Reproducibility gap

[escorciav]

I managed to reproduce all the testing/inference results. My results are below:

There is a discrepancy of 0.1%, 15.48% v.s. 15.6% claimed in the README. That looks reasonable to me. Could you please confirm so?

Namespace(gt_file='../../../../preprocess/caption_gt_test.json', pred_file='../sim_iter_5000.p', recall=[1, 5, 10], tiou=[0.1, 0.3, 0.5, 0.7])

tiou@0.1 :  [1, 5, 10]
0.639247311828
0.99247311828
0.99623655914

tiou@0.3 :  [1, 5, 10]
0.513709677419
0.948924731183
0.989784946237

tiou@0.5 :  [1, 5, 10]
0.337096774194
0.764784946237
0.922043010753

tiou@0.7 :  [1, 5, 10]
0.154838709677
0.447043010753
0.618279569892