mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-10-06 00:57:33 +08:00
71018fb62e
* Add ci case for min token and max token * 【CI case】include total_tokens in the last packet of completion interface stream output --------- Co-authored-by: xujing43 <xujing43@baidu.com>
42 lines
1.1 KiB
Python
42 lines
1.1 KiB
Python
#!/bin/env python3
|
|
# -*- coding: utf-8 -*-
|
|
# @author xujing43
|
|
# encoding=utf-8 vi:ts=4:sw=4:expandtab:ft=python
|
|
|
|
"""
|
|
Checking for /v1/completions parameters
|
|
"""
|
|
|
|
import json
|
|
|
|
from core import (
|
|
TEMPLATE,
|
|
URL,
|
|
build_request_payload,
|
|
send_request,
|
|
)
|
|
|
|
URL = URL.replace("/v1/chat/completions", "/v1/completions")
|
|
|
|
def test_completion_total_tokens():
|
|
data = {
|
|
"prompt": "你是谁",
|
|
"stream": True,
|
|
"stream_options": {"include_usage": True, "continuous_usage_stats": True},
|
|
}
|
|
|
|
payload = build_request_payload(TEMPLATE, data)
|
|
resp = send_request(URL, payload, stream=True)
|
|
last_data = None
|
|
for line in resp.iter_lines(decode_unicode=True):
|
|
if line.strip() == "data: [DONE]":
|
|
break
|
|
if line.strip() == "" or not line.startswith("data: "):
|
|
continue
|
|
line = line[len("data: "):]
|
|
last_data = json.loads(line)
|
|
usage = last_data["usage"]
|
|
total_tokens = usage["completion_tokens"] + usage["prompt_tokens"]
|
|
assert "total_tokens" in usage, "total_tokens 不存在"
|
|
assert usage["total_tokens"]== total_tokens, "total_tokens计数不正确"
|
|
|