---
license: apache-2.0
task_categories:
  - text-generation
language:
  - en
tags:
  - telecommunications
  - telecom
  - 3gpp
  - oran
  - gsma
  - rag
  - instruction-tuning
  - domain-specific
  - otel
size_categories:
  - 100K<n<1M
pretty_name: OTel-LLM
dataset_info:
  features:
    - name: anchor
      dtype: string
    - name: prompt
      dtype: string
    - name: completion
      dtype: string
    - name: prompt_type
      dtype: string
    - name: abstention
      dtype: bool
    - name: n_positive_chunks
      dtype: int64
    - name: n_negative_chunks
      dtype: int64
    - name: prompt_token_count
      dtype: int64
    - name: completion_token_count
      dtype: int64
---

# OTel-LLM

## Dataset Summary

OTel-LLM is an instruction-tuning dataset for training large language models to operate within end-to-end retrieval-augmented generation (RAG) pipelines in the telecommunications domain. It is part of the Open Telco (OTel) AI project, the largest open-source AI initiative in telecommunications, curated by over 100 domain experts from industry and academia.

The dataset pairs retrieved context passages with instruction-style prompts and reference completions, covering a broad range of telecom topics including 3GPP specifications, O-RAN architecture, GSMA standards, eSIM, roaming, network security, and related areas. The training data includes both standard answering examples and abstention examples, where the model is trained to refuse answering when the retrieved context is insufficient.

## Dataset Structure

Each sample contains the following fields:

| Field | Type | Description |
|---|---|---|
| `anchor` | string | The source passage or retrieved context chunk used to ground the generation. |
| `prompt` | string | The instruction-formatted input provided to the model, including the context and query. |
| `completion` | string | The reference output the model is trained to generate. |
| `prompt_type` | string | A categorical label indicating the type of instruction (e.g., question answering, summarization). |
| `abstention` | bool | Whether the sample is an abstention example, where the model should refuse to answer due to insufficient context. |
| `n_positive_chunks` | int64 | The number of context chunks in the prompt that are relevant to the query. |
| `n_negative_chunks` | int64 | The number of context chunks in the prompt that are irrelevant to the query (distractors). |
| `prompt_token_count` | int64 | Token count of the prompt. |
| `completion_token_count` | int64 | Token count of the completion. |

## Data Sources and Contributor Attribution

The OTel corpus was curated by 100+ domain experts from industry and academia. Each source class was contributed by a specific institutional partner:

| Source | Contributor | Raw samples |
|---|---|---:|
| arXiv telecom papers, 3GPP standards, telecom Wikipedia, telecom Common Crawl pages | Yale University | 681,172 |
| GSMA Permanent Reference Documents (PRDs), Discover portal, mixed telecom documents | GSMA | 158,006 |
| IETF RFC series | NetoAI | 100,751 |
| Industry whitepapers | Khalifa University | 62,000 |
| O-RAN specifications across working groups 1, 2, 4, 5, 6, 7, 8, 9, 10 | University of Leeds | 58,565 |
| O-RAN documents across working groups | The University of Texas at Dallas | 42,000 |
| **Total raw samples** | | **~1,102,494** |

After the four-stage cleaning pipeline (described below), the corpus was reduced to **326,767** higher-confidence examples across the OTel dataset family.

## Data Collection and Processing

The raw corpus comprised approximately 1.1 million samples. A four-stage cleaning pipeline reduced this to 326K samples across the OTel dataset family:

1. **Heuristic filtering**: Rule-based removal of malformed, duplicated, or low-quality entries.
2. **Reranking filtering**: Relevance scoring with a cross-encoder reranker to discard weakly aligned prompt-completion pairs.
3. **Embedding filtering**: Semantic quality assessment with a strong embedding model to remove outliers and near-duplicates in embedding space.
4. **Deduplication**: Final deduplication pass to ensure sample uniqueness.

## Intended Use

OTel-LLM is intended for supervised fine-tuning of large language models for telecom-domain RAG pipelines. Target applications include:

- Domain-adapted instruction-following models for telecom question answering.
- RAG systems that require grounded, context-faithful generation over telecom standards.
- Models that must learn to abstain from answering when retrieved context does not support a reliable response.

## Limitations

- **Domain scope**: The dataset covers telecommunications standards and specifications. It is not designed for general-purpose instruction tuning.
- **Language**: English only.
- **Temporal coverage**: The source documents reflect standards and specifications available at the time of collection. Newer revisions of 3GPP, GSMA, or O-RAN documents may not be represented.
- **Abstention heuristics**: Abstention labels are derived from the data generation pipeline and may not perfectly capture all cases where a model should refuse to answer.

## Resources

- [Project Page](https://huggingface.co/att)
- [OTel code](https://github.com/farbodtavakkoli/OTel)
- [OTel Selected Coverage](https://github.com/farbodtavakkoli/OTel/blob/main/docs/OTel-selected-coverage.md)
- [OTel Paper](https://arxiv.org/pdf/2610.07766)
- [OTel 1.0 Coverage](https://github.com/farbodtavakkoli/OTel/blob/main/docs/OTel-1.0-coverage.md)
- [OTel 2.0 Coverage](https://github.com/farbodtavakkoli/OTel/blob/main/docs/OTel-2.0-coverage.md)
- [OTel Docker](https://hub.docker.com/u/farbodatdocker)

## Citation

```bibtex
@inproceedings{tavakkoli2026otel,
  title     = {OTel: Open Telco AI Datasets, Benchmarks, and Models},
  author    = {Tavakkoli, Farbod and others},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS), Evaluations and Datasets Track},
  year      = {2026},
  url       = {https://github.com/farbodtavakkoli/OTel}
}
```