sparkapi-python 1.0.3

SparkDesk API and Client

SparkDesk API and Client

https://www.xfyun.cn/doc/spark/Web.html

接口说明 注意： 该接口可以正式使用。如您需要申请使用，请点击前往产品页面

计费说明：

服务引擎	单价
讯飞星火认知大模型V1.5	0.18元/万tokens
讯飞星火认知大模型V2.0	0.36元/万tokens
讯飞星火认知大模型V3.0	0.36元/万tokens

Installation

python3 -m pip install -U sparkapi-python

Configuration

copy example.env to ~/.sparkapi.env and edit the file with your credentials.

SPARK_APP_ID=<your-app-id>
SPARK_API_SECRET=<your-api-secret>
SPARK_API_KEY=<your-api-key>
SPARK_API_MODEL='v3.0'          # v1.5, v2.0, v3.0
SPARK_CHAT_MAX_TOKENS=4096      # v1.5: 1-4096, v2.0: 1-8192
SPARK_CHAT_TEMPERATURE=0.5      # 0-1
SPARK_CHAT_TOP_K=4              # 1-6

Quickstart

CommandLine

sparkapi --help

# start a chat session
sparkapi chat

# get completion from your prompt
sparkapi prompt 详细介绍一下科大讯飞，输出Markdown结果

Python

from sparkapi.core.api import SparkAPI
from sparkapi.core.config import SparkConfig
config = SparkConfig().model_dump()
api = SparkAPI(**config)

# start a chat session
api.chat()

# get completion from prompt
res = api.get_completion('hello')
print(''.join(res))

# get completion from messages
messages = [
    {'role': 'user', 'content': 'hello'},
    {'role': 'assistant', 'content': 'Hello! How can I assist'},
    {'role': 'user', 'content': 'write me a Python script of BubbleSort'},
]
res = api.get_completion_from_messages(messages)
print(''.join(res))

Changelog

[1.0.3] - 2023-10-26

• Add support for v3.0