HuggingFaceH4/llava-instruct-mix-vsft

---
dataset_info:
features:
- name: messages
list:
- name: content
list:
- name: index
dtype: int64
- name: text
dtype: string
- name: type
dtype: string
- name: role
dtype: string
- name: images
sequence: image
splits:
- name: train
num_bytes: 9992582190.928007
num_examples: 259155
- name: test
num_bytes: 525935525.39699405
num_examples: 13640
download_size: 11407075653
dataset_size: 10518517716.325
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
---

[theblackcat102/llava-instruct-mix](https://huggingface.co/datasets/theblackcat102/llava-instruct-mix) reformated for VSFT with TRL's SFT Trainer.

See https://github.com/huggingface/trl/blob/main/examples/scripts/vsft_llava.py.

原始信息汇总

数据集概述

数据集信息

特征

• messages
  • content
    • index: 数据类型为 int64
    • text: 数据类型为 string
    • type: 数据类型为 string
  • role: 数据类型为 string
• images: 序列类型为 image

数据分割

• train
  • 字节数: 9992582190.928007
  • 样本数: 259155
• test
  • 字节数: 525935525.39699405
  • 样本数: 13640

数据大小

• 下载大小: 11407075653
• 数据集大小: 10518517716.325

配置

• default
  • 数据文件
    • train: 路径为 data/train-*
    • test: 路径为 data/test-*