Local experiment tracking with model versioning - self-hosted W&B alternative with Artifacts
Project description
Runicorn
English | 简体中文
Local, open-source ML experiment tracking. 100% offline, zero telemetry. A modern self-hosted alternative to W&B.
✨ Highlights
| Feature | Description |
|---|---|
| 🏠 100% Local | Your data never leaves your machine |
| 📊 Real-time Visualization | Live metrics, logs, and GPU monitoring |
| 📦 Model Versioning | Git-like Artifacts with deduplication |
| 🌐 Remote Viewer | Access remote GPU servers via SSH (like VSCode Remote) |
| 🖥️ Desktop App | Native Windows app with auto-backend |
🚀 Quick Start
pip install runicorn
runicorn viewer # Open http://127.0.0.1:23300
import runicorn as rn
run = rn.init(project="my_project", name="exp_1")
for epoch in range(100):
loss = train_one_epoch()
run.log({"loss": loss, "epoch": epoch})
run.finish()
📦 Model Versioning
# Save
artifact = rn.Artifact("my-model", type="model")
artifact.add_file("model.pth")
run.log_artifact(artifact) # → v1, v2, v3...
# Load
artifact = run.use_artifact("my-model:latest")
model_path = artifact.download()
🌐 Remote Viewer
Access remote GPU servers without file sync:
runicorn viewer # → Click "Remote" → SSH credentials → Done!
| Old Sync (v0.4) | Remote Viewer (v0.5+) | |
|---|---|---|
| Wait | Minutes~Hours | Seconds |
| Storage | Required | Zero |
| Real-time | ❌ | ✅ |
📚 Documentation
| Resource | Link |
|---|---|
| User Guide | docs/user-guide/ |
| API Reference | docs/api/ |
| Changelog | CHANGELOG.md |
🆕 v0.7.0 (Latest)
- 🌐 Remote Viewer Hardening — Saved connections, health monitoring, reconnect states, and OpenSSH password support
- 🎨 Web UI Productization — Cleaner navigation, better compare flow, ZIP import/export preview, and unified recycle bin
- 📈 Logs & Monitoring — Virtualized logs, stronger dark-mode consistency, and backend-collected GPU telemetry history
- 🔌 Logging Compatibility — Better support for ImageNet meters, TensorBoard, and tensorboardX
- 🖥️ Desktop Improvements — Native remote-session windows in the current desktop workflow
License
MIT — see LICENSE
Version: v0.7.0 | Last Updated: 2026-03-15
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file runicorn-0.7.0.tar.gz.
File metadata
- Download URL: runicorn-0.7.0.tar.gz
- Upload date:
- Size: 1.6 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.5
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
68e622ab6196c2ffbaf4f7318c7c9e0f5f80f194da428d275f63d0ff21918e45
|
|
| MD5 |
d3a7cd2152ca5a4ac6996754a399db4d
|
|
| BLAKE2b-256 |
547ba8993f469facccb23b71bbf5595a745efe8c38e80f6b35f45d798d2c58f7
|
File details
Details for the file runicorn-0.7.0-py3-none-any.whl.
File metadata
- Download URL: runicorn-0.7.0-py3-none-any.whl
- Upload date:
- Size: 1.7 MB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.5
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
4461397c5ae6a81e38482042edcda63dfec30d6baccdbbd4d585660c3e9bc091
|
|
| MD5 |
218235ab1f07e3d23c10d60405492ce7
|
|
| BLAKE2b-256 |
5d8a95ecc9fdc56e5c79aec0cd300c62ef4f3a80a678747fd5b1a69e6c2753f6
|