allenai / spoc-robot-training

SPOC: Imitating Shortest Paths in Simulation Enables Effective Navigation and Manipulation in the Real World
https://spoc-robot.github.io/
Other
90 stars 7 forks source link

Issue with scripts/evaluate_pretrained_model.sh - Stuck at "Rate 0.000 tasks/s, ETA 5555555.556 hours" #44

Open yvfengZhong opened 2 months ago

yvfengZhong commented 2 months ago

Hello,

I've encountered an issue while trying to run the scripts/evaluate_pretrained_model.sh script as instructed in the TRAINING_README.md file.

python -m training.offline.online_eval --shuffle --eval_subset minival --output_basedir tmp_log \
 --test_augmentation --task_type ObjectNavType \
 --input_sensors raw_navigation_camera raw_manipulation_camera last_actions an_object_is_in_hand \
 nav_task_relevant_object_bbox manip_task_relevant_object_bbox nav_accurate_object_bbox manip_accurate_object_bbox \
 --house_set objaverse --wandb_logging False --num_workers 2 \
 --gpu_devices 0 --training_run_id SigLIP-ViTb-3-double-det-CHORES-S --local_checkpoint_dir $CKPT_DIR

However, the script seems to be stuck and continuously outputs the following message: Rate 0.000 tasks/s, ETA 5555555.556 hours

image

This appears to indicate that the script is not processing any tasks. I would appreciate any guidance on how to resolve this issue. Thank you for your help!

yvfengZhong commented 2 months ago

Specifically, when the code reaches https://github.com/allenai/spoc-robot-training/blob/1efeeaa8f862f7791712c42fddd7d3df7ced3d14/online_evaluation/online_evaluator_worker.py#L484, it unexpectedly jumps to https://github.com/allenai/spoc-robot-training/blob/1efeeaa8f862f7791712c42fddd7d3df7ced3d14/online_evaluation/online_evaluator.py#L531 instead of continuing to https://github.com/allenai/spoc-robot-training/blob/1efeeaa8f862f7791712c42fddd7d3df7ced3d14/online_evaluation/online_evaluator_worker.py#L503.

Additionally, the variable new_found_results remains 0 throughout, which might be related to the issue.

zzee-glitch commented 1 week ago

Hello, can you download the pre-trained model? If possible, can you share it? I can’t seem to download it here. Thank you