VLN-CE CMA_PM_DA_Aug checkpoint (mirror)

Mirror of the released CMA_PM_DA_Aug checkpoint of VLN-CE (Krantz et al., "Beyond the Nav-Graph: Vision and Language Navigation in Continuous Environments", ECCV 2020) and the R2R-VLNCE instruction vocabulary, for the arena-planners cma planner.

• Source: https://github.com/jacobkrantz/VLN-CE (commit 729d141, 2025-01-07), the CMA_PM_DA_Aug checkpoint linked from the README, and the instruction_vocab of the R2R_VLNCE_v1-3_preprocessed dataset.
• cma_pm_da_aug.pt is the upstream checkpoint re-saved as a tensor-only state_dict, loadable with torch.load(weights_only=True) into VLN-CE's CMAPolicy layout.
• License: both files derive from Matterport3D and are distributed under CC BY-NC-SA 3.0 US and the Matterport3D Terms of Use (http://kaldir.vc.in.tum.de/matterport/MP_TOS.pdf), for non-commercial academic use only. The VLN-CE code is MIT licensed (Copyright (c) 2020 Jacob Krantz, Erik Wijmans, Arjun Majumdar, Dhruv Batra, Stefan Lee).

Files

File	sha256
cma_pm_da_aug.pt	57a49d7008f3449b754fd618621918660fd1f3f8231754cb1d5d6d3d71eb88b3
r2r_vocab.json	1a289293317423f903f51ea11de72c44487cb27d9c9204eadce9489008574c26

Citation

@inproceedings{krantz_vlnce_2020,
  title={Beyond the Nav-Graph: Vision and Language Navigation in Continuous Environments},
  author={Jacob Krantz and Erik Wijmans and Arjun Majundar and Dhruv Batra and Stefan Lee},
  booktitle={European Conference on Computer Vision (ECCV)},
  year={2020}
}