如何将文本向量化

安装环境

pip install openai==1.51.2

实现原理

使用兼容openai的客户端，直接调用embedding服务。实现代码如下：

import openai

base_url = "http://127.0.0.1:9999/v1"

input = "How are you?"

client = openai.Client(
    base_url=base_url,
    api_key="EMPTY"
)

# models = client.models.list()
#
# print(models)

resp = client.embeddings.create(
    model="bge-base-zh-v1.5",
    encoding_format="float",
    input=input
)

data = resp.data

embedding = data[0].embedding

print(embedding)

执行结果

D:\Anaconda3\envs\demo\python.exe E:/Work/Code/demo/get_embeddings.py
[0.0028676323127001524, -0.03512047231197357, 0.010429189540445805, -0.022234726697206497, 0.023326072841882706, -0.03975958004593849, 0.025496214628219604, 0.0247025266289711, -0.0029035459738224745, 0.03305435925722122, -0.023743389174342155, 0.015931695699691772, -0.01457991823554039, -0.0686136931180954, -0.013840625993907452, -0.0111240791156888, -0.04825543239712715, 0.025986848399043083, -0.01409992203116417, -0.0507720522582531, 0.05778298154473305, 0.04618502035737038, 0.022715436294674873, 0.0017536708619445562, 0.0442514643073082, -0.025457104668021202, 0.0022458313032984734, 0.019642774015665054, 0.0300145223736763, 0.03186137601733208, 
......
-0.04568396508693695, -0.010739356279373169, -0.06107804924249649, -0.003004075726494193, -0.011063020676374435, -0.0210788045078516, -0.021335413679480553, -0.002127290004864335, 0.0013641789555549622, -0.014254610985517502, 0.04622962698340416, -0.032668013125658035, -0.0046304636634886265, -0.008929592557251453, -0.01895640790462494, 0.011647721752524376, 0.023981524631381035, -0.06814616918563843, 0.029181446880102158, 0.034045781940221786, 0.05161082372069359, -0.02784288488328457, 0.009662372060120106, -0.001818168442696333, -0.008052065037190914, -0.03914210945367813, -0.04090451821684837, -0.014385311864316463]

Process finished with exit code 0