33 lines
875 B
Python
33 lines
875 B
Python
from dotenv import dotenv_values, load_dotenv
|
|
|
|
load_dotenv('.env')
|
|
env = dotenv_values('.env')
|
|
|
|
def run_test(model_name: str):
|
|
from evalscope import TaskConfig, run_task
|
|
from evalscope.constants import EvalType
|
|
task_cfg = TaskConfig(
|
|
model=model_name,
|
|
api_url='https://dashscope.aliyuncs.com/compatible-mode/v1',
|
|
api_key=env.get('DASHSCOPE_API_KEY'),
|
|
eval_type=EvalType.OPENAI_API,
|
|
datasets=[
|
|
'gsm8k',
|
|
'iquiz',
|
|
'ifeval',
|
|
'gpqa_diamond',
|
|
],
|
|
generation_config={
|
|
'temperature': 0.7,
|
|
'stream': True,
|
|
},
|
|
eval_batch_size=10,
|
|
limit=10, # 设置为10条数据进行测试
|
|
use_cache=f'examples/viz/{model_name}' # 缓存路径
|
|
)
|
|
|
|
run_task(task_cfg=task_cfg)
|
|
|
|
run_test('qwen-plus')
|
|
run_test('qwen3-max')
|