metadata
dataset_info:
features:
- name: schema_version
dtype: string
- name: candidate_id
dtype: string
- name: author_name
dtype: string
- name: selection_protocol
struct:
- name: described_in
dtype: string
- name: K
dtype: int64
- name: strategy
dtype: string
- name: target_words
dtype: int64
- name: decontam_threshold_cos_vs_ryan
dtype: float64
- name: decontam_cache_version
dtype: string
- name: n_eligible_comments
dtype: int64
- name: n_after_decontam
dtype: int64
- name: chosen_n_words
list: int64
- name: chosen_max_cos_vs_ryan
list: float64
- name: system
dtype: string
- name: few_shots
list:
- name: kind
dtype: string
- name: post_title
dtype: string
- name: text
dtype: string
- name: source_id
dtype: string
- name: endpoint
dtype: string
- name: sampler_args
struct:
- name: max_tokens
dtype: int64
- name: temperature
dtype: float64
- name: top_p
dtype: float64
- name: seed
dtype: int64
- name: 'n'
dtype: int64
splits:
- name: train
num_bytes: 10395
num_examples: 1
download_size: 28937
dataset_size: 10395
configs:
- config_name: default
data_files:
- split: train
path: data/train-*