Svngoku/histoire-general-afrique-global-adaption

---
annotations_creators: []
language:
- fr
- en
language_creators: []
license: apache-2.0
multilinguality:
- monolingual
pretty_name: Svngoku/Histoire-General-Afrique-Global
size_categories:
- 1K<n<10K
source_datasets:
- extended|Svngoku/Histoire-General-Afrique-Global
tags:
- adaption
- instruction-tuning
- history
task_categories:
- question-answering
task_ids: []
---


This dataset is a remastered version of this [dataset](https://huggingface.co/datasets/Svngoku/Histoire-General-Afrique-Global) prepared using [Adaption's](https://adaptionlabs.ai/app/auth) Adaptive Data platform.

# Svngoku/Histoire-General-Afrique-Global

This dataset contains French-language text excerpts detailing the political, social, and economic history of Africa from the 16th to the 18th centuries. The content covers specific regions such as the Lower Guinea Coast and the Zambezi, discussing topics like ethnic migrations, kingdom formations, and trade dynamics. Each sample consists of a single text field derived from historical academic literature.

I started with books from the General History of Africa that had already been split into smaller text chunks.

The initial dataset contained only two fields for each example: the chunk itself and the title of the page it came from.

I used the chunk as the prompt and the page title as context.

Then, I used Adaptive Data to generate the completion and improve the output through hallucination control and alignment steps.

I also added a custom system prompt to better guide how the data was adapted.

### Dataset size

There are 1,315 data points in this dataset. This is an instruction tuning dataset.

### API Code

```py
# Step 1: Create dataset from HuggingFace
dataset = client.datasets.create_from_huggingface(
url="Svngoku/Histoire-General-Afrique-Global",
files=["train.parquet"],
)

# Step 2: Run augmentation
job = client.datasets.run(
dataset.dataset_id,
column_mapping={
"prompt": "content",
"context": ["source"]
},
)
```

### Quality of Remastered Dataset

The final quality is A, with a relative quality improvement of 800.0%.

### Domain
- History (8%)

### Language
- French (100%)

### Tone
- Analytical (16%)
- Informative (8%)
- Historic (4%)

### Evaluation Results

- **Quality Gains:**
<img src="https://proteus-prod-public.s3.us-east-1.amazonaws.com/temp/146fcd3c-49e2-40b7-852e-5ec97aa00eaf.png" alt="QualityGains" style="max-width: 50%; display: block; margin-left: auto; margin-right: auto;" />

- **Grade Improvement:**
<img src="https://proteus-prod-public.s3.us-east-1.amazonaws.com/temp/0e923b0d-e7e0-430c-83c2-97458cfa67e8.png" alt="Grade" style="max-width: 50%; display: block; margin-left: auto; margin-right: auto;" />

- **Percentile Chart:**
<img src="https://proteus-prod-public.s3.us-east-1.amazonaws.com/temp/030d6eca-efef-40bd-a4a8-27656e6e9326.png" alt="Percentile Chart" style="max-width: 50%; display: block; margin-left: auto; margin-right: auto;" />