Datasetpublic

BuRabea/v2v-autonomous-driving-qa

V2V & Autonomous Driving QA Dataset This dataset contains processed research papers focused on V2V communication, autonomous driving, sensor fusion, ADAS, and related fields.It provides structured text sections and automatically generated question–answer (QA) pairs, making it useful for fine-tuning language models, evaluation, or RAG pipelines. 📊 Dataset Summary ~22,457 papers processed Sections extracted: abstract, introduction, conclusion 55k+ QA pairs… See the full description on the dataset page: https://huggingface.co/datasets/BuRabea/v2v-autonomous-driving-qa.

sourceHugging Facemitupdated 1y agoView on Hugging Face

0likes206downloads

Dataset Card

V2V & Autonomous Driving QA Dataset

This dataset contains processed research papers focused on V2V communication, autonomous driving, sensor fusion, ADAS, and related fields. It provides structured text sections and automatically generated question–answer (QA) pairs, making it useful for fine-tuning language models, evaluation, or RAG pipelines.

📊 Dataset Summary

—~22,457 papers processed
—Sections extracted: abstract, introduction, conclusion
—55k+ QA pairs generated using predefined templates

📂 Dataset Structure

Each entry in the dataset includes:

Field	Description
`instruction`	The prompt/question template
`input`	Context text used for generating the QA pair
`output`	The answer to the question
`source_paper`	Name or source of the paper
`paper_id`	Unique identifier (arXiv ID or local)
`source_section`	Section where the QA was generated (abstract, intro, conclusion, etc.)
`source_file`	Original file name used during preprocessing

🔎 Example Entry

json

{
  "paper_id": "2305.12345",
  "instruction": "What is the main contribution of the paper?",
  "input": "This paper introduces a new protocol for V2V communication that improves reliability …",
  "output": "The paper proposes a new V2V communication protocol that improves reliability in high-interference environments.",
  "source_paper": "A Novel V2V Communication Protocol for Cooperative Driving",
  "source_section": "abstract",
  "source_file": "2305.12345.pdf"
}

🚀 Usage

Load with the Hugging Face datasets library:

python

from datasets import load_dataset  

dataset = load_dataset("BuRabea/v2v-autonomous-driving-qa")  
print(dataset["train"][0])

✅ Intended Uses

—Fine-tuning LLMs in autonomous driving / V2V domains
—Question answering or summarization over technical papers
—RAG systems for domain-specific retrieval
—Evaluation of models on extracting knowledge from research papers

⚠️ Limitations

—Extracted sections may be incomplete (e.g., only first 10 pages of some PDFs)
—Equations, tables, and scanned PDFs may be missing or truncated
—QA pairs are template-based → not exhaustive or fully human-verified

📜 License & Citation

License: MIT (for dataset processing). Original papers remain under their respective copyrights.

bibtex

@dataset{burabea_v2v_autonomous_driving_qa_2025,  
  author       = {BuRabea},  
  title        = {V2V & Autonomous Driving QA Dataset},  
  year         = {2025},  
  url          = {https://huggingface.co/datasets/BuRabea/v2v-autonomous-driving-qa}  
}

📬 Contact

For issues, feedback, or contributions, open a discussion in this repo or contact BuRabea.