Datasets:

Tasks:
Other
Modalities:
Text
Formats:
parquet
ArXiv:
Libraries:
Datasets
pandas
License:
alexrs commited on
Commit
f9ea045
·
verified ·
1 Parent(s): 7942c27

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -10
README.md CHANGED
@@ -144,9 +144,9 @@ configs:
144
  path: demographics/train-*
145
  tags: []
146
  ---
147
- ![Aya Header](https://huggingface.co/datasets/CohereForAI/aya_dataset/resolve/main/aya_header.png)
148
  # Dataset Summary
149
- The `Aya Dataset` is a multilingual instruction fine-tuning dataset curated by an open-science community via [Aya Annotation Platform](https://aya.for.ai/) from Cohere For AI. The dataset contains a total of 204k human-annotated prompt-completion pairs along with the demographics data of the annotators.<br>
150
  This dataset can be used to train, finetune, and evaluate multilingual LLMs.
151
 
152
  - **Curated by:** Contributors of [Aya Open Science Intiative](https://aya.for.ai/).
@@ -156,11 +156,11 @@ This dataset can be used to train, finetune, and evaluate multilingual LLMs.
156
 
157
  | Name | Explanation |
158
  |------|--------------|
159
- | [aya_dataset](https://huggingface.co/datasets/CohereForAI/aya_dataset) | Human-annotated multilingual instruction finetuning dataset, comprising over 204K instances across 65 languages. |
160
- | [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, providing 513M instances for various tasks.|
161
- | [aya_collection_language_split](https://huggingface.co/datasets/CohereForAI/aya_collection_language_split) | Aya Collection structured based on language level subsets. |
162
- | [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.|
163
- | [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.|
164
 
165
  # Dataset
166
  The `Aya Dataset` comprises of two types of data:
@@ -174,10 +174,10 @@ To load this dataset consisting of both prompt-completions and demographics data
174
  from datasets import load_dataset
175
 
176
  # Load the annotations dataset
177
- aya_dataset = load_dataset("CohereForAI/aya_dataset")
178
 
179
  # Load the demographics dataset
180
- aya_demographics = load_dataset("CohereForAI/aya_dataset", "demographics")
181
  ```
182
 
183
  ## Data Fields
@@ -359,7 +359,7 @@ The dataset covers 65 languages: 28 high-resource, 12 mid-resource, and 31 low-r
359
 
360
 
361
  ## Authorship
362
- - **Publishing Organization:** [Cohere For AI](https://cohere.com/research)
363
  - **Industry Type:** Not-for-profit - Tech
364
  - **Contact Details:** https://aya.for.ai/
365
 
 
144
  path: demographics/train-*
145
  tags: []
146
  ---
147
+ ![Aya Header](https://huggingface.co/datasets/CohereLabs/aya_dataset/resolve/main/aya_header.png)
148
  # Dataset Summary
149
+ The `Aya Dataset` is a multilingual instruction fine-tuning dataset curated by an open-science community via [Aya Annotation Platform](https://aya.for.ai/) from Cohere Labs. The dataset contains a total of 204k human-annotated prompt-completion pairs along with the demographics data of the annotators.<br>
150
  This dataset can be used to train, finetune, and evaluate multilingual LLMs.
151
 
152
  - **Curated by:** Contributors of [Aya Open Science Intiative](https://aya.for.ai/).
 
156
 
157
  | Name | Explanation |
158
  |------|--------------|
159
+ | [aya_dataset](https://huggingface.co/datasets/CohereLabs/aya_dataset) | Human-annotated multilingual instruction finetuning dataset, comprising over 204K instances across 65 languages. |
160
+ | [aya_collection](https://huggingface.co/datasets/CohereLabs/aya_collection) | Created by applying instruction-style templates from fluent speakers to 44 datasets, including translations of 19 instruction-style datasets into 101 languages, providing 513M instances for various tasks.|
161
+ | [aya_collection_language_split](https://huggingface.co/datasets/CohereLabs/aya_collection_language_split) | Aya Collection structured based on language level subsets. |
162
+ | [aya_evaluation_suite](https://huggingface.co/datasets/CohereLabs/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.|
163
+ | [aya_redteaming](https://huggingface.co/datasets/CohereLabs/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.|
164
 
165
  # Dataset
166
  The `Aya Dataset` comprises of two types of data:
 
174
  from datasets import load_dataset
175
 
176
  # Load the annotations dataset
177
+ aya_dataset = load_dataset("CohereLabs/aya_dataset")
178
 
179
  # Load the demographics dataset
180
+ aya_demographics = load_dataset("CohereLabs/aya_dataset", "demographics")
181
  ```
182
 
183
  ## Data Fields
 
359
 
360
 
361
  ## Authorship
362
+ - **Publishing Organization:** [Cohere Labs](https://cohere.com/research)
363
  - **Industry Type:** Not-for-profit - Tech
364
  - **Contact Details:** https://aya.for.ai/
365