File size: 1,111 Bytes
863346f
b910aca
 
 
 
863346f
9fe0361
b910aca
863346f
 
b910aca
863346f
 
b910aca
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
---
title: CommonLID Leaderboard
emoji: 🌍
colorFrom: indigo
colorTo: pink
sdk: gradio
sdk_version: 5.9.1
python_version: 3.12
app_file: app.py
pinned: false
license: apache-2.0
---

# CommonLID Leaderboard

Live results table for the [CommonLID](https://huggingface.co/datasets/commoncrawl/CommonLID)
benchmark and its `commonlid_nano` slice. Models are ranked by **macro
F1**; click any row for per-language detail.

## How the data flows

1. Each run writes `<dataset>/<model>/{summary.json, predictions.jsonl}` to
   `data/results/` via `commonlid run …`.
2. The contents of `data/results/` are uploaded to
   [`commoncrawl/commonlid-results`](https://huggingface.co/datasets/commoncrawl/commonlid-results)
   with `huggingface-cli upload`.
3. This Space loads the dataset on boot and renders one tab per benchmark.

## Local preview

```bash
uv sync --extra leaderboard
uv run commonlid leaderboard serve --local-dir ./data/results
```

## Pinning a specific results revision

Set `COMMONLID_RESULTS_REVISION` (or `--revision` on the CLI) to lock the
Space to a specific commit on the results dataset.