search

Ram81 / habitat-imitation-baselines

Code for training embodied agents using imitation learning at scale in Habitat-Lab
MIT License
34 stars 7 forks source link

about the model weight on hm3d #14

Closed KarelZhang closed 1 year ago

KarelZhang commented 1 year ago

hi @Ram81 , Do you have the model weight based on hm3d dataset? If you do, can you share with us? Thanks!

Ram81 commented 1 year ago

Hi @KarelZhang , you can look at PIRLNav. It has imitation and RL finetuning results on HM3D. The policy architecture used it different but you can use the data provided as part of PIRLNav release to train policy architecture used in Habitat-Web

Ram81 commented 1 year ago

Closing this issue as it is resolved