README.md

---
language:
  - en
license: other
license_name: 3gpp
license_link: https://www.3gpp.org/specifications-technologies/legal-matters
tags:
  - telecommunications
  - 3gpp
  - 5g
  - nr
  - lte
  - standards
  - specifications
pretty_name: 3GPP Specifications
configs:
  - config_name: rel-18
    default: true
    data_files:
      - split: 21_series
        path: data/rel-18/21_series-*.parquet
      - split: 22_series
        path: data/rel-18/22_series-*.parquet
      - split: 23_series
        path: data/rel-18/23_series-*.parquet
      - split: 24_series
        path: data/rel-18/24_series-*.parquet
      - split: 26_series
        path: data/rel-18/26_series-*.parquet
      - split: 27_series
        path: data/rel-18/27_series-*.parquet
      - split: 28_series
        path: data/rel-18/28_series-*.parquet
      - split: 29_series
        path: data/rel-18/29_series-*.parquet
      - split: 31_series
        path: data/rel-18/31_series-*.parquet
      - split: 32_series
        path: data/rel-18/32_series-*.parquet
      - split: 33_series
        path: data/rel-18/33_series-*.parquet
      - split: 36_series
        path: data/rel-18/36_series-*.parquet
      - split: 37_series
        path: data/rel-18/37_series-*.parquet
      - split: 38_series
        path: data/rel-18/38_series-*.parquet
  - config_name: rel-19
    data_files:
      - split: 21_series
        path: data/rel-19/21_series-*.parquet
      - split: 22_series
        path: data/rel-19/22_series-*.parquet
      - split: 23_series
        path: data/rel-19/23_series-*.parquet
      - split: 24_series
        path: data/rel-19/24_series-*.parquet
      - split: 26_series
        path: data/rel-19/26_series-*.parquet
      - split: 27_series
        path: data/rel-19/27_series-*.parquet
      - split: 28_series
        path: data/rel-19/28_series-*.parquet
      - split: 29_series
        path: data/rel-19/29_series-*.parquet
      - split: 31_series
        path: data/rel-19/31_series-*.parquet
      - split: 32_series
        path: data/rel-19/32_series-*.parquet
      - split: 33_series
        path: data/rel-19/33_series-*.parquet
      - split: 34_series
        path: data/rel-19/34_series-*.parquet
      - split: 35_series
        path: data/rel-19/35_series-*.parquet
      - split: 36_series
        path: data/rel-19/36_series-*.parquet
      - split: 37_series
        path: data/rel-19/37_series-*.parquet
      - split: 38_series
        path: data/rel-19/38_series-*.parquet
---

# 3GPP Specifications

938 3GPP technical specifications and reports decomposed into self-contained
sections with tables and figures inlined. Pre-validation dataset.

Releases: Rel-18, Rel-19

## Usage

```python
from datasets import load_dataset

# Load Rel-18 NR radio specs
ds = load_dataset("GSMA/3GPP", "rel-18", split="38_series")

# Get a specific section
section = ds.filter(lambda r: r["spec_id"] == "38331" and r["clause"] == "5.2.1")[0]
print(section["body"])  # self-contained: tables and figures are inlined

# Reconstruct full spec
spec = ds.filter(lambda r: r["spec_id"] == "38331").sort("document_order")
full_text = "\n\n".join(spec["body"])

# Load Rel-19
ds19 = load_dataset("GSMA/3GPP", "rel-19", split="38_series")
```

## Schema

| Column | Type | Description |
|--------|------|-------------|
| `spec_id` | string | Spec identifier, e.g. `38331` |
| `spec_number` | string | Dotted form, e.g. `38.331` |
| `spec_type` | string | `TS` or `TR` |
| `title` | string | Full spec title |
| `release` | string | `Rel-18` or `Rel-19` |
| `clause` | string | Section number, e.g. `5.2.1` |
| `section_title` | string | Section heading |
| `parent_clause` | string | Parent section, e.g. `5.2` |
| `depth` | int32 | Heading depth (1-6) |
| `body` | string | Self-contained markdown (tables/figures inlined) |
| `body_chars` | int32 | Character count |
| `document_order` | int32 | Position within spec |
| `images` | list[Image] | Images referenced in this section |
| `image_hashes` | list[string] | MD5 hashes of images |

## Original DOCX Files

The original 3GPP Word documents (549 files for Rel-18, 442 files for Rel-19) are available under `original/`:

```
original/
  rel-18/
  rel-19/
```

Filenames follow the 3GPP convention: `{spec_id}-{version}[_{part}].docx`.
Large specs may be split across multiple files (e.g., `38101-1-j10_cover.docx`,
`38101-1-j10_s00-0504.docx`).

### Download a single file

```python
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    "GSMA/3GPP",
    "original/rel-18/38331-i00.docx",
    repo_type="dataset",
)
```

## License

See [3GPP Legal Matters](https://www.3gpp.org/specifications-technologies/legal-matters).