ECG hybrid · 4×

Reproduce the ecg-hybrid-4x configuration. Prepare its dataset before running the workflow; use the linked configuration to inspect the exact settings.

Overview

• Modality: ECG
• Structure: codec
• Target compression ratio: 4×
• Sample rate: 256 Hz
• Config: Operating point declared in the registry; no training config.
• Run name: ecg_hybrid_256hz_04x_golden
• Hugging Face target: Ambiq/compressionkit-ecg-hybrid-4x-v1.0

Dataset & License

• Dataset: PTB-XL (dataset_id: ptb-xl)
• License: CC BY 4.0 (open)
• Notes: Auto-downloaded on first use.

The lifecycle runner pre-flights dataset availability before training (see the dataset contract).

Reproduction

# Single command, end-to-end.
uv run compressionkit golden run ecg-hybrid-4x

Results land under results/ecg_hybrid_256hz_04x_golden/; deploy artifacts under results/ecg_hybrid_256hz_04x_golden/deploy/.

Evaluation Metrics

See the modality model zoo for the full metrics table:

• ECG models

Evaluation writes quality_scorecard.json and summary.json under its results/<run>/.

Deploy Artifacts

The deploy package contains the following artifacts, depending on the export configuration:

• spiht_config.json / spiht_app_config.h — codec parameters (language-neutral + C header).
• c_sources/spiht.[ch] — portable C99 SPIHT reference.
• sample_stimulus.npz / reference_vectors.npz — license-safe test frames and known-good encode/decode vectors.
• model_card.json, deploy_manifest.json — metadata.
• denoiser_gain_model.tflite / .h — INT8 wavelet-gain denoiser (embeddable, LiteRT).
• denoiser_gain_model.keras, hybrid_manifest.json — float32 Python reference denoiser and pipeline stage order.

Customization Notes

• This operating point is declared directly in compressionkit/experiments/registry.py (no training YAML) — add a new registry entry to explore a neighbouring operating point.
• To resume publishing without retraining, pass --skip-train to compressionkit golden run ecg-hybrid-4x --publish.