---
dataset_info:
  features:
  - name: text
    dtype: string
  - name: token_idx
    dtype: int64
  - name: hidden_states
    sequence: float64
  splits:
  - name: train
configs:
- config_name: default
  data_files:
  - split: train
    path: 
    - c4/data_*
    - cosmopedia_web/data_*
    - redpajama_cc/data_*
---