Rucci-2053183's workspace
Runs
7
Name
3 visualized
State
Notes
User
Tags
Created
Runtime
Sweep
current_seq
decoder_output_dim
description
enable_gs_lr_scheduler
enable_on_plateau_phase2
enable_plateau_detection
enable_wandb
encoder_hidden_dim
encoder_input_dim
exp_name
global_iterations
iteration_first_timestep
iteration_others_timestep
lambda1_correction_weight
lambda2_correction_weight
lr_cam_c
lr_cam_m
lr_log_scales
lr_logit_opacities
lr_means3D
lr_rgb_colors
lr_seg_colors
lr_unnorm_rotations
plateau_threshold
profile_keys
scheduler_lr_gamma_cam_c
scheduler_lr_gamma_cam_m
scheduler_lr_gamma_log_scales
scheduler_lr_gamma_logit_opacities
scheduler_lr_gamma_means3D
scheduler_lr_gamma_rgb_colors
scheduler_lr_gamma_seg_colors
scheduler_lr_gamma_unnorm_rotations
scheduler_lr_step_size_cam_c
scheduler_lr_step_size_cam_m
scheduler_lr_step_size_log_scales
scheduler_lr_step_size_logit_opacities
scheduler_lr_step_size_means3D
scheduler_lr_step_size_rgb_colors
scheduler_lr_step_size_seg_colors
scheduler_lr_step_size_unnorm_rotations
subset_of_points
tile_size
train_file
Failed
-
rucci-2053183
2s
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Finished
-
rucci-2053183
32s
-
basketball
12
Set as 0 the reconstruction loss, use only the mse from the oprimal params
false
true
-
true
128
14
No Scheduler + Corrective Network 04
1
10000
10
1
0.5
0.0001
0.0001
0.01
0.5
0.0016
0.025
0
0.01
0.0001
-
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
10
10
10
10
10
10
10
10
100000
40
train_v25.py
Finished
-
rucci-2053183
26s
-
basketball
12
Set as 0 the reconstruction loss, use only the mse from the oprimal params
false
true
-
true
128
14
No Scheduler + Corrective Network 03
1
10000
10
0
1
0.0001
0.0001
0.01
0.5
0.0016
0.025
0
0.01
0.0001
-
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
10
10
10
10
10
10
10
10
100000
40
train_v25.py
Finished
-
rucci-2053183
1m 20s
-
basketball
12
Add a term to the loss that penalizes the difference between the optimal parameters (from the GS training with small lr) and the ones you get after applying the prediction of the model
false
true
-
true
128
14
No Scheduler + Corrective Network 02
1
10000
10
1
1
0.0001
0.0001
0.01
0.5
0.0016
0.025
0
0.01
0.0001
-
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
10
10
10
10
10
10
10
10
100000
40
train_v25.py
Finished
-
rucci-2053183
2m 31s
-
basketball
12
GS optimization + Corrective network - LambdaLR scheduler
true
-
true
true
128
13
GS optimization + Corrective network - 01
1
10000
10
-
-
0.0001
0.0001
0.01
0.5
0.0016
0.025
0
0.01
0.0001
-
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
10
10
10
10
10
10
10
10
100000
40
train_v25.py
Finished
-
rucci-2053183
1m 48s
-
basketball
12
GS optimization + Corrective network - Baseline Fixed LR
-
-
true
true
128
13
GS optimization + Corrective network - 01
1
10000
10
-
-
0.0001
0.0001
0.001
0.05
0.00016
0.0025
0
0.001
0.0001
["loss","correction_loss"]
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
100000
40
train_v25.py
Finished
-
rucci-2053183
2m 22s
-
basketball
12
GS optimization + Corrective network - Baseline Fixed LR
-
-
true
true
128
13
Baseline Fixed LR
1
10000
10
-
-
0.0001
0.0001
0.001
0.05
0.00016
0.0025
0
0.001
0.0001
["loss","correction_loss"]
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
100000
40
train_v25.py
1-7
of 7