UCLAML/synthetic_data_rpo-iter1_score

---
dataset_info:
features:
- name: chosen_probs
dtype: float64
- name: chosen_probs_win
dtype: float64
- name: chosen_probs_lose
dtype: float64
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 95930511
num_examples: 20379
- name: test
num_bytes: 2348004
num_examples: 500
download_size: 50912186
dataset_size: 98278515
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
---
# Dataset Card for "synthetic_data_rpo-iter1_score"

[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)

The dataset includes multiple features such as chosen_probs, chosen_probs_win, chosen_probs_lose, etc., with data type float64. Additionally, the dataset contains two lists, chosen and rejected, each containing fields content and role with data type string. The dataset is divided into train and test splits, containing 20379 and 500 samples respectively. The download size of the dataset is 50912186 bytes, and the total size is 98278515 bytes. The configuration name of the dataset is default, with data file paths being data/train-* and data/test-* respectively.