Datasets:

Modalities:
Tabular
Text
Formats:
parquet
ArXiv:
Libraries:
Datasets
Dask
License:
MarziehFadaee commited on
Commit
c18f0ee
1 Parent(s): b6dcae7

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -0
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