PAct x SceneTransporter OT Ablation

27Stage1 runs

10Stage2 runs

23OT diagnostic runs

37total parsed logs

路线状态

Phase 1: Stage1 cross-attention can be patched by OT-gated routing without changing the official checkpoint format.
Phase 2: 2D masks are now soft OT priors and edge maps enter the cost, so segmentation errors no longer monopolize part assignment.
Phase 3: A virtual patch channel gives occluded or weakly visible parts an explicit route to learned prior information.
Phase 4: Stage2 now exposes spatial mask moments and a zero-initialized articulation residual adapter for joint prediction.

可视化

消融表

Run	Phase	Steps	Loss	Art L2	Gate	Edge delta	Virtual mass	Mask parts	Empty	Note
Stage1 baseline smoke `pact_finetune_route_32_20260516/stage1_smoke`	baseline	1	0.03043	-	-	-	-	-	-	Official-equivalent trainer, no OT router.
OT first block `pact_stage1_ot_router_first_diag_smoke_20260520`	phase1	1	0.0304	-	0.981	-	-	-	-	One Stage1 cross-attention block patched with OT gating.
OT mask prior `pact_stage1_ot_router_maskprior_smoke_20260520`	phase2	1	0.02921	-	0.2925	-	-	-	-	Mask demoted from hard condition to OT cost prior.
OT mask prior 5-step `pact_stage1_ot_router_maskprior_5step_sgd1e8_20260520`	phase2	5	0.04972	-	0.2306	-	-	-	-	SGD stability check for soft mask prior.
OT RGB edge `pact_stage1_ot_router_rgb_edge_1step_20260520`	phase2	1	0.02921	-	0.2925	0.001399	-	-	-	RGB Sobel edge regularized cost.
OT edge 5-step `pact_stage1_ot_router_edge_fixed_5step_20260520`	phase2	5	0.04973	-	0.231	0.001319	-	-	-	Mask-edge regularized 5-step stability check.
OT virtual patch `pact_stage1_ot_router_virtual005_rgb_fixedgate_5step_20260520`	phase3	5	0.03779	-	0.2607	0.00119	0.04999	-	-	Virtual image token with visible/virtual gate normalization.
OT first-third `pact_stage1_ot_router_firstthird_1step_20260520`	phase1	1	0.02923	-	0.2901	-	-	-	-	Patch first third of Stage1 cross-attention blocks.
OT edge first-third `pact_stage1_ot_router_edge_firstthird_fixed_1step_20260520`	phase2	1	0.02924	-	0.2905	0.001509	-	-	-	First-third patch plus edge-regularized cost.
Stage2 baseline smoke `pact_finetune_route_32_20260516/stage2_smoke`	baseline	1	3.2046	0.09492	-	-	-	-	-	Official-equivalent Stage2 smoke.
Stage2 spatial diag `pact_stage2_spatial_diag_smoke_20260521`	phase4	1	2.3856	0.1107	-	-	-	5.0	1.0	Mask-derived spatial moment diagnostics for articulation rows.
Stage2 spatial adapter 1-step `pact_stage2_spatial_adapter_smoke_20260521`	phase4	1	7.4112	0.1336	-	-	-	3.0	1.0	Zero-initialized articulation residual adapter smoke.
Stage2 spatial adapter 5-step `pact_stage2_spatial_adapter_5step_20260521`	phase4	5	3.3670	0.1141	-	-	-	6.0	2.0	Spatial adapter 5-step SGD stability check.
Matrix Stage1 baseline `pact_ot_matrix_20260521_stage1_baseline`	matrix	8	1.0631	-	-	-	-	-	-	Quick matrix baseline with the official-equivalent Stage1 path.
Matrix Stage1 OT first `pact_ot_matrix_20260521_stage1_ot_first`	matrix	8	1.0631	-	0.9751	0	0	-	-	Quick matrix with one OT-gated cross-attention block.
Matrix Stage1 mask prior `pact_ot_matrix_20260521_stage1_mask_prior`	matrix	8	1.0631	-	0.3237	0	0	-	-	Quick matrix with mask labels as a soft OT prior.
Matrix Stage1 RGB edge `pact_ot_matrix_20260521_stage1_rgb_edge`	matrix	8	1.0631	-	0.3237	0.001583	0	-	-	Quick matrix with RGB Sobel edge-regularized OT cost.
Matrix Stage1 virtual patch `pact_ot_matrix_20260521_stage1_virtual_patch`	matrix	8	1.0334	-	0.1462	0.001448	0.04999	-	-	Quick matrix with a virtual patch channel.
Matrix Stage1 first-third edge `pact_ot_matrix_20260521_stage1_firstthird_edge`	matrix	3	1.0303	-	0.3603	0.001898	0	-	-	Quick matrix with OT injected into the first third of Stage1 blocks.
Matrix Stage2 spatial diag `pact_ot_matrix_20260521_stage2_spatial_diag`	matrix	5	3.4941	0.2005	-	-	-	4.0	2.0	Quick matrix Stage2 spatial diagnostics.
Matrix Stage2 spatial adapter `pact_ot_matrix_20260521_stage2_spatial_adapter`	matrix	5	3.3674	0.1141	-	-	-	6.0	2.0	Quick matrix Stage2 spatial adapter.
Full Stage1 baseline `pact_ot_matrix_20260521_full_stage1_baseline`	full	60	1.0311	-	-	-	-	-	-	Full matrix baseline with final checkpoint export.
Full Stage1 OT first `pact_ot_matrix_20260521_full_stage1_ot_first`	full	60	1.0311	-	0.9707	0	0	-	-	Full matrix with one OT-gated cross-attention block.
Full Stage1 mask prior `pact_ot_matrix_20260521_full_stage1_mask_prior`	full	60	1.0311	-	0.271	0	0	-	-	Full matrix with mask labels as a soft OT prior.
Full Stage1 RGB edge `pact_ot_matrix_20260521_full_stage1_rgb_edge`	full	60	1.0311	-	0.2712	0.001356	0	-	-	Full matrix with RGB Sobel edge-regularized OT cost.
Full Stage1 virtual patch `pact_ot_matrix_20260521_full_stage1_virtual_patch`	full	60	1.0240	-	0.29	0.001643	0.04999	-	-	Full matrix with a virtual patch channel.
Full Stage1 first-third edge `pact_ot_matrix_20260521_full_stage1_firstthird_edge`	full	20	1.0325	-	0.1809	0.002062	0	-	-	Full matrix with OT injected into the first third of Stage1 blocks.
Full Stage2 spatial diag `pact_ot_matrix_20260521_full_stage2_spatial_diag`	full	40	1.5619	0.06205	-	-	-	10.0	3.0	Full matrix Stage2 spatial diagnostics.
Full Stage2 spatial adapter `pact_ot_matrix_20260521_full_stage2_spatial_adapter`	full	40	5.9822	0.157	-	-	-	6.0	1.0	Full matrix Stage2 spatial adapter.
Official full Stage1 baseline `pact_ot_matrix_official_20260521_full_stage1_baseline`	official-full	60	0.04795	-	-	-	-	-	-	Official-initialized full matrix baseline with final checkpoint export.
Official full Stage1 OT first `pact_ot_matrix_official_20260521_full_stage1_ot_first`	official-full	60	0.04791	-	0.9351	0	0	-	-	Official-initialized full matrix with one OT-gated cross-attention block.
Official full Stage1 mask prior `pact_ot_matrix_official_20260521_full_stage1_mask_prior`	official-full	60	0.04789	-	0.2683	0	0	-	-	Official-initialized full matrix with mask labels as a soft OT prior.
Official full Stage1 RGB edge `pact_ot_matrix_official_20260521_full_stage1_rgb_edge`	official-full	60	0.0479	-	0.2686	0.001104	0	-	-	Official-initialized full matrix with RGB Sobel edge-regularized OT cost.
Official full Stage1 virtual patch `pact_ot_matrix_official_20260521_full_stage1_virtual_patch`	official-full	60	0.05485	-	0.2864	0.0009899	0.04999	-	-	Official-initialized full matrix with a virtual patch channel.
Official full Stage1 first-third edge `pact_ot_matrix_official_20260521_full_stage1_firstthird_edge`	official-full	20	0.04858	-	0.1845	0.001843	0	-	-	Official-initialized full matrix with OT injected into the first third of Stage1 blocks.
Official full Stage2 spatial diag `pact_ot_matrix_official_20260521_full_stage2_spatial_diag`	official-full	40	1.5618	0.06206	-	-	-	10.0	3.0	Official-initialized full matrix Stage2 spatial diagnostics.
Official full Stage2 spatial adapter `pact_ot_matrix_official_20260521_full_stage2_spatial_adapter`	official-full	40	5.9811	0.157	-	-	-	6.0	1.0	Official-initialized full matrix Stage2 spatial adapter.

文件

ablation_summary.csv · ablation_summary.json