anki flashcards.
378.7K stars · Python
Curated list of project-based tutorials
290.1K stars · Python
The agent that grows with you
265K stars · Python
A feature-rich command-line audio/video downloader
209.4K stars · Python
Public repository for Agent Skills
198.1K stars · Python