Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,28 @@
|
|
1 |
-
---
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
# This is metadata for the Hugging Face Hub
|
3 |
+
# Add relevant tags to make it discoverable
|
4 |
+
language: ttj
|
5 |
+
tags:
|
6 |
+
- translation
|
7 |
+
- generation
|
8 |
+
- grammar
|
9 |
+
- rutooro
|
10 |
+
license: creativeml-openrail-m
|
11 |
+
---
|
12 |
+
|
13 |
+
# Rutooro Multitask Dataset
|
14 |
+
|
15 |
+
This dataset contains a collection of instruction-response pairs for fine-tuning a Large Language Model (LLM) on the Rutooro language. The dataset is prepared for a multi-task learning approach, including:
|
16 |
+
|
17 |
+
* **Translation:** English to Rutooro.
|
18 |
+
* **Monolingual Generation:** Continued stories and prose in Rutooro.
|
19 |
+
* **Grammar Instructions:** Explanations of Rutooro grammar rules.
|
20 |
+
|
21 |
+
### Data Source
|
22 |
+
|
23 |
+
The data was sourced from [mention your source, e.g., "manual transcription of local texts," or "openly licensed books like 'Engeso zaitu ez'Obuhangwa.'"]
|
24 |
+
|
25 |
+
### Structure
|
26 |
+
|
27 |
+
The dataset is a JSONL file with the following format:
|
28 |
+
`{"instruction": "...", "completion": "...", "task_type": "..."}`
|