Geonwoohong/pile-uncopyrighted-test-tokenized-gpt2

Name: Geonwoohong/pile-uncopyrighted-test-tokenized-gpt2
Creator: Geonwoohong
Published: 2025-11-14 15:07:32
License: 暂无描述

Hugging Face2025-11-14 更新2025-12-20 收录

下载链接：

https://hf-mirror.com/datasets/Geonwoohong/pile-uncopyrighted-test-tokenized-gpt2

下载链接

链接失效反馈

官方服务：

资源简介：

--- license: other ---

提供机构：

Geonwoohong

Geonwoohong/pile-uncopyrighted-train-tokenized-gpt2

--- license: other ---

Hugging Face2025-11-14 更新10

alancooney/sae-monology-pile-uncopyrighted-tokenizer-gpt2

--- license: mit dataset_info: features: - name: input_ids sequence: int32 splits: - name: train num_bytes: 31522321636 num_examples: 30663737 download_size: 17207116581 datase

Hugging Face2024-01-02 更新50

apollo-research/sae-monology-pile-uncopyrighted-tokenizer-gpt2

--- dataset_info: features: - name: input_ids sequence: int32 splits: - name: train num_bytes: 39285716200.0 num_examples: 9581882 download_size: 16728794109 dataset_size: 3928

Hugging Face2024-02-09 更新40

Geonwoohong/pile-uncopyrighted-6b-tokenized-gpt2

Pile-Uncopyrighted的预标记子集，使用GPT-2 BPE编码，适用于Mixture-of-Experts（MoE）实验。每个记录包含固定长度的1024个标记序列，适用于解码器语言模型的训练。数据集包括Pile-CC、USPTO Backgrounds、Wikipedia (en)、GitHub、PubMed Abstracts和StackExchange等未版权子集。

Hugging Face2025-10-21 更新30

apollo-research/monology-pile-uncopyrighted-tokenizer-gpt2

--- dataset_info: features: - name: input_ids sequence: int32 splits: - name: train num_bytes: 196198407000.0 num_examples: 47853270 download_size: 83576335505 dataset_size: 19

Hugging Face2024-03-07 更新90