---
language:
- en
pretty_name: "quickb"
tags:
- quickb
- text-chunking
- 1K<n<10K
task_categories:
- text-generation
- text-retrieval
task_ids:
- document-retrieval 
library_name: quickb
---

# quickb

Generated using [QuicKB](https://github.com/AdamLucek/quickb), a tool developed by [Adam Lucek](https://huggingface.co/AdamLucek).

QuicKB optimizes document retrieval by creating fine-tuned knowledge bases through an end-to-end pipeline that handles document chunking, training data generation, and embedding model optimization.

### Chunking Configuration
- **Chunker**: RecursiveTokenChunker
- **Parameters**:
  - **chunk_size**: `400`
  - **chunk_overlap**: `50`
  - **length_type**: `'character'`
  - **separators**: `['\n\n', '\n', '.', '?', '!', ' ', '']`
  - **keep_separator**: `True`
  - **is_separator_regex**: `False`

### Dataset Statistics
- Total chunks: 3,800
- Average chunk size: 43.1 words
- Source files: 1



### Dataset Structure
This dataset contains the following fields:

- `text`: The content of each text chunk
- `source`: The source file path for the chunk
- `id`: Unique identifier for each chunk