microsoft / SwinBERT

Research code for CVPR 2022 paper "SwinBERT: End-to-End Transformers with Sparse Attention for Video Captioning"
https://arxiv.org/abs/2111.13196
MIT License
237 stars 35 forks source link

One data typo #17

Open RyanLiut opened 2 years ago

RyanLiut commented 2 years ago

Hi,

The data reported on MSRVTT open-book (red mark) was wrong (it seems to be mixed with VATEX open-book). According to openbook paper (The second pic), B4, M, R should be 42.8, 29.3, and 61.7. image


Paper: open-book image

kevinlin311tw commented 2 years ago

Sorry for the confusion. We will fix the typo and update our arxiv paper accordingly. Thank you very much for pointing out the typo.