d2g_generated / README.md
voorhs's picture
Duplicate from DeepPavlov/d2g_generated
555649c
metadata
dataset_info:
  features:
    - name: graph
      struct:
        - name: edges
          list:
            - name: source
              dtype: int64
            - name: target
              dtype: int64
            - name: utterances
              sequence: string
        - name: nodes
          list:
            - name: id
              dtype: int64
            - name: is_start
              dtype: bool
            - name: label
              dtype: string
            - name: utterances
              sequence: string
    - name: topic
      dtype: string
    - name: dialogs
      list:
        - name: id
          dtype: string
        - name: messages
          list:
            - name: participant
              dtype: string
            - name: text
              dtype: string
  splits:
    - name: train
      num_bytes: 4865580
      num_examples: 402
  download_size: 658444
  dataset_size: 4865580
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
language:
  - en
pretty_name: d2g_g

Dialogue2Graph

This is a dialogue dataset. Every dialogue graph has corresponding set of dialogues. It is intended for machine learning research and experimentation.

This dataset is generated with LLMs to test dialogue2graph task.

Usage

from datasets import load_dataset

dataset = load_dataset("DeepPavlov/d2g_generated", token=True)

This dataset is designed to develop dialogue2graph solution.