Hugging Face Trackio

1---
2name: hugging-face-trackio
3description: Track and visualize ML training experiments with Trackio. Use when logging metrics during training (Python API) or retrieving/analyzing logged metrics (CLI). Supports real-time dashboard visualization, HF Space syncing, and JSON output for automation.
4---
5 
6# Trackio - Experiment Tracking for ML Training
7 
8Trackio is an experiment tracking library for logging and visualizing ML training metrics. It syncs to Hugging Face Spaces for real-time monitoring dashboards.
9 
10## Two Interfaces
11 
12| Task | Interface | Reference |
13|------|-----------|-----------|
14| **Logging metrics** during training | Python API | [references/logging_metrics.md](references/logging_metrics.md) |
15| **Retrieving metrics** after/during training | CLI | [references/retrieving_metrics.md](references/retrieving_metrics.md) |
16 
17## When to Use Each
18 
19### Python API → Logging
20 
21Use `import trackio` in your training scripts to log metrics:
22 
23- Initialize tracking with `trackio.init()`
24- Log metrics with `trackio.log()` or use TRL's `report_to="trackio"`
25- Finalize with `trackio.finish()`
26 
27**Key concept**: For remote/cloud training, pass `space_id` — metrics sync to a Space dashboard so they persist after the instance terminates.
28 
29→ See [references/logging_metrics.md](references/logging_metrics.md) for setup, TRL integration, and configuration options.
30 
31### CLI → Retrieving
32 
33Use the `trackio` command to query logged metrics:
34 
35- `trackio list projects/runs/metrics` — discover what's available
36- `trackio get project/run/metric` — retrieve summaries and values
37- `trackio show` — launch the dashboard
38- `trackio sync` — sync to HF Space
39 
40**Key concept**: Add `--json` for programmatic output suitable for automation and LLM agents.
41 
42→ See [references/retrieving_metrics.md](references/retrieving_metrics.md) for all commands, workflows, and JSON output formats.
43 
44## Minimal Logging Setup
45 
46```python
47import trackio
48 
49trackio.init(project="my-project", space_id="username/trackio")
50trackio.log({"loss": 0.1, "accuracy": 0.9})
51trackio.log({"loss": 0.09, "accuracy": 0.91})
52trackio.finish()
53```
54 
55### Minimal Retrieval
56 
57```bash
58trackio list projects --json
59trackio get metric --project my-project --run my-run --metric loss --json
60```
61