mansaripo/ClimbMix_SYNTH_shuffled

---
dataset_info:
features:
- name: text
dtype: string
splits:
- name: train
num_bytes: 1132201222189
num_examples: 631223639
download_size: 1132201222189
dataset_size: 1132201222189
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

# ClimbMix + SYNTH Shuffled

A globally shuffled combination of
[gvlassis/ClimbMix](https://huggingface.co/datasets/gvlassis/ClimbMix) and
[PleIAs/SYNTH](https://huggingface.co/datasets/PleIAs/SYNTH),
interleaved by file-size weight.

| Source | Rows | Size | Weight |
|--------|------|------|--------|
| ClimbMix | 553,315,056 | ~955 GB | ~85.4% |
| SYNTH | 77,908,583 | ~164 GB | ~14.6% |
| **Total** | **631,223,639** | **~1,119 GB** | **100%** |

Both sources were individually shuffled before interleaving.

数据集信息：
特征：
- 名称：text
数据类型：字符串
数据划分：
- 名称：train（训练集）
字节数：1132201222189
样本数量：631223639
下载大小：1132201222189
数据集总大小：1132201222189
配置项：
- 配置名称：default（默认配置）
数据文件：
- 数据划分：train
文件路径：data/train-*

# ClimbMix + SYNTH 混洗数据集

本数据集为[gvlassis/ClimbMix](https://huggingface.co/datasets/gvlassis/ClimbMix)与[PleIAs/SYNTH](https://huggingface.co/datasets/PleIAs/SYNTH)的组合，按照文件大小权重进行交错排列，并完成全局混洗。

| 数据源 | 样本数量 | 数据大小 | 权重占比 |
|--------|----------|----------|----------|
| ClimbMix | 553,315,056 | ~955 GB | ~85.4% |
| SYNTH | 77,908,583 | ~164 GB | ~14.6% |
| **总计** | **631,223,639** | **~1,119 GB** | **100%** |

两个数据源在进行交错排列前均已各自完成混洗。