Leaderboard [ERROR] Evaluation failed. No available key left!

[ymxyll]

[ERROR] Evaluation failed. No available key left! The team name in your submission is [xxx].

Thank you for providing such a convenient testing service! After I submitted submission.json to the test server, the following error occurred. Does this mean that the GPT key on the test server has run out of balance?

[ChonghaoSima]

Some of your answers contain too many tokens, please check your json file size before submitting.

Error message: This model's maximum context length is 16385 tokens. However, your messages resulted in 24345 tokens. Please reduce the length of the messages

[ymxyll]

Some of your answers contain too many tokens, please check your json file size before submitting.

Error message: This model's maximum context length is 16385 tokens. However, your messages resulted in 24345 tokens. Please reduce the length of the messages

Thank you for your patient response to my question. I'd like to ask how to resolve the following error, which seems to be caused by an incorrect format of the GPT response score.

[ERROR] Evaluation failed. could not convert string to float: 'The rating for your answer is: 60' The team name in your submission is [ssl-asuka].

[Leo-Yuyang]

@ymxyll Im meetting the similar error: "could not convert string to float". My last four submissions always failed because of this reason. I think the eval code used in the test server is not very robust. A simple try again logic could easily solve this problem but we cannot access the code

[ChonghaoSima]

Usually people encounter such error with inappropriate answer in the submission json file. We do have several try-catch logic in our server, meaning that your submission is constantly causing such error. My suggestion would be testing with train data on your side and see if there is similar result.