README.md

---
title: Time Series Compression Benchmark
emoji: 📊
colorFrom: green
colorTo: gray
sdk: static
app_file: index.html
pinned: false
short_description: Compression rates and encoding/decoding costs
---

# THULab Time Series Compression Benchmark

Interactive, static leaderboard of the existing `web_compression` project results, inspired by the grouped leaderboard interface of [GIFT-Eval](https://huggingface.co/spaces/Salesforce/GIFT-Eval).

The Space serves published result snapshots. It does not execute compression experiments or accept raw dataset uploads. It has no runtime dependency on the project's Django server.

## Update the results

From the source project checkout, regenerate the snapshots with:

```bash
python3 deploy/huggingface-compressionbenchmark/export_results.py
python3 deploy/huggingface-compressionbenchmark/build_ui_payload.py
```

Publish `README.md`, `index.html`, `style.css`, `app.js`, `data.json`, `ui-data.json`, `manifest.json`, `export_results.py`, and `build_ui_payload.py` to the static Space. The manifest records the exact source-file identities and metric boundaries. The compact browser payload retains the full snapshot's measurements and SHA-256 identity.

Only measurements in the authoritative current project scope are exported. Missing metrics and missing coverage are displayed explicitly. Existing results do not imply independently verified round trips or uniform hardware environments unless those checks are recorded in the source.

## Interface

- Overall and per-dataset leaderboards with runtime, data-type, precision, family, and method filters
- Sortable metrics, optional columns, and CSV download of the selected ranking
- Compression rate versus encoding/decoding cost plots
- Dataset-level method details, snapshot provenance, and metric definitions

The default family view selects the best measured compression-rate configuration independently for each dataset. Switch to **Exact configuration** to hold a configuration fixed. Runtime results stay separate. The paper snapshot tab preserves the frozen source table and is separate from the current report aggregation.

Source project: https://github.com/xjz17/web_compression