metadata
license: odc-by
dataset_info:
- config_name: default
features:
- name: post_id
dtype: string
- name: domain
dtype: string
- name: upvote_ratio
dtype: float64
- name: seconds_difference
dtype: float64
- name: score_ratio
dtype: float64
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: score_chosen
dtype: int64
- name: score_rejected
dtype: int64
splits:
- name: train
num_bytes: 123987035
num_examples: 89971
- name: validation
num_bytes: 7709028
num_examples: 4936
- name: test
num_bytes: 7793116
num_examples: 4938
download_size: 88713712
dataset_size: 139489179
- config_name: pair-STS-scored
features:
- name: post_id
dtype: string
- name: domain
dtype: string
- name: upvote_ratio
dtype: float64
- name: seconds_difference
dtype: float64
- name: score_ratio
dtype: float64
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: score_chosen
dtype: int64
- name: score_rejected
dtype: int64
- name: cosine_similarity
dtype: float64
splits:
- name: train
num_bytes: 124706803
num_examples: 89971
- name: validation
num_bytes: 7748516
num_examples: 4936
- name: test
num_bytes: 7832620
num_examples: 4938
download_size: 89455007
dataset_size: 140287939
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
- config_name: pair-STS-scored
data_files:
- split: train
path: pair-STS-scored/train-*
- split: validation
path: pair-STS-scored/validation-*
- split: test
path: pair-STS-scored/test-*
tags:
- dpo
>>> dataset = load_dataset("pszemraj/SHP-2-dpo-100k_sample")
>>> dataset
DatasetDict({
train: Dataset({
features: ['post_id', 'domain', 'upvote_ratio', 'seconds_difference', 'score_ratio', 'prompt', 'chosen', 'rejected', 'score_chosen', 'score_rejected'],
num_rows: 89971
})
validation: Dataset({
features: ['post_id', 'domain', 'upvote_ratio', 'seconds_difference', 'score_ratio', 'prompt', 'chosen', 'rejected', 'score_chosen', 'score_rejected'],
num_rows: 4936
})
test: Dataset({
features: ['post_id', 'domain', 'upvote_ratio', 'seconds_difference', 'score_ratio', 'prompt', 'chosen', 'rejected', 'score_chosen', 'score_rejected'],
num_rows: 4938
})
})