Why did you set config.length to 50?

[JunRepos]

Upon reviewing the evaluation.py file, I noticed that only submissions from index 0 to config.length are considered. In cases where the number of submissions increases, wouldn’t the performance measurement for each question become imbalanced due to the exclusion of later submissions? I also observed in the dataset that there are frequent instances where the submission length exceeds 50, so I wanted to inquire about this.

[YangAzure]

I only limited the valid tracing to the first 50 submissions due to computational resource limit -- feel free to make this number bigger and see if it works for more later submissions.