MarioBarbeque's picture
a second dataset for evaluating the partial correctness of the second finetuning of the flan-T5-large model on a downsampled version of the DeepMind LingAlg 1D dataset
c0462b6 verified
metadata
dataset_info:
  features:
    - name: predicted_tokens
      sequence: int64
    - name: label_tokens
      sequence: int64
    - name: decoded_prediction
      dtype: string
    - name: decoded_label
      dtype: string
  splits:
    - name: train
      num_bytes: 837445
      num_examples: 10000
  download_size: 93540
  dataset_size: 837445
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*