Py之tiktoken：tiktoken的简介、安装、使用方法之详细攻略

目录

tiktoken的简介

1、性能：tiktoken比一个类似的开源分词器快3到6倍

tiktoken的安装

tiktoken的使用方法

1、基础用法

(1)、用于OpenAI模型的快速BPE标记器

(2)、帮助可视化BPE过程的代码

---

tiktoken的简介

tiktoken是一个用于OpenAI模型的快速BPE标记器。

1、性能：tiktoken比一个类似的开源分词器快3到6倍

tiktoken的安装

pip install tiktoken

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple tiktoken

C:\Windows\system32>pip install -i https://pypi.tuna.tsinghua.edu.cn/simple tiktoken
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Collecting tiktoken
  Downloading https://pypi.tuna.tsinghua.edu.cn/packages/91/cf/7f3b821152f7abb240950133c60c394f7421a5791b020cedb190ff7a61b4/tiktoken-0.5.1-cp39-cp39-win_amd64.whl (760 kB)
     |████████████████████████████████| 760 kB 726 kB/s
Requirement already satisfied: regex>=2022.1.18 in d:\programdata\anaconda3\lib\site-packages (from tiktoken) (2022.3.15)
Requirement already satisfied: requests>=2.26.0 in d:\programdata\anaconda3\lib\site-packages (from tiktoken) (2.31.0)
Requirement already satisfied: charset-normalizer<4,>=2 in d:\programdata\anaconda3\lib\site-packages (from requests>=2.26.0->tiktoken) (2.0.12)
Requirement already satisfied: urllib3<3,>=1.21.1 in d:\programdata\anaconda3\lib\site-packages (from requests>=2.26.0->tiktoken) (1.26.9)
Requirement already satisfied: idna<4,>=2.5 in d:\programdata\anaconda3\lib\site-packages (from requests>=2.26.0->tiktoken) (3.3)
Requirement already satisfied: certifi>=2017.4.17 in d:\programdata\anaconda3\lib\site-packages (from requests>=2.26.0->tiktoken) (2021.10.8)
Installing collected packages: tiktoken
Successfully installed tiktoken-0.5.1

tiktoken的使用方法

1、基础用法

(1)、用于OpenAI模型的快速BPE标记器

import tiktoken
enc = tiktoken.get_encoding("cl100k_base")
assert enc.decode(enc.encode("hello world")) == "hello world"

# To get the tokeniser corresponding to a specific model in the OpenAI API:
enc = tiktoken.encoding_for_model("gpt-4")

(2)、帮助可视化BPE过程的代码

from tiktoken._educational import *

# Train a BPE tokeniser on a small amount of text
enc = train_simple_encoding()

# Visualise how the GPT-4 encoder encodes text
enc = SimpleBytePairEncoding.from_tiktoken("cl100k_base")
enc.encode("hello world aaaaaaaaaaaa")