Siesher's picture
Upload dataset
08b2408 verified
metadata
dataset_info:
  - config_name: gspo
    features:
      - name: prompt
        dtype: string
      - name: answer
        dtype: string
      - name: domain
        dtype: string
      - name: type
        dtype: string
      - name: difficulty
        dtype: string
    splits:
      - name: train
        num_bytes: 6562375
        num_examples: 14250
      - name: test
        num_bytes: 345388
        num_examples: 750
    download_size: 1501037
    dataset_size: 6907763
  - config_name: preference
    features:
      - name: prompt
        dtype: string
      - name: chosen
        dtype: string
      - name: rejected
        dtype: string
      - name: domain
        dtype: string
      - name: chosen_model
        dtype: string
      - name: rejected_model
        dtype: string
      - name: chosen_questions
        dtype: int64
      - name: rejected_questions
        dtype: int64
    splits:
      - name: train
        num_bytes: 50107976
        num_examples: 11967
      - name: test
        num_bytes: 2637923
        num_examples: 630
    download_size: 20538999
    dataset_size: 52745899
  - config_name: sft
    features:
      - name: instruction
        dtype: string
      - name: output
        dtype: string
      - name: domain
        dtype: string
      - name: type
        dtype: string
      - name: difficulty
        dtype: string
      - name: teacher_model
        dtype: string
      - name: source
        dtype: string
    splits:
      - name: train
        num_bytes: 209870717
        num_examples: 38330
      - name: test
        num_bytes: 11049285
        num_examples: 2018
    download_size: 88505027
    dataset_size: 220920002
configs:
  - config_name: gspo
    data_files:
      - split: train
        path: gspo/train-*
      - split: test
        path: gspo/test-*
  - config_name: preference
    data_files:
      - split: train
        path: preference/train-*
      - split: test
        path: preference/test-*
  - config_name: sft
    data_files:
      - split: train
        path: sft/train-*
      - split: test
        path: sft/test-*