search

MrZihan / GridMM

Official implementation of GridMM: Grid Memory Map for Vision-and-Language Navigation (ICCV'23).
60 stars 1 forks source link

pretraining on R2R-CE #10

Closed Bowen-sdu closed 6 months ago

Bowen-sdu commented 6 months ago

How to perform pretraining on R2R-CE？

MrZihan commented 6 months ago

GridMM is pre-trained on the R2R dataset (Matterport3D simulator) and then fine-tuned on the R2R-CE (Habitat simulator). Perhaps there are some differences between the images in these two datasets, but the instructions in the R2R and R2R-CE are consistent.

Bowen-sdu commented 6 months ago

Thank you very much for your reply. It has been very helpful to me.

GridMM is pre-trained on the R2R dataset (Matterport3D simulator) and then fine-tuned on the R2R-CE (Habitat simulator). Perhaps there are some differences between the images in these two datasets, but the instructions in the R2R and R2R-CE are consistent.