the results are not as expected.

[goldtimes]

in the bash
python run_slam.py configs/TUM_RGBD/rgbd_dataset_freiburg1_desk.yaml --input_path /media/kilox/PS2000/photo_slam_data/data/TUM/rgbd_dataset_freiburg1_desk --output_path output/TUM/desk

[VladimirYugay]

Thanks for your interest in our work!

Can you please report the ATE and PSNR obtained on this sequence?

[goldtimes]

here is ATE.json and PSNR
{"compared_pose_pairs": 592, "rmse": 0.062046918602192254, "mean": 0.05791550874710083, "median": 0.05207182094454765, "std": 0.022262418642640114, "min": 0.0, "max": 0.0988922268152237}

[VladimirYugay]

ATE seems normal (we report ATE aligned). What is the average PSNR and depth l1?

[goldtimes]

do you mean rendering_metrics.json file?
here it's contents
{"psnr": 24.996692822847752, "lpips": 0.14584543960804194, "ssim": 0.9370739935450472, "depth_l1_train_view": 0.18504035514142356, "num_renders": 519}

[VladimirYugay]

Yes. I think the results reported in Table 2 accurately match your results. So I'd say that's the expected result you got!