Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,67 @@
+---
+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 (3072d, 100k)
+
+Generated embeddings dataset for vector database training and evaluation.
+
+## Dataset Summary
+
+This dataset contains 100,000 text samples with vector embeddings (3072 dimensions)
+generated from the BioASQ-BEIR dataset using Qwen/Qwen3-Embedding-8B.
+
+## Dataset Structure
+- **Base dataset**: 100,000 samples with embeddings
+- **Embedding dimension**: 3072
+
+## Repository Structure
+
+### parquet/
+- `base.parquet` - Main dataset with text and embeddings
+
+## Usage
+
+### Loading with HuggingFace Datasets
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("maknee/bioasq_bier_3072_100k")
+base_data = dataset['train']
+
+import numpy as np
+embeddings = np.array(base_data['embedding'])
+texts = base_data['text']
+```
+
+## Dataset Information
+
+- **Source**: BioASQ-BEIR
+- **Size**: 100,000 samples
+- **Dimension**: 3072
+- **Format**: Parquet
+
+## Citation
+
+```bibtex
+@dataset{huggingface_embeddings_maknee_bioasq_bier_3072_100k,
+  title={BioASQ-BEIR Vector Database Embeddings Dataset},
+  author={Henry Zhu},
+  year={2026},
+  url={https://huggingface.co/datasets/maknee/bioasq_bier_3072_100k}
+}
+```
+
+## License
+
+MIT License.