Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Question about the MUJOCO PUSH accuracy #43

Open
shentt67 opened this issue Nov 25, 2024 · 0 comments
Open

Question about the MUJOCO PUSH accuracy #43

shentt67 opened this issue Nov 25, 2024 · 0 comments

Comments

@shentt67
Copy link

Thanks for your excellent work! I have questions about the results in Table 17. Is there a scientific notation unit missed since my reproduction results are about 0.0023 in MSE?
image

Just like in Table 18, where the scientific notation is provided:
image

Thanks for your time. Here are my test logs:
Training Time: 616.6786675453186
Training Peak Mem: 22735.2109375
Training Params: 1904986
mse: 0.002326280577108264
Inference Time: 2.2704453468322754
Inference Params: 1904986
Testing on noisy data (image)...
10%|█ | 1/10 [00:02<00:19, 2.18s/it]mse: 0.002326280577108264
mse: 1.2753043174743652
30%|███ | 3/10 [00:06<00:15, 2.21s/it]mse: 1.7230396270751953
mse: 2.0257174968719482
50%|█████ | 5/10 [00:11<00:11, 2.22s/it]mse: 2.305746078491211
60%|██████ | 6/10 [00:13<00:08, 2.22s/it]mse: 2.6151041984558105
70%|███████ | 7/10 [00:15<00:06, 2.22s/it]mse: 2.7902309894561768
mse: 2.8953778743743896
90%|█████████ | 9/10 [00:19<00:02, 2.23s/it]mse: 2.9822394847869873
mse: 3.0394344329833984
relative robustness (image, MSE): 0.0
effective robustness (image, MSE): 0.0
100%|██████████| 10/10 [00:22<00:00, 2.22s/it]
0%| | 0/10 [00:00<?, ?it/s]Plot saved as My method-gentle push image-image-MSE
Testing on noisy data (proprio)...
mse: 0.002326280577108264
20%|██ | 2/10 [00:04<00:19, 2.40s/it]mse: 0.07984583079814911
30%|███ | 3/10 [00:07<00:16, 2.39s/it]mse: 0.13955439627170563
40%|████ | 4/10 [00:09<00:14, 2.39s/it]mse: 0.20418862998485565
50%|█████ | 5/10 [00:11<00:11, 2.38s/it]mse: 0.26611292362213135
60%|██████ | 6/10 [00:14<00:09, 2.38s/it]mse: 0.32232439517974854
70%|███████ | 7/10 [00:16<00:07, 2.38s/it]mse: 0.3833681344985962
mse: 0.4626495838165283
90%|█████████ | 9/10 [00:21<00:02, 2.38s/it]mse: 0.5830950736999512
100%|██████████| 10/10 [00:23<00:00, 2.38s/it]
mse: 0.7971442341804504
relative robustness (proprio, MSE): 0.0
effective robustness (proprio, MSE): 0.0
0%| | 0/10 [00:00<?, ?it/s]Plot saved as My method-gentle push proprio-proprio-MSE
Testing on noisy data (haptics)...
10%|█ | 1/10 [00:02<00:21, 2.37s/it]mse: 0.002326280577108264
mse: 0.0036232746206223965
30%|███ | 3/10 [00:07<00:16, 2.38s/it]mse: 0.0056821503676474094
40%|████ | 4/10 [00:09<00:14, 2.37s/it]mse: 0.0069841258227825165
50%|█████ | 5/10 [00:11<00:11, 2.37s/it]mse: 0.007524526212364435
60%|██████ | 6/10 [00:14<00:09, 2.37s/it]mse: 0.007330612279474735
70%|███████ | 7/10 [00:16<00:07, 2.38s/it]mse: 0.006867253687232733
mse: 0.006039843894541264
90%|█████████ | 9/10 [00:21<00:02, 2.37s/it]mse: 0.005186266731470823
100%|██████████| 10/10 [00:23<00:00, 2.38s/it]
mse: 0.0038046431727707386
relative robustness (haptics, MSE): 0.0
effective robustness (haptics, MSE): 0.0
0%| | 0/10 [00:00<?, ?it/s]Plot saved as My method-gentle push haptics-haptics-MSE
Testing on noisy data (controls)...
mse: 0.002326280577108264
20%|██ | 2/10 [00:04<00:19, 2.38s/it]mse: 0.06037278473377228
mse: 0.056742727756500244
40%|████ | 4/10 [00:09<00:14, 2.38s/it]mse: 0.0527164526283741
mse: 0.04934581369161606
60%|██████ | 6/10 [00:14<00:09, 2.38s/it]mse: 0.046973370015621185
70%|███████ | 7/10 [00:16<00:07, 2.37s/it]mse: 0.04735881835222244
mse: 0.04990621283650398
90%|█████████ | 9/10 [00:21<00:02, 2.38s/it]mse: 0.060220394283533096
mse: 0.0837954431772232
relative robustness (controls, MSE): 0.0
effective robustness (controls, MSE): 0.0
100%|██████████| 10/10 [00:23<00:00, 2.38s/it]
0%| | 0/10 [00:00<?, ?it/s]Plot saved as My method-gentle push controls-controls-MSE
Testing on noisy data (multimodal)...
mse: 0.002326280577108264
20%|██ | 2/10 [00:04<00:18, 2.37s/it]mse: 0.40776482224464417
mse: 0.629368782043457
40%|████ | 4/10 [00:09<00:14, 2.38s/it]mse: 0.7287225127220154
mse: 0.7674158811569214
60%|██████ | 6/10 [00:14<00:09, 2.37s/it]mse: 0.8007391691207886
mse: 0.8004142045974731
80%|████████ | 8/10 [00:18<00:04, 2.37s/it]mse: 0.8466270565986633
90%|█████████ | 9/10 [00:21<00:02, 2.37s/it]mse: 0.9301233291625977
100%|██████████| 10/10 [00:23<00:00, 2.37s/it]
mse: 1.0777761936187744
relative robustness (multimodal, MSE): 0.0
effective robustness (multimodal, MSE): 0.0
Plot saved as My method-gentle push multimodal-multimodal-MSE

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant