maknee/bioasq_bier_2048_1m

---
license: mit
task_categories:
- feature-extraction
- text-retrieval
language:
- en
configs:
- config_name: default
data_files:
- split: train
path: "parquet/base.parquet"
---

# BioASQ-BEIR Vector Database Dataset (2048d, 1M)

Generated embeddings dataset for vector database training and evaluation.

## Dataset Summary

This dataset contains 1,000,000 text samples with vector embeddings (2048 dimensions)
generated from the BioASQ-BEIR dataset using Qwen/Qwen3-Embedding-8B.

## Dataset Structure
- **Base dataset**: 1,000,000 samples with embeddings
- **Embedding dimension**: 2048

## Repository Structure

### parquet/
- `base.parquet` - Main dataset with text and embeddings

## Usage

```python
from datasets import load_dataset

dataset = load_dataset("maknee/bioasq_bier_2048_1m")
base_data = dataset["train"]

import numpy as np
embeddings = np.array(base_data["embedding"])
texts = base_data["text"]
```

## Dataset Information

- **Source**: BioASQ-BEIR
- **Size**: 1,000,000 samples
- **Dimension**: 2048
- **Format**: Parquet

## Citation

```bibtex
@dataset{huggingface_embeddings_maknee_bioasq_bier_2048_1m,
title={BioASQ-BEIR Vector Database Embeddings Dataset},
author={Henry Zhu},
year={2026},
url={https://huggingface.co/datasets/maknee/bioasq_bier_2048_1m}
}
```

## License

MIT License.