Datasets:
Tasks:
Question Answering
Modalities:
Text
Formats:
json
Languages:
English
Size:
10K - 100K
ArXiv:
License:
metadata
license: apache-2.0
task_categories:
- question-answering
language:
- en
size_categories:
- 10K<n<100K
Used the technique from Unnatural Instructions: Tuning Language Models with (Almost) No Human Labor and Mixtral8x7B (Base Model) to generate this diverse, fully-synthetic, fully open-source set of 30,000 conversation starters.
See also: unnaturalhermes-questions-100k, a distinct set of 100k examples just like this, if you want more training data.