---
dataset_info:
  - config_name: index
    data_files: index.parquet
    default: true
  - config_name: embeddings_eng
    data_files: embeddings_eng.parquet
  - config_name: embeddings_fra
    data_files: embeddings_fra.parquet
  - config_name: embeddings_ita
    data_files: embeddings_ita.parquet
  - config_name: embeddings_pol
    data_files: embeddings_pol.parquet
  - config_name: embeddings_spa
    data_files: embeddings_spa.parquet
  - config_name: intervention_by_chapter
    data_files: intervention_by_chapter.parquet
  - config_name: intervention_summary
    data_files: intervention_summary.parquet
configs:
  - config_name: index
    data_files: index.parquet
    default: true
  - config_name: embeddings_eng
    data_files: embeddings_eng.parquet
  - config_name: embeddings_fra
    data_files: embeddings_fra.parquet
  - config_name: embeddings_ita
    data_files: embeddings_ita.parquet
  - config_name: embeddings_pol
    data_files: embeddings_pol.parquet
  - config_name: embeddings_spa
    data_files: embeddings_spa.parquet
  - config_name: intervention_by_chapter
    data_files: intervention_by_chapter.parquet
  - config_name: intervention_summary
    data_files: intervention_summary.parquet
license: cc-by-4.0
task_categories:
  - text-classification
  - feature-extraction
language:
  - en
  - fr
  - it
  - pl
  - es
  - grc
tags:
  - bible
  - translation-studies
  - embeddings
  - interlinear
pretty_name: "SIGHUM Interlinear Vector Baselines"
size_categories:
  - 10K<n<100K
---

# SIGHUM Interlinear Vector Baselines

Companion dataset for the paper *Degree Zero of Translation: Using Interlinear Baselines to Quantify Translator Intervention* (SIGHUM 2026).

## Contents

| Config | Description |
|---|---|
| **index** (default) | Translation metadata: 79 translations across 5 languages with strategy labels |
| embeddings_{lang} | Chapter-level Qwen3-Embedding-8B vectors (4096-dim) per translation × chapter |
| intervention_by_chapter | L2 distance to interlinear baseline per translation × chapter |
| intervention_summary | Per-translation summary statistics (mean, std, median distance) |

## Quick Start

```python
from datasets import load_dataset

# Browse translations
index = load_dataset("mrapacz/sighum-interlinear-vector-baselines", "index", split="train")
print(index.to_pandas())

# Load English chapter-level embeddings
eng = load_dataset("mrapacz/sighum-interlinear-vector-baselines", "embeddings_eng", split="train")

# Load intervention distances
intervention = load_dataset("mrapacz/sighum-interlinear-vector-baselines", "intervention_by_chapter", split="train")
```

## Languages

English (17), French (15), Italian (13), Polish (17), Spanish (17) — 74 prose translations + 5 interlinear baselines.

## Model

All embeddings computed with [Qwen/Qwen3-Embedding-8B](https://huggingface.co/Qwen/Qwen3-Embedding-8B).

## Links

- **Code & Poster**: [github.com/mrapacz/sighum-interlinear-vector-baselines](https://github.com/mrapacz/sighum-interlinear-vector-baselines)
- **Targum Corpus**: [huggingface.co/datasets/mrapacz/targum-corpus](https://huggingface.co/datasets/mrapacz/targum-corpus)