metadata
dataset_info:
features:
- name: seed
dtype: int64
- name: needle
dtype: int64
- name: position
dtype: float64
- name: context_length
dtype: int64
- name: haystack_text
dtype: string
- name: needle_token_position
dtype: int64
splits:
- name: train
num_bytes: 3052189401
num_examples: 1408
download_size: 159846856
dataset_size: 3052189401
configs:
- config_name: default
data_files:
- split: train
path: data/train-*