Hi PPGeo team, thanks for the well-documented repo! I'm curious about the training curves for the fine-tuning stage of the model on the NuScenes dataset:
- val_loss_l1 vs step
- val_loss_l2_1s vs step
- val_loss_l2_2s vs step
- val_loss_l2 vs step
|
self.log('val_loss_l1', loss_l1.item(), sync_dist=True) |
|
self.log('val_loss_l2_1s', loss_l2_1s.item(), sync_dist=True) |
|
self.log('val_loss_l2_2s', loss_l2_2s.item(), sync_dist=True) |
|
self.log('val_loss_l2', loss_l2.item(), sync_dist=True) |
The paper doesn't say where to find them. Will your team be releasing this information? I appreciate your time and consideration.
Hi PPGeo team, thanks for the well-documented repo! I'm curious about the training curves for the fine-tuning stage of the model on the NuScenes dataset:
PPGeo/nuscenes_planning/train_planning.py
Lines 75 to 78 in c5e684a
The paper doesn't say where to find them. Will your team be releasing this information? I appreciate your time and consideration.