Request to provide the missing test dataset.

[wyt2000]

Hello! Thank you for your excellent work. I have successfully reproduced PlanSearch's performance on MBPP+. However, I noticed that the test datasets LiveCodeBench and HumanEval+ in your format are missing on HuggingFace. Could you kindly provide the dataset files codegenning/F_livecodebench_lite_v3 (codegenning/B_livecodebench_lite_v3) and codegenning/F_human_eval_plus (codegenning/B_human_eval_plus_v2)? I truly appreciate your assistance!

[YerbaPage]

Hi, same question here — I’m also looking for the test datasets, as mentioned above.

Would it be possible to provide them? If it’s not possible to share the datasets directly, could you kindly provide some guidance or scripts on how to generate them?

Also, @wyt2000 did you manage to get these datasets or figure out a workaround? Any tips would be greatly appreciated. Thank you!

[Yijihuashi]

Hi, I’m also trying to reproduce your results and would really appreciate access to the missing test datasets codegenning/F_livecodebench_lite_v3, codegenning/B_livecodebench_lite_v3, codegenning/F_human_eval_plus, and codegenning/B_human_eval_plus_v2.

It would be super helpful if you could share the files or provide guidance on how to generate them. Thanks again for your great work!