metadata
license: apache-2.0
task_categories:
- question-answering
language:
- zh
pretty_name: Outlook
size_categories:
- 1K<n<10K
The main propose of developing this dataset is to improve creative writing ability and human-like understanding.merged with world qa dataset(80%),result in coherent self-cognition ability,but still in experiment.
NEED FURTHER DEVELOPMENT
Contact me to unlock the dataset.
Can be used in reasoning model. It do have coherent reasoning trace.(Generated by ds v3.1 and refined by a private model)
Alpaca format. SFT or pt only