jxm's picture
Upload README.md with huggingface_hub
153fbbc
metadata
dataset_info:
  features:
    - name: input_ids
      sequence: int32
    - name: attention_mask
      sequence: int8
    - name: labels
      sequence: int64
    - name: length
      dtype: int64
    - name: embedder_input_ids
      sequence: int64
    - name: embedder_attention_mask
      sequence: int64
    - name: frozen_embeddings
      sequence: float32
    - name: idx
      dtype: int64
    - name: str_original
      dtype: string
    - name: str_reconstruction
      dtype: string
  splits:
    - name: train
      num_bytes: 13289065
      num_examples: 100
  download_size: 0
  dataset_size: 13289065
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Dataset Card for "llama-7b__model__one_million_instructions__reconstructions_sample"

More Information needed