jpthu17 / DiCoSA

[IJCAI 2023] Text-Video Retrieval with Disentangled Conceptualization and Set-to-Set Alignment
Apache License 2.0
49 stars 2 forks source link

question about QB-norm inference #10

Open musicman217 opened 1 month ago

musicman217 commented 1 month ago

Hello, I found that you used QB-Norm postprocessing in inference stage while there is no mention about qb-norm in paper, can you show the result without qb-norm? thank you for your work!