GSMA/leaderboard

---
license: apache-2.0
pretty_name: Open Telco Leaderboard Scores
language:
- en
task_categories:
- text-classification
- question-answering
tags:
- telecommunications
- 5g
- llm-evaluation
- benchmark
- leaderboard
configs:
- config_name: default
data_files:
- split: train
path: leaderboard_scores.csv
---

# Open Telco Leaderboard Scores

Benchmark scores for **84 models** across 7 telecom-domain benchmarks, sourced from the MWC leaderboard.

This dataset publishes **scores only** (no energy metrics).

## Files

- `leaderboard_scores.csv`: Flat table for the dataset viewer.
- `leaderboard_scores.json`: Structured JSON with per-model benchmark scores and standard errors.

## Schema (`leaderboard_scores.csv`)

Core columns:

- `model` — Model name
- `provider` — Model provider (e.g. OpenAI, Google, Meta)
- `rank` — Rank by average score (descending)
- `average` — Mean of available benchmark scores
- `benchmarks_completed` — Number of benchmarks with scores

One column per benchmark — each cell contains `[score, stderr]` as a JSON tuple, or empty if not evaluated:

Benchmarks:

- `teleqna` — Telecom Q&A (multiple choice)
- `teletables` — Table understanding
- `oranbench` — O-RAN knowledge
- `srsranbench` — srsRAN knowledge
- `telemath` — Telecom math problems
- `telelogs` — Telecom log analysis
- `three_gpp` — 3GPP specification knowledge

## Usage

```python
from datasets import load_dataset

ds = load_dataset("GSMA/leaderboard", split="train")
print(ds.column_names)
print(ds[0])
```