Wang Yongpan
Wang Yongpan
ok, i get it. thanks for your reply
hi, I found that the exebench has three optimization options (O0, O3, Os). How can I evaluate your tool on different options (O0, O1, O2, O3) similar to your experiments?
OK, how can i obtain the source code of exebench? i can not find them from huggingface.
I can not find the source code from this link. I just found the below files: train_not_compilable: 2.357M train_synth_compilable: 2.308373M train_real_compilable: 0.675074M train_synth_simple_io: 0.550116M train_real_simple_io: 0.043769M train_synth_rich_io: 0.097250M valid_synth: 5k...
ok, I noticed that your paper tested the re-executability rate of the exebench dataset. Can I ask how you achieved it?
And I found that if i use option **-fno-builtin**, i can get your results. Did you use this to get the results?