language:
- en
license: mit
size_categories: - 1K<n<10K
configs: - config_name: default
data_files:- split: train
path: data/train-* - split: test
path: data/test-*
dataset_info:
features: - name: dataset
dtype: string - name: chosen
list:- name: content
dtype: string - name: role
dtype: string
- name: content
- name: rejected
list:- name: content
dtype: string - name: role
dtype: string
- name: content
- name: chosen_rating
dtype: float64 - name: rejected_rating
dtype: float64
splits: - name: train
num_bytes: 41362946
num_examples: 6750 - name: test
num_bytes: 4586808
num_examples: 750
download_size: 24232011
dataset_size: 45949754
tags:
- split: train
- distilabel
- synthetic
- dpo
- argilla
Argilla DPO Mix 7K Dataset
A small cocktail combining DPO datasets built by Argilla with distilabel. The goal of this dataset is having a small, high-quality DPO dataset by filtering only highly rated chosen responses.
Datasets mixed
As already mentioned, this dataset mixes the following datasets:
argilla/distilabel-capybara-dpo-7k-binarized: random sample of highly scored chosen responses (>=4).argilla/distilabel-intel-orca-dpo-pairs: random sample of highly scored chosen responses (>=8).argilla/ultrafeedback-binarized-preferences-cleaned: random sample of highly scored chosen responses (>=4).
The samples have been randomly selected from the original datasets with a proportion of 0.33 each, as can be seen via the dataset column of the dataset.
Next steps
- Adding more samples
- Use data selection techniques to improve the diversity, usefulness, and complexity of the dataset.