比较两张图片的相似程度是否超过给定阈值。 参数: img1: PIL.Image.Image - 第一张图片 img2: PIL.Image.Image - 第二张图片 threshold: float - 相似度阈值，默认是0.9（即90%）

比较两张图片的尺寸比例是否在允许的误差范围内。

Compare two resources based on request rate, token usage and price Args: other (Resource): The other resource to compare with

build llm from config Args: llm_type (str): llm type, only support openai config (Dict[str, Any]): the dict of llm config R

根据task_id匹配并运行相应的单元测试函数。

extract code from assistant content Args: rsp (str): the response content from assistant code_start_token (str, optional): the to

接受用户传入的messages和tools[可选]，返回新的message和debug_info Args: messages: chatml messages tools = openai tools description

检查图像是否是灰度图。

parse output from code interpreter Args: output (str): the output from code interpreter mode: the mode of the output, could be pr

把传入的ToolProperty转换为OpenAI接口需要的dict格式 Args: property (ToolProperty): ToolProperty对话 Returns: Dict: 参数的dict形式描述

Get a client from the pool and return it as a context manager.

Get a client from the pool and return it as a context manager.

Get a client from the pool and return it as a context manager.

Get a client from the pool and return it as a context manager.

Put the resource back in the pool.

Put the resource back in the pool.

Put the resource back in the pool.

Put the resource back in the pool.

Compare two resources based on request rate and token usage Args: other (Resource): The other resource to compare with Ret

判断两张图片是否相似，使用ORB特征匹配。 :param image_path1: 第一张图片的文件路径 :param image_path2: 第二张图片的文件路径 :param threshold: 匹配特征点的阈值，默认为0.75 :return:

根据task_id，判断是否需要输出文件， 如果需要，则检测输出文件是否在output文件夹内，模糊匹配，返回匹配的文件名称 如果没有输出文件，则直接failed 如果不需要，这一环节直接pass 好像不是很必要