p95 Python SDK

Track ML experiments locally. No server setup required.

Install

pip install p95

uv install p95

## Agentic skill
npx skills add numerataz/instrument-experiment

Usage

from p95 import Run

with Run(project="my-project", name="experiment-1") as run:
    run.log_config({"learning_rate": 0.001, "epochs": 10})

    for epoch in range(10):
        loss = train_one_epoch()
        run.log_metrics({"loss": loss}, step=epoch)

View Results

pnf --logdir ~/.p95/logs

Opens a dashboard at http://localhost:6767

API

run.log_metrics({"loss": 0.5, "accuracy": 0.85}, step=epoch)  # Log metrics
run.log_config({"lr": 0.001})                                  # Log config
run.add_tags(["baseline"])                                     # Add tags

Environment Variables

Variable	Description	Default
P95_LOGDIR	Where to save logs	~/.p95/logs
P95_API_KEY	p95 cloud API Key	none
P95_URL	p95 cloud URL	https://api.p.ninetyfive.gg

License

MIT