Aim
Self-hosted, open-source ML training metadata tracker with a powerful exploratory web UI.
Aim
Aim is a self-hosted, open-source experiment tracking tool. It logs training metadata — metrics, hyperparameters, text, images, audio, video — and provides a powerful web UI for exploring thousands of runs simultaneously. Unlike hosted solutions, all data stays on your infrastructure.
Key Features
- Local-first: all tracking data stays in a local
.aimrepo — no account needed - Exploratory UI: query runs with a SQL-like expression language and visualize anything
- Multi-modal logging: metrics, images, audio, video, text, distributions, figures
- Remote tracking server: centralize tracking for a team without SaaS
- Deep integrations: PyTorch Lightning, Keras, XGBoost, Optuna, Hugging Face, Comet
- Python SDK for fine-grained control
Quick Start
from aim import Run
run = Run()
run["hparams"] = {"lr": 0.001, "batch_size": 32}
for step in range(100):
run.track(loss, name="loss", step=step)
run.track(acc, name="accuracy", step=step)
# Launch the UI
aim up
Install via ai-supply
npx ai-supply add aim-training-metadata-ui
Curated mirror of the open-source Aim (Apache-2.0). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/aim-training-metadata-ui/check). Click a policy:
Consume Aim programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/aim-training-metadata-ui
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/aim-training-metadata-ui/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add aim-training-metadata-ui
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/aim-training-metadata-ui/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "aim-training-metadata-ui" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.