21 lines
643 B
Python
21 lines
643 B
Python
|
import requests
|
||
|
|
from ollama import Client, ResponseError
|
||
|
|
import tiktoken
|
||
|
|
|
||
|
|
class AICore:
|
||
|
|
modelMaxTokens = 128000
|
||
|
|
# 初始化 DeepSeek 使用的 Tokenizer (cl100k_base)
|
||
|
|
encoder = tiktoken.get_encoding("cl100k_base")
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
def __init__(self, model):
|
||
|
|
#初始化ollama客户端
|
||
|
|
ollamaClient = Client(host='http://localhost:11434', headers={'x-some-header': 'some-value'})
|
||
|
|
response = ollamaClient.show(model)
|
||
|
|
modelMaxTokens = response.modelinfo['qwen2.context_length']
|
||
|
|
|
||
|
|
def getPromptToken(self, prompt)-> int:
|
||
|
|
tokens = self.encoder.encode(prompt)
|
||
|
|
return len(tokens)
|
||
|
|
|