MarziehFadaee
commited on
Commit
•
c18f0ee
1
Parent(s):
b6dcae7
Update README.md
Browse files
README.md
CHANGED
@@ -5658,6 +5658,7 @@ This collection incorporates instruction-style templates from fluent speakers an
|
|
5658 |
| [aya_collection](https://huggingface.co/datasets/CohereForAI/aya_collection) | Created by applying instruction-style templates from fluent speakers to 44 datasets, including translations of 19 instruction-style datasets into 101 languages. This collection structured based on dataset level subsets. An alternative version of the collection structured by language subsets is also available.|
|
5659 |
| [aya_collection_language_split](https://huggingface.co/datasets/CohereForAI/aya_collection_language_split) | Aya Collection structured based on language level subsets. |
|
5660 |
| [aya_evaluation_suite](https://huggingface.co/datasets/CohereForAI/aya_evaluation_suite) | A diverse evaluation set for multilingual open-ended generation, featuring 250 culturally grounded prompts in 7 languages, 200 translated prompts in 24 languages, and human-edited versions selected for cross-cultural relevance from English Dolly in 6 languages.|
|
|
|
5661 |
|
5662 |
|
5663 |
# Dataset
|
|
|
5658 |
| [aya_collection](https://huggingface.co/datasets/CohereForAI/aya_collection) | Created by applying instruction-style templates from fluent speakers to 44 datasets, including translations of 19 instruction-style datasets into 101 languages. This collection structured based on dataset level subsets. An alternative version of the collection structured by language subsets is also available.|
|
5659 |
| [aya_collection_language_split](https://huggingface.co/datasets/CohereForAI/aya_collection_language_split) | Aya Collection structured based on language level subsets. |
|
5660 |
| [aya_evaluation_suite](https://huggingface.co/datasets/CohereForAI/aya_evaluation_suite) | A diverse evaluation set for multilingual open-ended generation, featuring 250 culturally grounded prompts in 7 languages, 200 translated prompts in 24 languages, and human-edited versions selected for cross-cultural relevance from English Dolly in 6 languages.|
|
5661 |
+
| [aya_redteaming](https://huggingface.co/datasets/CohereForAI/aya_redteaming)| A red-teaming dataset consisting of harmful prompts in 8 languages across 9 different categories of harm with explicit labels for "global" and "local" harm.|
|
5662 |
|
5663 |
|
5664 |
# Dataset
|