您可以透過以下實作來實現 Multiple Functions 的場景。
如以下的例子是撈取對話歷史加上 system prompt後，餵入兩個 functions (summarization & extract_keywords) 讓模型依據對話歷史回傳對應 function 的 arguments。
| 程式碼區塊 | ||
|---|---|---|
| ||
import json
response = await chat(
conversation=CURRENT_CONVERSATION + [
{
"role": "system",
"content": "Only use the functions you have been provided with."
}
],
functions=[
{
"name": "summarization",
"description": "summarize the conversation",
"parameters": {
"type": "object",
"properties": {
"summary": {
"type": "string",
"description": "the summary of the conversation"
}
},
"required": ["summary"]
}
},
{
"name": "extract_keywords",
"description": "extracting from the conversation",
"parameters": {
"type": "object",
"properties": {
"keywords": {
"type": "array",
"items": {
"type": "string",
"description": "one keyword extracted from the conversation"
}
}
},
"required": ["keywords"]
}
}
]
)
if 'function_call' in response:
if response['function_call']['name'] == 'summarization':
print("The summary of the converstaion is ", json.loads(response['function_call']['arguments'])['summary'])
elif response['function_call']['name'] == 'extracted_keywords':
print("The keywords of the converstaion are ", json.loads(response['function_call']['arguments'])['keywords'])
else:
print(response['content']) |