Releasing model outputs

[peteanderson80]

Hi Jialu,

Great paper!

I was wondering if you would consider sharing the model outputs (predicted paths) for RxR val-unseen and R2R val-unseen? This would be useful to dig into the performance in more detail. Sharing the outputs of your agent saves some time and avoids randomness compared to re-training the agent from scratch.

Cheers,
Peter

[jialuli-luka]

Hi Peter,

Thanks for your interest in EnvEdit! You can access the model outputs from this link. These outputs are from the models we use to submit to R2R and RxR leaderboard. Please let me know if you need trained model/model outputs of other models.

Best,
Jialu

[peteanderson80]

Thanks Jialu, this is super helpful!