Spaces:
Running
Running
File size: 5,185 Bytes
a984ba9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 |
#!/usr/bin/python3
# -*- coding: utf-8 -*-
import argparse
from datetime import datetime
import json
import os
from pathlib import Path
import sys
import time
from zoneinfo import ZoneInfo # Python 3.9+ 自带，无需安装
pwd = os.path.abspath(os.path.dirname(__file__))
sys.path.append(os.path.join(pwd, "../"))
from openai import OpenAI
from project_settings import environment, project_path
def get_args():
parser = argparse.ArgumentParser()
parser.add_argument(
"--model_name",
default="qwen3-max-2025-09-23",
# default="qwen3-max-preview",
# default="qwen-plus-2025-12-01",
type=str
)
parser.add_argument(
"--eval_dataset_name",
default="agent-lingoace-zh-400-choice.jsonl",
type=str
)
parser.add_argument(
"--eval_dataset_dir",
default=(project_path / "data/dataset").as_posix(),
type=str
)
parser.add_argument(
"--eval_data_dir",
default=(project_path / "data/eval_data").as_posix(),
type=str
)
parser.add_argument(
"--client",
default="shenzhen_sase",
type=str
)
parser.add_argument(
"--service",
default="aliyun_api_key",
type=str
)
parser.add_argument(
"--create_time_str",
default="null",
# default="20250812_092418",
type=str
)
parser.add_argument(
"--interval",
default=1,
type=int
)
args = parser.parse_args()
return args
def main():
args = get_args()
eval_dataset_dir = Path(args.eval_dataset_dir)
eval_dataset_dir.mkdir(parents=True, exist_ok=True)
eval_data_dir = Path(args.eval_data_dir)
eval_data_dir.mkdir(parents=True, exist_ok=True)
if args.create_time_str == "null":
tz = ZoneInfo("Asia/Shanghai")
now = datetime.now(tz)
create_time_str = now.strftime("%Y%m%d_%H%M%S")
# create_time_str = "20250724_090615"
else:
create_time_str = args.create_time_str
eval_dataset = eval_dataset_dir / args.eval_dataset_name
model_name_ = args.model_name.replace("/", "#")
output_file = eval_data_dir / f"aliyun_choice/aliyun/{model_name_}/{args.client}/{args.service}/{create_time_str}/{args.eval_dataset_name}"
output_file.parent.mkdir(parents=True, exist_ok=True)
api_key = environment.get(args.service, dtype=str)
client = OpenAI(
base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
# Read your Ark API Key from the environment variable.
api_key=api_key
)
total = 0
total_correct = 0
# finished
finished_idx_set = set()
if os.path.exists(output_file.as_posix()):
with open(output_file.as_posix(), "r", encoding="utf-8") as f:
for row in f:
row = json.loads(row)
idx = row["idx"]
total = row["total"]
total_correct = row["total_correct"]
finished_idx_set.add(idx)
print(f"finished count: {len(finished_idx_set)}")
with open(eval_dataset.as_posix(), "r", encoding="utf-8") as fin, open(output_file.as_posix(), "a+", encoding="utf-8") as fout:
for row in fin:
row = json.loads(row)
idx = row["idx"]
prompt = row["prompt"]
response = row["response"]
if idx in finished_idx_set:
continue
finished_idx_set.add(idx)
try:
time.sleep(args.interval)
print(f"sleep: {args.interval}")
time_begin = time.time()
completion = client.chat.completions.create(
model=args.model_name,
messages=[
{"role": "user", "content": prompt},
],
# 由于 enable_thinking 非 OpenAI 标准参数，需要通过 extra_body 传入
extra_body={"enable_thinking": False},
stream=False,
)
time_cost = time.time() - time_begin
print(f"time_cost: {time_cost}")
except Exception as e:
print(f"request failed, error type: {type(e)}, error text: {str(e)}")
continue
# print(f"completion: {completion}")
prediction = completion.choices[0].message.content
rid = completion.id
correct = 1 if prediction == response else 0
total += 1
total_correct += correct
score = total_correct / total
row_ = {
"idx": idx,
"rid": rid,
"prompt": prompt,
"response": response,
"prediction": prediction,
"correct": correct,
"total": total,
"total_correct": total_correct,
"score": score,
"time_cost": time_cost,
}
row_ = json.dumps(row_, ensure_ascii=False)
fout.write(f"{row_}\n")
fout.flush()
return
if __name__ == "__main__":
main()
|