Skip to main content
mooey5775
Projects
mixed_media
Reports
Overfit-to-1-Demo Test for SegLess TFN

Overfit-to-1-Demo Test for SegLess TFN

Daniel Seita
Created on October 10|Last edited on October 11
Getting back to reports now to make it possibly a bit easier to digest.
Here, we're trying SEGLESS_VANILLA and SEGLESS_WEIGHTED_CONSISTENCY and seeing if they can overfit to one demonstration. I think they can do it. They seem to be all driving their losses to 0.
First, log-scale. Then log-scale.

Train Curves


0100200300400500Step0.0010.010.1
BC04_MMOneSphere_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_4DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_03_56_0001 segless weighted consistency
BC04_PourWater_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_3DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_03_17_0001 segless weighted consistency
BC04_PourWater6D_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_13_47_01_0001 segless weighted consistency
BC04_MMOneSphere_v02_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_13_43_01_0001 segless weighted consistency
BC04_PourWater6D_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_35_04_0001 segless vanilla
BC04_PourWater_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_3DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_34_34_0001 segless vanilla
BC04_MMOneSphere_v02_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_20_00_0001 segless vanilla
BC04_MMOneSphere_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_4DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_19_18_0001 segless vanilla
0100200300400500Step0.0010.010.1
BC04_MMOneSphere_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_4DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_03_56_0001 segless weighted consistency
BC04_PourWater_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_3DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_03_17_0001 segless weighted consistency
BC04_PourWater6D_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_13_47_01_0001 segless weighted consistency
BC04_MMOneSphere_v02_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_13_43_01_0001 segless weighted consistency
BC04_PourWater6D_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_35_04_0001 segless vanilla
BC04_PourWater_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_3DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_34_34_0001 segless vanilla
BC04_MMOneSphere_v02_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_6DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_20_00_0001 segless vanilla
BC04_MMOneSphere_v01_ntrain_0001_PCL_PNet2_svd_pointwise_segless_ee2flow_segless_4DoF_ar_8_hor_100_keep_0s_scalePCL_noScaleTarg_debug_2022_10_10_14_19_18_0001 segless vanilla
segless vanilla
4
segless weighted consistency
4


Created with ❤️ on Weights & Biases.

https://wandb.ai/mooey5775/mixed_media/reports/Overfit-to-1-Demo-Test-for-SegLess-TFN--VmlldzoyNzcwNTcx
Made with Weights & Biases. Sign up or log in to create reports like this one.