rubricreward/mmr3-bpo
收藏Hugging Face2026-02-10 更新2026-04-05 收录
下载链接：
https://hf-mirror.com/datasets/rubricreward/mmr3-bpo
下载链接
链接失效反馈官方服务：
资源简介：
---
dataset_info:
features:
- name: id
dtype: string
- name: source_dataset
dtype: string
- name: languages
sequence: string
- name: inputs
list:
- name: role
dtype: string
- name: content
list:
- name: type
dtype: string
- name: content
dtype: string
- name: candidates
list:
- name: role
dtype: string
- name: content
list:
- name: type
dtype: string
- name: content
dtype: string
- name: rubrics
list:
- name: modality_mixture
struct:
- name: image
dtype: bool
- name: text
dtype: bool
- name: audio
dtype: bool
- name: video
dtype: bool
- name: format
dtype:
class_label:
names:
'0': pointwise
'1': pairwise
'2': binary
'3': ''
- name: task
dtype: string
- name: scores
list:
- name: score
dtype: string
- name: score_description
dtype: string
- name: teacher_evaluations
list:
- name: teacher_model
dtype: string
- name: eval_detail
list:
- name: scores
list:
- name: model_reasoning
dtype: string
- name: explanation
dtype: string
- name: score
dtype: string
splits:
- name: train
num_bytes: 26787095
num_examples: 13881
- name: validation
num_bytes: 594073
num_examples: 298
- name: test
num_bytes: 330351
num_examples: 200
download_size: 11971561
dataset_size: 27711519
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
---
提供机构：
rubricreward
