metadata
dataset_info:
features:
- name: context
dtype: string
- name: context_length
dtype: int64
- name: depth_percent
dtype: int64
- name: gpt4o_response
dtype: string
- name: num_retrieved_needles
dtype: float64
splits:
- name: train
num_bytes: 152171977
num_examples: 350
download_size: 91717493
dataset_size: 152171977
configs:
- config_name: default
data_files:
- split: train
path: data/train-*