README.md

metadata

dataset_info:
  features:
    - name: message_tree_id
      dtype: string
    - name: prompter_message_id
      dtype: string
    - name: assistant_message_id
      dtype: string
    - name: instruction
      dtype: string
    - name: input
      dtype: string
    - name: output
      dtype: string
    - name: assistant_detoxify
      struct:
        - name: identity_attack
          dtype: float64
        - name: insult
          dtype: float64
        - name: obscene
          dtype: float64
        - name: severe_toxicity
          dtype: float64
        - name: sexual_explicit
          dtype: float64
        - name: threat
          dtype: float64
        - name: toxicity
          dtype: float64
    - name: prompter_detoxify
      struct:
        - name: identity_attack
          dtype: float64
        - name: insult
          dtype: float64
        - name: obscene
          dtype: float64
        - name: severe_toxicity
          dtype: float64
        - name: sexual_explicit
          dtype: float64
        - name: threat
          dtype: float64
        - name: toxicity
          dtype: float64
    - name: assistant_labels
      struct:
        - name: count
          sequence: int32
        - name: name
          sequence: string
        - name: value
          sequence: float64
    - name: prompter_labels
      struct:
        - name: count
          sequence: int32
        - name: name
          sequence: string
        - name: value
          sequence: float64
  splits:
    - name: train
      num_bytes: 10327876
      num_examples: 4744
  download_size: 4458529
  dataset_size: 10327876
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: apache-2.0
task_categories:
  - text-generation
language:
  - ja
tags:
  - self-rewarding
  - oasst1
size_categories:
  - 1K<n<10K

Dataset Card for Dataset Name

• kunishou/oasst1-89k-ja
• OpenAssistant/oasst1

を元に、self-rewardingのIFT(Instruction Fine-Tuning data)を作成しました。 Self-rewardingの論文では3,200件に絞り込んでいますが、このデータは4,744件となっています。
さらに人手でのフィルタリングをする必要があるかもしれません。

Dataset Details

Dataset Description

• Curated by: HachiML
• Language(s) (NLP): Japanese
• License: Apache-2.0

Filtering Rule

以下のルールで絞り込んでいます。

• First Conversational Turn
• rank 0 の回答をもつ
• Single Turn Conversation