metadata
dataset_info:
features:
- name: text
dtype: string
splits:
- name: train
num_bytes: 1294193
num_examples: 1000
download_size: 397427
dataset_size: 1294193
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
Syntetic emails. Split by inquiry and answer email
Consume and run into a notebook colab
You can switch the dataset there from mlabonne/guanaco-llama2-1k
to spielhoelle/film-equipment-order-emails