This synthetic dataset was generated to improvement model thinking and reflection in systematical way.
Columns
question
answer_with_tags
text (processed for Llama 3.1)
Dataset
DatasetDict({
train: Dataset({
features: ['question', 'answer_with_tags', 'text'],
num_rows: 17129
})
test: Dataset({
features: ['question', 'answer_with_tags', 'text'],
num_rows: 52
})
})
inspired by mattshumer (sharedGPT dataset) and… See the full description on the dataset page:
https://huggingface.co/datasets/mychen76/synthetic-think-and-reflection_v1.
