How to efficiently create, maintain, and filter examples in Langsmith's dataset

[MillionWang0724]

During use, it was found that the management tool for examples is not very user-friendly. Here are some issues that need to be addressed:

1. Difficulty in adding new use cases
   Adding and updating test cases is troublesome. When the test set is first created, it is allowed to upload CSV test case files in bulk. After creation, adding new test cases requires manual addition of individual cases
2. Poor interaction experience and lack of filtering and other abilities
   The interaction interface of use case management is not ideal, and there are certain difficulties in querying, locating, and filtering use cases, which can only be searched page by page, resulting in a poor experience
3. It is difficult to observe and create when there are many parameters
   Langsmith only has input and output columns for the parameters of the test data, but in reality, input contains a lot of content. When there are many fields that need to be maintained, the observation of test cases is not intuitive, and creating them is also quite cumbersome.
4. Troubles during batch maintenance
   When a common field in the set of use cases is updated in bulk, updating it on Langsmith is very cumbersome. It requires finding some use cases at once, finding the corresponding fields, and modifying them one by one, which may easily lead to omissions.