diff --git a/.gitattributes b/.gitattributes
index f018faef677751be25afe81b044b3a8da6a6ad3b..9aa6808d93f98f37c43c7da626df439a58d5e103 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -11352,3 +11352,9 @@ anchorgen_baselines_20260923/input_cond/results/smoke_009_661_side/review_009_66
 anchorgen_baselines_20260923/input_cond/results/smoke_009_661_side/review_009_661/opt_drag_5pct.obj filter=lfs diff=lfs merge=lfs -text
 anchorgen_baselines_20260923/input_cond/results/smoke_009_661_side/review_009_661/source_009_R512.obj filter=lfs diff=lfs merge=lfs -text
 anchorgen_baselines_20260923/input_cond/results/smoke_009_661_side/review_009_661/target_661_R512.obj filter=lfs diff=lfs merge=lfs -text
+anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/run-1jk61ovk.wandb filter=lfs diff=lfs merge=lfs -text
+anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/run-e6xhvf8a.wandb filter=lfs diff=lfs merge=lfs -text
+anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/run-42v9byh1.wandb filter=lfs diff=lfs merge=lfs -text
+anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/run-xp14m7ij.wandb filter=lfs diff=lfs merge=lfs -text
+anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/run-3zi7cjs8.wandb filter=lfs diff=lfs merge=lfs -text
+anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/run-3mjkolf3.wandb filter=lfs diff=lfs merge=lfs -text
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_022__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_022__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..67d38185e80ad63b7251b0ad516457f56dcd9919
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_022__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32ef9cdbf9d9fc6c2a13baec87c33960eec2d93c31e7dcc9e5c3b559215eaa40
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_550__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_550__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1e4e036307b673ee21d75a1d20d942d8c83cc4cc
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_550__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1162e0482859118225aec51649d7aa430aef0f784676a054919d10551ada8809
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_681__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_681__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0e4774749037b7d16d03905eca40b1c33a13cef2
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_141__E_S_WW_WM_681__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1f76beb86a7460b7fa17f027580b29daec9d13e611a2684bca7745683e12834
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_033__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_033__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..069e8b675d0f426245dd2fd0615a4f03d9fe3e83
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_033__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cff989ca3b80568f6275c1cadf0ba7dfd367fdae46e654ce7cb2dcc8a076081
+size 9813
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_134__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_134__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..49a7c58a0833611abe77646b87659842227a9754
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_134__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a1df9ad9faa0c1fd331833c0cffef169e2cf937ed984a99966700db3366c609
+size 9813
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_444__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_444__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9a5754762358f1571ede902025c4ac3f7954011c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WWC_WM_444__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ef9f07753327a9e9f5bc69f4ec5afa7f1731e4c841670b8e5c0ed782e1300b5
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WW_WM_080__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WW_WM_080__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c3b204882f4566c20e1a666c663a7be9c77f68c0
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__E_S_WW_WM_080__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e2407791569675a5284e68aed75d4b7378dcad64f17692005ee55fca10f8c32
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__F_S_WWC_WM_631__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__F_S_WWC_WM_631__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c48581d1d8dd03dcbd1b4f17b55288f334086382
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__F_S_WWC_WM_631__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ceedbf8db9f678a17ac9df113a4e1095a559f7600f56ae68495adab20e89a576
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__N_S_WW_WM_441__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__N_S_WW_WM_441__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..367bc635e4d60a224dadb018bb76a39c49ac9567
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_211__N_S_WW_WM_441__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f4e77bb4076c3346414a3b84b67a0bf5ae4c666ec7ad8f065b8c64d896a4ebd
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WWC_WM_450__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WWC_WM_450__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5be0492a042ba4ba613ff37cc6afffd842cd21e7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WWC_WM_450__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29ec8ce33c89fc09e6afc96778da4224be8a644d41322cf89eb36118e9f79051
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WWC_WM_662__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WWC_WM_662__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7c05d9bdeff41c9d667e2ab703221176a906dedc
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WWC_WM_662__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb0ff4395c02f18124b6d557b0c5ed882a1c50fdf5c81bf85c44703f5f2b8f36
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_190__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_190__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3d409666bebbb8b5cc4d575e80677cb5723bb4d9
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_190__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df7b64ae73d3ee0d4407a602843e7ef7ee236c59f879787df8dab7c7eee3a35d
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_563__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_563__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..cf26495168160def24a389970195b70de3c15983
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_563__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2006dd0c117a692646e68999d6dae1edb538088a870d9cca1f991c6f7473c20b
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_583__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_583__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..481179099e1c0aa3b277e76f08eb1c802dc77a4d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__E_S_WW_WM_583__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:660db1ab235c42aad3aaf74db3b9b1e0910aee257f97d54fa581db5bd013ddf9
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__F_S_WWC_WM_486__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__F_S_WWC_WM_486__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3a58e49464bd6ee8843c73a554e0f4a6e9ccce10
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_330__F_S_WWC_WM_486__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:305c9601164aa26b3ca76cf8e06a9307002137198be693c5e94bc67ee84e2cb0
+size 9813
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WWC_WM_267__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WWC_WM_267__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..39d383b79fc41f065dbabb119e10df39c578f1ba
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WWC_WM_267__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2172893cd4b43f8ca23d6f932cf5fcdfffd6980007cd6098bf37dfdb43fdd9e2
+size 9813
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_022__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_022__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fcbdd535553befdc8271093eb771daffd0464139
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_022__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e7aa937a320aaf7a5dd663ce613bf506973f9532a04c5890d3dff035c58252f
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_072__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_072__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b72d08d9ec9c545c2efb37ce46d1f0afc9befe5b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_072__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8be9e17fb0875ce7f699fa4b8736baac2592b8f2a345e0b40c65bd380c532cb4
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_170__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_170__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..645033a4bce412903689a46dfcad2ea47d510edb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_170__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:250854f34dcd84e403f9194e0e79a02ff893b8a8fe19ca37a6f4dec37160903c
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_181__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_181__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..791ebdf278adc0ddcf3e1547fbda711b824998f0
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_181__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3aa33eb3dbc6d6880c1de55987680d59588ebe57da9a22505e7ca88876b0107d
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_302__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_302__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..decce9c8becb6a18286a16e342c2a7d13983b01d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_372__E_S_WW_WM_302__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:351a4e1e8a00bec8eed278b50015043c2758577333bce2efcb96fcfa40d7e260
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WWC_WM_013__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WWC_WM_013__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..11fc7991c2d1ec96545b3f42273f02e5588671ed
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WWC_WM_013__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efa035cfd91bab91ea6bfc40734e979889fc2c2dde75ed3a4c8a4299a43ec2ec
+size 9813
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WWC_WM_272__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WWC_WM_272__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1ed1f300168ad11ef0dc341209625ead8d49ac6d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WWC_WM_272__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a55b6bb2c1556d17cd6a6af0d28813ce874bb3e93b6599956880de52a310c31
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WW_WM_204__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WW_WM_204__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2bce4901936e24b3353b93663777f18ef6299304
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WW_WM_204__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff2204191814675a82f9d526a5c6e409c5b9465305e983803929a34a77f0a0d2
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WW_WM_696__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WW_WM_696__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1e974a41660ad8e0b41a7a12268edb64e41a7f49
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__E_S_WW_WM_696__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:473a769029dc643598e7c55165c4e731509be72fbc9fdd5c9229ad9e7a0d3f38
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__F_S_WWC_WM_631__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__F_S_WWC_WM_631__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f4622881182bcd51037a5d27bbbbc87f61ff5f1d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__F_S_WWC_WM_631__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e0895ee60cf2654fbd09b3e04a96087dd023bf8f149d3da329051b49cced0dc
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__N_S_WWS_WM_399__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__N_S_WWS_WM_399__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..515742e6746b4fa27385781c6fb6590408a8e294
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_430__N_S_WWS_WM_399__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9ae30f56430fa76c94e1ad9f7934ea08084a6a9d1d81f5f3b287c5a32eb6354
+size 9813
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WWC_WM_369__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WWC_WM_369__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..55a7e09d99d1b4d5777e44f802d82444f1f5c3c6
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WWC_WM_369__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd6d3446db781ba5a326a0f773ba8f802bfb757f4ed7126964dbe769e1511f7d
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_091__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_091__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6cd20fee60f9d51c546a7f054a1f686648f6b3ae
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_091__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08b819372ded54c17862e6daefe20479c1f70d2ce7b404961152dc2a5c2844ed
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_191__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_191__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..28225dab3802ed94dd987d576555f856f972f97d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_191__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c5cca7ad5bf8d4de809fb65d02313f006413c631fc99307c140e52ffe58c3ca2
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_360__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_360__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..84b45a1ef0d2816e572a9afe9ce33a29d12ccf76
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_360__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9c59d18fdabc8a4575756ff7b8d10086b97e8a48e39ccd755cf624770628aad
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_378__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_378__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..53685fd5c54d433df5059f47d54220ac0464bebd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__E_S_WW_WM_378__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c2b2712123611d800b46921d647a4ca09df8fa22578b134a73a0c8fe831d986
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__F_S_WWS_WM_536__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__F_S_WWS_WM_536__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ca381ed695880578857f3596e7c106123a845e44
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WWS_WM_461__F_S_WWS_WM_536__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c748ee2e1d6431985d11f61c7bcf5ce885f25879da57b5fcb7df3a6ad8311001
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_363__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_363__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d22bfda9af2ec1dc216b24c75287cfb473877031
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_363__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e38884099d0810be2f84fff7182693569f9bfe3c7a808d547d03783b32107821
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_563__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_563__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..53dbef9fc4ac395c9595da55ecbd5aff02bdf17d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_563__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39c8c10f2706d005bef3ec1a3efba6abae231456330a075b0c2a8d7f3b7fdd54
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_583__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_583__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..41835ffdeef8e4430e98d59753b729063f96f91c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WWC_WM_583__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4651e65a726bc93639273d72bb28966181631d5d2671222812c76a695cd7cbc2
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WW_WM_370__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WW_WM_370__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6cb95f09018e0f83ed9f629628865c9f4556b476
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WW_WM_370__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2a1fb98957e73f665d70fb0f2c4392abd421ab6f7b30052135deca7395f5ba1
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WW_WM_649__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WW_WM_649__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..45a9ac6b7e5d9e1e8ca7be728a4e57d4c351dd45
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__E_S_WW_WM_649__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71e6e8841b79c4076aa016364527a48cc937779c92485e686283b95756da5484
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__N_S_WW_WM_179__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__N_S_WW_WM_179__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e3012be2375e6d8edb9034ba030e30e025536370
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_053__N_S_WW_WM_179__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a18d5d563ece241a1e72d0accf10d2fd393a3dc2b1c18f34ab9090943e15e44f
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WWC_WM_121__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WWC_WM_121__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3d4a965090a106510f0d41146868e766d0387dbf
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WWC_WM_121__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65e5697a1b999c69324870cfe141a979f65c33f4ca9e045d6003cabe5297e1f6
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WWC_WM_440__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WWC_WM_440__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e59b4a4aa33e44a6d6b8dcc93be90b8432d0d4e7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WWC_WM_440__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8af58d01b36281378a643bba8584473fe14171491290de2bbdaf38bddb57d683
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_101__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_101__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e3fa35440b1b8b465240c77d92b3574ab7b3ce3b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_101__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6572c37076c087bd813690084bcac52af75b8c4c2c59467db165b258c4042631
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_633__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_633__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2b922fbd0a41e2b4fb5b8e3c1c5077d1547b9a9d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_633__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3525e4d6498a7caeea214b96f9378b1514e7bf668563d76fc4ce8154c22cbfc5
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_667__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_667__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..90a3afd57d9dc56d9173ab7f9eaff549e275401a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__E_S_WW_WM_667__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b00998abf871d19149a4ecce45ef2fb3d30ce9d87e491bcb402ac0de03d00573
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__N_S_WWS_WM_405__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__N_S_WWS_WM_405__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0a0136b3ac7c40e88f00a510adb164086a95f768
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_109__N_S_WWS_WM_405__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70eff192ad221c49b1ca67f386061a063d15a76129059cd15565a1dad6d12eb0
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_031__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_031__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..62bb89d7cceae214f4aa2d05a22b5648e807954d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_031__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da85c7c8ef05babd160d3ba30244dd6c99206a7183c41ad58c2a04cf339b7afd
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_105__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_105__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6a9b138a53c54c5f5d77d05fded98a4a8166a7ea
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_105__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a2f8cb532fb9f95c1e5cff9b553c1c69bff75de0aaec8e56ba8bf7e9c562424
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_181__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_181__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..672b5ce4aad3723f3c46e8af185eb1772724d3a1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_181__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10bfcd6e9175d034425dc6e650806d7f50eab18c3b4ca2dac43c9d09e6b56736
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_411__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_411__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fbfa5c8e992880c14424024451087cd97c42ae1c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_411__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddc8c27a3ad950073e23f87d17d3c244caa78a3fecec8358ef2c652c74da3016
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_416__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_416__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4d69246630dc1674ddeba8972295436d4c9af69a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__E_S_WW_WM_416__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c8b540bd8f9854fd18f437f47d85a00138dc119d7b3ba9195f8a3ce0c673b76
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__F_S_WWC_WM_231__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__F_S_WWC_WM_231__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a05a1e1564913370dc9a37fb4fadd1ccabbebcc1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_126__F_S_WWC_WM_231__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfa2b824be638d03b803199b2fa159f4a08d4515e745ad3b3ae14e3206c7f9f1
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WWC_WM_556__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WWC_WM_556__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1dcf82e691f8836ed725baec158b157280e51c65
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WWC_WM_556__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eab1dc75afba89788ec6bf990f2a5b6688a82c86ae46779ac082fbf79001653d
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WWC_WM_602__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WWC_WM_602__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1f78bd9f695428130bb8289816a3fc338d73c94a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WWC_WM_602__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcecfe87f999b5b3f8739eb71684802de91b9239e16e0bc00e04fb4741d45855
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_444__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_444__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..563c0b59d957a74ea75830073ff82635c38c64fa
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_444__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:001fe7e1f925bd6e4f3982a81c681e2cdf4f6cc76d361603ed96c048c2d45dab
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_551__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_551__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bac6dd915dc2ff702c6d378265a310d6a6d7b1a4
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_551__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ecd30dc6edc0e20fa697bf0c1a678c053800ec0c76d5049d5568f76408f6778
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_610__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_610__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..da3de9318048275670dfaf0791d5d8acb11621ce
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__E_S_WW_WM_610__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e40b3b6d993add14455dd9e4066262c97c97837d52bfeb0b14bf21b5db13e16f
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__F_S_WWC_WM_198__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__F_S_WWC_WM_198__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b248c452789679732caaade95663c2727acc54a3
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_200__F_S_WWC_WM_198__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:715d18fe553734eb03ab295e64f6272f3c3c45377ab95447d6c55ce18ebc463f
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_365__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_365__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f0dfb012cf520487ef6f4140655ddc89d547dc3d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_365__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daf7836d8fafeca3ba5c0fe08c745aa7af4fb10f678b9549b938efd5e74e1c6a
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_382__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_382__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1e41c66882da89c4ed4c8cd19dcc179c295653f6
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_382__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e0002baca83726c1a8c7abe2b7954834c05d584ff65328e37e441a33144d372
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_487__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_487__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..21ef8da2c431c99060a07ea5760651af5f62cfec
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WWC_WM_487__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28dc4518e5dc6ff40a49067ff6cb4e405ab4fd6d1f121aeb0e6a7431ce0da819
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WW_WM_422__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WW_WM_422__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d0c49affa132f64c335d8fef34c3754385cddc43
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WW_WM_422__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:563aea2d094e764d5c52ea01ecfbaadc5d78ec9027cc5555cc04e33cc0c8eb63
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WW_WM_604__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WW_WM_604__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..227179b5709510be812f832e17004e5a26260713
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__E_S_WW_WM_604__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55be88f8270091b774ff40df6026804470dd0a439709b509d2b5a8e4c255cf34
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__F_S_WWC_WM_295__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__F_S_WWC_WM_295__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..364df908c0ead19216844950c9f1096fda734e5c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_278__F_S_WWC_WM_295__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5166f5b3e60b6f33a557782c4574f08f4d3dc00b4ead9fad2d843dac03bd0eff
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_124__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_124__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b3231354354d020781580f13cf463c14e7ac8383
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_124__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f30667ee516d44a1054cbb54911746f9154ee92b8bf8f8cfab144564439f337
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_313__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_313__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0017c66963986fc36e6ad44704d558c6220c092a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_313__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abc42f4f8dc315b9782c9d28f1622845994c418475867da7684d5864ae5db0c1
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_437__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_437__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..542a63f68d69839dde8d5509658bae323b2bda06
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WWC_WM_437__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51862b3cd87ca34cb8cf584c93a1a6f7f0ace56ef8cbd0aa16c10ed4013bbea6
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_190__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_190__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d00df0c0015e6fa882e5ae073cfac581590b1508
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_190__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6f49918758ed9d9b31d209a3fdb91f744301eb9bcac0b40cd3966ba41f1a584
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_256__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_256__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d874f3bae6c3320d1b51486cc3a9806015178994
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_256__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2697b03251e6a55366cc0cdd1f449245f684f15b89b8c94719340e65dc11fce2
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_402__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_402__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7765588bddf61c32f4fcfb762790460f8deb1313
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_313__E_S_WW_WM_402__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:badd2806f9c097fa813d49e81c988212756499c1f75ab4302ba2d4851cf445cc
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WWC_WM_664__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WWC_WM_664__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..06d204af9e158c29f2f46c3b3031cef45b896d2f
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WWC_WM_664__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b2cd73a3e69bf66a81e9815dfdbf0ebb605760d5d48c67680fe2e79ddec7b3c
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_086__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_086__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d19d167e9d7029c88161dab5e01c2ce98bc90c5e
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_086__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41ce232c987a2877dea1872fe27524a06e5f8245f53e820ec7dd502f38c22ebe
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_135__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_135__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..73545602812fc4756e528db3279c431cffb40b4d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_135__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b92cf70b2a4662541c352d83d2d6e878b7fd980c822e6e0428fdc35baef4bcf
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_444__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_444__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..08729fe35ca17f41d8123b8a3cf8de2e17658acd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_444__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91b1aebf4de7e07cbeaf60725751747d5559857b777bf50a62932822a328abf8
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_530__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_530__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..becaaea18d2011cf46f738e90e1dffd0ef3460ce
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__E_S_WW_WM_530__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cf2ac247f86088525289b5193b55c0661d25e35d3a0f67343fd1b624fffad0d
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__F_S_WWC_WM_013__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__F_S_WWC_WM_013__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e0d4ce4b5b8ca01c40473c64b6c9faf94e328488
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_404__F_S_WWC_WM_013__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0fd3ee10728f7d956d133af6df2a1d99b2f9e0f471c03080a28a20934d986cd
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_261__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_261__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..dfe0d836bc774324fef8c0f6100baa40932b2f9d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_261__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91fa70354743741fbd0057aaec6c0557a748c0add6090b57e3088e2d480e6a3b
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_584__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_584__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..091fa10a8acc1ae459c4c2e242f62bdcd0f6fcef
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_584__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26170d480044d3604961cd52fa18ce82c34a34a0420085f4e14f68eefb4f573d
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_624__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_624__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..cc393fea781126e245b73582e8781d4370a15c71
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WWC_WM_624__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5adb156737a25d9980b63b91cc2bd6946ae145bca47d726c927664c7083d7ff
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_199__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_199__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e79b485ae34f17e4d6508b0e2de67dd9387ecc7f
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_199__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4f0a505ef30468350d203b1fde8271c243ed82e95e903bac82007e1583b5500
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_219__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_219__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fd39af34c5d93d747dbf82940852f16dcbf97e00
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_219__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ecf0b412a16193c3ba30e700a0582074088214a0ba32067151e729b9222f1e4
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_468__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_468__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..de0685341d2f34114c38dcba7f0967a18fe2f317
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_456__E_S_WW_WM_468__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb211cd8ba27ce43f777cfd00e15c42be8621d87a261a4555592fc20ab39a17f
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_256__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_256__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c846693819fff54ecb33ca1d6afd40d249c736b3
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_256__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b2bbc21dd7948f2d2b62250a387a74e153de55d8599d59a10cf76ecc8a2703d
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_517__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_517__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0024cd545304eed698eb6d93441b647433a98191
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_517__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13da5cfcbedeac35928229395783fde4ca10688495254dd6a21d60cb254fa8f2
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_582__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_582__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..827e1f907b19700c870c767294696b8c234bbf4b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_582__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:489766e0df6139b0c66d3f67ab0fd0ad09c1d241b1c0ae6d8f2e959e275d4135
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_641__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_641__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e4c1adc9ea6df39a73e7b81bdb4832fe584a349b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WWC_WM_641__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21bb4c568f59a78d815bb1cbf0c993814b8894280ec2886b9d96ca76d4bc906c
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WW_WM_411__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WW_WM_411__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ee8b6075f8d5e8dd01fa55615ae5c1a052efb6b4
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__E_S_WW_WM_411__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:876af47029168740e741006f35808b2bd4495ab76c89ac81f6c5998e49f2abd9
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__F_S_WWC_WM_275__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__F_S_WWC_WM_275__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ad338a2fdaac47e044f5fe31c8baac7584ebd4c4
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_481__F_S_WWC_WM_275__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:662c640a03742b42b929781b39d4bc35e6b0f9787d62a9fe571ad51a9c5902ed
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_181__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_181__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1aa63db132b9eb9d10fc2b8a13408d72088bf801
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_181__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddaeeed9440f9edaad3fb51b0ccd397a818605821c16bd996b6d9e53f7bfd803
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_217__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_217__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..be3951991ee45c0ce27dfd249b1b2b63df76a8ea
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_217__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c74df812e66e5c90c6034ddd8352b000482a998b5b7420b25bc62b88751a25bd
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_405__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_405__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..00c75e6cede23163bd4ee31a421dee44cb17a0db
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_405__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3520860b66e179c8242b84d8908b2b9df342824b5f16e0194a7c92a0bab8dff4
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_691__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_691__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7a1e86f7e6630056ef661cfb8d8d29673f2d08fd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__E_S_WW_WM_691__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:352572129cf847524a67fe5a541ae0ac6050bd62d26009557160338f1e6eb189
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__F_S_WWC_WM_231__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__F_S_WWC_WM_231__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b33a6a9cf80a8f43d019235a984fb8624f285d61
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__F_S_WWC_WM_231__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5df419ff49f51881b072d581acb3e1c4e15ef96e84427f4f9c45e293d55c95f4
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__F_S_WWC_WM_281__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__F_S_WWC_WM_281__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..68db3e8428e0f37ac873adadf82340d2fe6e606d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_557__F_S_WWC_WM_281__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:352d42474af5a546a97ffc11dd242d2c13e1f67adbbe79b2347847ea4d61b5c7
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_005__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_005__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..70f09f581cc647fc0b621debeca4b3f60ab2ff3e
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_005__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f87936349a9cd1c49232fe1a44d62c6c465115fee7deb821d2e05c361be21eab
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_162__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_162__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b3d62ce1af14aca7c0cbccc944e58676f6c7dc6a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_162__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ac6f015fa79f30b5988f9e30be2f3b641ed62a8cbcc36da306bd9f907338699
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_181__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_181__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..63fd67f1b96651dc58e5524e21d86e38ceb0385d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_181__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b41b9dac33b4e33771360405ad77c28579cfa0678eab4af1037886894b4a7fb
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_374__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_374__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..de365ff9ef6a40464348f3c5b1fb7d03d34ead85
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_374__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6648d4d5d8f76d2ead74f71a2651987cbe9d440d39a7d869b34ac3ba8eb25c77
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_675__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_675__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f9b3394fab6db318b05c130437f1236b82758429
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__E_S_WW_WM_675__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1448ebd5992e7b75eb01bf0ce9f3b7374dba61e6150e1295df8a11c0dd5f19fb
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__F_S_WWC_WM_497__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__F_S_WWC_WM_497__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bb1ffd8e1b9a187482b979ccf77a168d6ae01df5
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_574__F_S_WWC_WM_497__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1b09dfabe7ea46fcf9f5ebe65fd0e96edd4d352d203fc9df203f6e4804cdf87
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_050__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_050__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..39dbb4da94725ad0087ac63346e0eed6025fca4b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_050__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ad2613e46c4f5f76d06d83173cd0e7bd405a4458310e1fd450b3404d8d9d492
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_459__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_459__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..020e0b6f7cc3c378c5d4586fca8abb1ef291e6b8
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_459__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d40993bc4774459eb2b02d10ec92881e7a6e2f3032ac995b77d0468bdd741fcc
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_662__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_662__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7421fa90a249ad9a5f222b189a6764b466ed41ed
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WWC_WM_662__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f957ea71ad17e6e6ec786c93e76e4d9bd67fc04175569d6ab9e1f8f8d6d63b0
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_014__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_014__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1da7c0c264542d2dc490ab4bf4080ddee3bf6a90
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_014__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52870d573ab2efbe6ae86391dc836ef99bf0f895aca442f7ab3290852aeb6e9e
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_155__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_155__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2bea83f4f9df125e6d7dc687f31b8122fc9d70e0
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_155__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a9aab6da9d4354bb2de9b75c40f7e0cdf328efb259e215283440553c0c37155
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_583__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_583__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1fc587b8d935c109435d6066e29f245bf82730db
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_585__E_S_WW_WM_583__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c23d42901e6d6d1fc47bf33e9350b03025b5bd9057a34829e28dbe897f40146
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_088__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_088__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8023ff747347119ae343c52e95585db52e496a3b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_088__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fcf175db6f8f093e76ae05ec4f1492ca9b41ccc663c1be99ab638c5af934b77
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_319__rear.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_319__rear.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6c002314ab1dbe537349734b80758a978278df79
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_319__rear.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4709062b1c6d4d2b3df177100d565a85f7b21bc7245c2db8e3f780553c561c61
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_332__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_332__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8ccba6e52a3b42f598dc9b96057632c5bf1ed540
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_332__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14d60b8fcf5b504292f04b074513d0579e0891c2da3fe5c1e44df90a406d782d
+size 9808
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_582__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_582__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3a2878a1192c85ae02c6c902cb1dcd89abfc8694
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WWC_WM_582__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96c31bd9fbf04f5704e91d146a7a765bb50ac5954c558d3178b500fd77e28415
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WW_WM_042__side.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WW_WM_042__side.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8b21518182f42919f5209d3f64bdaeb44432b757
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WW_WM_042__side.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22d943f0ef629061aa8228396e368bd0fb8f6e75b3b643f8e68fb1d4861814f5
+size 9798
diff --git a/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WW_WM_242__front.pt b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WW_WM_242__front.pt
new file mode 100644
index 0000000000000000000000000000000000000000..dab279cf1eb32f9b65bb57a830cad33685929945
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/results/source_init_cache/plain__N_S_WW_WM_619__E_S_WW_WM_242__front.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:398b24581c0c3ebf6048d6e0083d23f0ec422f1b1c03e7a678e944f50c773aa7
+size 9803
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/code_provenance.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/code_provenance.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d65c77fa6bb7025a1484f4ad82c6b647f7a2315
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/code_provenance.json
@@ -0,0 +1,7 @@
+{
+  "file": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "commit": "81d95989e1279e432c4145933bf9ce02c3f70f77",
+  "branch": "flow-field",
+  "train_cond_flow_sha256": "095053598f77923a8c8fbdd63bb77280bd641d5f8ca004bc100e63b82bae7616",
+  "seiler_cond_backbone_sha256": "908f1f3b3756ac054e640b4e6e9fe7a40e2130adcf7ba0ce09432f74783dd97c"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_best.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_best.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e313ad0b3b41c68140da328d44c526fd92603e60
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_best.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cba6977001d79e0ae15ac401e3d20142dd070d18da6b07cbd2484d45a14ee5ae
+size 329404702
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_last.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_last.pt
new file mode 100644
index 0000000000000000000000000000000000000000..581163312614fde857af974f2084841651fb4e22
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_last.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76bee5e98093417ddcab2d1d8908098a8b8164ee0cf0e640dde15ab1e70404e1
+size 988212490
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_step_005000.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_step_005000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..52180eea69a8a4dcadc8fe235c00ab8ec6895b65
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_step_005000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49befdf7002302ba351c93889c4ae6893a1440f2cd454951f218d6058530f03a
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_step_010000.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_step_010000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f611d56d86c9df78ab702e9dc14d2be86baf3b93
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/model_step_010000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bd26c6b05fbd7a9db6b0b7911e368e9b1c71c251845d2171e827116057f31f2
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/run_config.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/run_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..752b628591b735914fa3ab5fe53bf0e5bd38b97c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/run_config.json
@@ -0,0 +1,46 @@
+{
+  "mode": "train",
+  "out_dir": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front",
+  "paired_latents": "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz",
+  "image_view": "front",
+  "seed": 20260526,
+  "val_fraction": 0.05,
+  "max_val": 128,
+  "val_overlap_train": true,
+  "shape_scale": 0.0,
+  "image_scale": 0.0,
+  "hidden_size": 512,
+  "depth": 12,
+  "num_heads": 8,
+  "head_dim": 64,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "pos_embed": "2d",
+  "attention_backend": "sdpa",
+  "gradient_checkpointing": false,
+  "cond_dropout": 0.1,
+  "batch_size": 128,
+  "num_workers": 2,
+  "steps": 10000,
+  "lr": 0.0001,
+  "weight_decay": 0.01,
+  "log_every": 100,
+  "eval_every": 500,
+  "eval_noise_seed": 20260712,
+  "save_every": 5000,
+  "resume": true,
+  "sample_count": 8,
+  "sample_steps": 50,
+  "sample_seed": 20260712,
+  "sample_checkpoint": "model_best.pt",
+  "guidance_scale": 1.0,
+  "decode_mesh": false,
+  "shape_resolution": 512,
+  "shape_max_batch": 32768,
+  "wandb": true,
+  "wandb_project": "drivaer-qwen3v-input-cond",
+  "wandb_entity": "kanydao-epfl",
+  "wandb_run_name": "drivaer3v-inputcond-cfg-front-0816c",
+  "wandb_tags": "drivaer,qwen3v,input-conditional,channel-concat,front",
+  "wandb_mode": "online"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/summary.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..d100ad98a5edbd7ad7881c7e78410baf99e5d321
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/summary.json
@@ -0,0 +1,10 @@
+{
+  "image_view": "front",
+  "best_val_shape": 0.825316309928894,
+  "steps": 10000,
+  "train_samples": 3051,
+  "val_samples": 128,
+  "shape_scale": 1.7132293052897125,
+  "image_scale": 1.6039202568304356,
+  "conditioning": "sketch_latent_channel_concat"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/train_history.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/train_history.json
new file mode 100644
index 0000000000000000000000000000000000000000..2635fde1783c92bcb310330fd25bac019e235e88
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/train_history.json
@@ -0,0 +1,692 @@
+[
+  {
+    "step": 1,
+    "train_shape": 2.0095300674438477,
+    "lr": 9.999999765596898e-05,
+    "elapsed_sec": 1.1659798622131348
+  },
+  {
+    "step": 1,
+    "val_shape": 2.00606107711792
+  },
+  {
+    "step": 100,
+    "train_shape": 1.2359693050384521,
+    "lr": 9.997656161737219e-05,
+    "elapsed_sec": 30.328721523284912
+  },
+  {
+    "step": 200,
+    "train_shape": 0.9719560146331787,
+    "lr": 9.990626960034288e-05,
+    "elapsed_sec": 58.69812893867493
+  },
+  {
+    "step": 300,
+    "train_shape": 0.8980196118354797,
+    "lr": 9.978919331864633e-05,
+    "elapsed_sec": 87.07923674583435
+  },
+  {
+    "step": 400,
+    "train_shape": 0.9003630876541138,
+    "lr": 9.962544831243784e-05,
+    "elapsed_sec": 115.46795678138733
+  },
+  {
+    "step": 500,
+    "train_shape": 0.8763077259063721,
+    "lr": 9.94151961782692e-05,
+    "elapsed_sec": 143.8689079284668
+  },
+  {
+    "step": 500,
+    "val_shape": 0.9301149845123291
+  },
+  {
+    "step": 600,
+    "train_shape": 0.9271390438079834,
+    "lr": 9.915864440961283e-05,
+    "elapsed_sec": 173.21740293502808
+  },
+  {
+    "step": 700,
+    "train_shape": 0.8747949004173279,
+    "lr": 9.885604619209063e-05,
+    "elapsed_sec": 201.6245141029358
+  },
+  {
+    "step": 800,
+    "train_shape": 0.8422240018844604,
+    "lr": 9.85077001536101e-05,
+    "elapsed_sec": 230.04091548919678
+  },
+  {
+    "step": 900,
+    "train_shape": 0.8330157995223999,
+    "lr": 9.811395006965495e-05,
+    "elapsed_sec": 258.45489835739136
+  },
+  {
+    "step": 1000,
+    "train_shape": 0.832108736038208,
+    "lr": 9.767518452401991e-05,
+    "elapsed_sec": 286.88052129745483
+  },
+  {
+    "step": 1000,
+    "val_shape": 0.8935505151748657
+  },
+  {
+    "step": 1100,
+    "train_shape": 0.8795543909072876,
+    "lr": 9.71918365253258e-05,
+    "elapsed_sec": 316.2235383987427
+  },
+  {
+    "step": 1200,
+    "train_shape": 0.8634320497512817,
+    "lr": 9.666438307969207e-05,
+    "elapsed_sec": 344.75169587135315
+  },
+  {
+    "step": 1300,
+    "train_shape": 0.8227393627166748,
+    "lr": 9.60933447199893e-05,
+    "elapsed_sec": 373.1705160140991
+  },
+  {
+    "step": 1400,
+    "train_shape": 0.836679220199585,
+    "lr": 9.547928499213621e-05,
+    "elapsed_sec": 401.59855556488037
+  },
+  {
+    "step": 1500,
+    "train_shape": 0.8759798407554626,
+    "lr": 9.482280989894765e-05,
+    "elapsed_sec": 430.0254828929901
+  },
+  {
+    "step": 1500,
+    "val_shape": 0.8809015154838562
+  },
+  {
+    "step": 1600,
+    "train_shape": 0.8686296939849854,
+    "lr": 9.412456730208378e-05,
+    "elapsed_sec": 459.3630220890045
+  },
+  {
+    "step": 1700,
+    "train_shape": 0.8044033050537109,
+    "lr": 9.338524628268747e-05,
+    "elapsed_sec": 487.78540873527527
+  },
+  {
+    "step": 1800,
+    "train_shape": 0.8579736351966858,
+    "lr": 9.260557646134585e-05,
+    "elapsed_sec": 516.1995508670807
+  },
+  {
+    "step": 1900,
+    "train_shape": 0.8203479051589966,
+    "lr": 9.178632727804172e-05,
+    "elapsed_sec": 544.6289138793945
+  },
+  {
+    "step": 2000,
+    "train_shape": 0.7766510248184204,
+    "lr": 9.092830723281015e-05,
+    "elapsed_sec": 573.0551993846893
+  },
+  {
+    "step": 2000,
+    "val_shape": 0.8681105971336365
+  },
+  {
+    "step": 2100,
+    "train_shape": 0.873916506767273,
+    "lr": 9.003236308784539e-05,
+    "elapsed_sec": 602.3896069526672
+  },
+  {
+    "step": 2200,
+    "train_shape": 0.862097978591919,
+    "lr": 8.90993790318501e-05,
+    "elapsed_sec": 630.8128626346588
+  },
+  {
+    "step": 2300,
+    "train_shape": 0.8179246187210083,
+    "lr": 8.813027580744691e-05,
+    "elapsed_sec": 659.4647295475006
+  },
+  {
+    "step": 2400,
+    "train_shape": 0.7950730919837952,
+    "lr": 8.712600980251722e-05,
+    "elapsed_sec": 687.992662191391
+  },
+  {
+    "step": 2500,
+    "train_shape": 0.9208627939224243,
+    "lr": 8.60875721063612e-05,
+    "elapsed_sec": 716.5881881713867
+  },
+  {
+    "step": 2500,
+    "val_shape": 0.8620848655700684
+  },
+  {
+    "step": 2600,
+    "train_shape": 0.8543415069580078,
+    "lr": 8.501598753161292e-05,
+    "elapsed_sec": 745.9313373565674
+  },
+  {
+    "step": 2700,
+    "train_shape": 0.8070603609085083,
+    "lr": 8.391231360287367e-05,
+    "elapsed_sec": 774.363422870636
+  },
+  {
+    "step": 2800,
+    "train_shape": 0.8095806241035461,
+    "lr": 8.277763951306297e-05,
+    "elapsed_sec": 802.7957773208618
+  },
+  {
+    "step": 2900,
+    "train_shape": 0.7554308176040649,
+    "lr": 8.161308504851662e-05,
+    "elapsed_sec": 831.2304058074951
+  },
+  {
+    "step": 3000,
+    "train_shape": 0.7861917018890381,
+    "lr": 8.04197994838928e-05,
+    "elapsed_sec": 859.6500146389008
+  },
+  {
+    "step": 3000,
+    "val_shape": 0.8567577004432678
+  },
+  {
+    "step": 3100,
+    "train_shape": 0.8349908590316772,
+    "lr": 7.919896044797652e-05,
+    "elapsed_sec": 889.0078327655792
+  },
+  {
+    "step": 3200,
+    "train_shape": 0.7498753070831299,
+    "lr": 7.79517727615026e-05,
+    "elapsed_sec": 917.4341547489166
+  },
+  {
+    "step": 3300,
+    "train_shape": 0.8059127330780029,
+    "lr": 7.667946724814286e-05,
+    "elapsed_sec": 945.8617522716522
+  },
+  {
+    "step": 3400,
+    "train_shape": 0.8071276545524597,
+    "lr": 7.538329951983173e-05,
+    "elapsed_sec": 974.2893459796906
+  },
+  {
+    "step": 3500,
+    "train_shape": 0.8160433769226074,
+    "lr": 7.406454873762866e-05,
+    "elapsed_sec": 1002.7025361061096
+  },
+  {
+    "step": 3500,
+    "val_shape": 0.8492396473884583
+  },
+  {
+    "step": 3600,
+    "train_shape": 0.7519837617874146,
+    "lr": 7.272451634934115e-05,
+    "elapsed_sec": 1032.2714231014252
+  },
+  {
+    "step": 3700,
+    "train_shape": 0.7974459528923035,
+    "lr": 7.136452480515233e-05,
+    "elapsed_sec": 1060.7027056217194
+  },
+  {
+    "step": 3800,
+    "train_shape": 0.8061593770980835,
+    "lr": 6.99859162525224e-05,
+    "elapsed_sec": 1089.1293604373932
+  },
+  {
+    "step": 3900,
+    "train_shape": 0.7961065769195557,
+    "lr": 6.859005121165148e-05,
+    "elapsed_sec": 1117.5458691120148
+  },
+  {
+    "step": 4000,
+    "train_shape": 0.8077917098999023,
+    "lr": 6.71783072328101e-05,
+    "elapsed_sec": 1145.9755835533142
+  },
+  {
+    "step": 4000,
+    "val_shape": 0.8462536334991455
+  },
+  {
+    "step": 4100,
+    "train_shape": 0.783572793006897,
+    "lr": 6.575207753686356e-05,
+    "elapsed_sec": 1175.330373287201
+  },
+  {
+    "step": 4200,
+    "train_shape": 0.7855088114738464,
+    "lr": 6.431276964033111e-05,
+    "elapsed_sec": 1203.7463212013245
+  },
+  {
+    "step": 4300,
+    "train_shape": 0.7716313600540161,
+    "lr": 6.286180396633663e-05,
+    "elapsed_sec": 1232.175327539444
+  },
+  {
+    "step": 4400,
+    "train_shape": 0.7758926749229431,
+    "lr": 6.140061244282311e-05,
+    "elapsed_sec": 1260.5982868671417
+  },
+  {
+    "step": 4500,
+    "train_shape": 0.7754744291305542,
+    "lr": 5.9930637089412255e-05,
+    "elapsed_sec": 1289.0241222381592
+  },
+  {
+    "step": 4500,
+    "val_shape": 0.8421667814254761
+  },
+  {
+    "step": 4600,
+    "train_shape": 0.8085465431213379,
+    "lr": 5.845332859430572e-05,
+    "elapsed_sec": 1318.4387729167938
+  },
+  {
+    "step": 4700,
+    "train_shape": 0.6961475610733032,
+    "lr": 5.697014488263067e-05,
+    "elapsed_sec": 1346.9597284793854
+  },
+  {
+    "step": 4800,
+    "train_shape": 0.7941526174545288,
+    "lr": 5.548254967764354e-05,
+    "elapsed_sec": 1375.387183189392
+  },
+  {
+    "step": 4900,
+    "train_shape": 0.7978916168212891,
+    "lr": 5.3992011056212214e-05,
+    "elapsed_sec": 1403.9837183952332
+  },
+  {
+    "step": 5000,
+    "train_shape": 0.7639372944831848,
+    "lr": 5.250000000000107e-05,
+    "elapsed_sec": 1432.4146976470947
+  },
+  {
+    "step": 5000,
+    "val_shape": 0.8401537537574768
+  },
+  {
+    "step": 5100,
+    "train_shape": 0.7927569150924683,
+    "lr": 5.100798894378995e-05,
+    "elapsed_sec": 1463.6940205097198
+  },
+  {
+    "step": 5200,
+    "train_shape": 0.7865331768989563,
+    "lr": 4.951745032235864e-05,
+    "elapsed_sec": 1492.1158480644226
+  },
+  {
+    "step": 5300,
+    "train_shape": 0.7408822774887085,
+    "lr": 4.8029855117371535e-05,
+    "elapsed_sec": 1520.5466668605804
+  },
+  {
+    "step": 5400,
+    "train_shape": 0.8233519196510315,
+    "lr": 4.654667140569645e-05,
+    "elapsed_sec": 1548.97239112854
+  },
+  {
+    "step": 5500,
+    "train_shape": 0.7616298198699951,
+    "lr": 4.506936291058995e-05,
+    "elapsed_sec": 1577.4015789031982
+  },
+  {
+    "step": 5500,
+    "val_shape": 0.8372927904129028
+  },
+  {
+    "step": 5600,
+    "train_shape": 0.8095413446426392,
+    "lr": 4.359938755717899e-05,
+    "elapsed_sec": 1606.74946975708
+  },
+  {
+    "step": 5700,
+    "train_shape": 0.7651619911193848,
+    "lr": 4.213819603366514e-05,
+    "elapsed_sec": 1635.1806976795197
+  },
+  {
+    "step": 5800,
+    "train_shape": 0.8319501876831055,
+    "lr": 4.0687230359670296e-05,
+    "elapsed_sec": 1663.7003738880157
+  },
+  {
+    "step": 5900,
+    "train_shape": 0.8165658712387085,
+    "lr": 3.924792246313749e-05,
+    "elapsed_sec": 1692.135326385498
+  },
+  {
+    "step": 6000,
+    "train_shape": 0.7386670112609863,
+    "lr": 3.7821692767190845e-05,
+    "elapsed_sec": 1720.563426733017
+  },
+  {
+    "step": 6000,
+    "val_shape": 0.8345818519592285
+  },
+  {
+    "step": 6100,
+    "train_shape": 0.743924617767334,
+    "lr": 3.640994878834947e-05,
+    "elapsed_sec": 1750.066606760025
+  },
+  {
+    "step": 6200,
+    "train_shape": 0.7950394153594971,
+    "lr": 3.501408374747862e-05,
+    "elapsed_sec": 1778.4939451217651
+  },
+  {
+    "step": 6300,
+    "train_shape": 0.7536830306053162,
+    "lr": 3.363547519484887e-05,
+    "elapsed_sec": 1806.9204504489899
+  },
+  {
+    "step": 6400,
+    "train_shape": 0.7760335206985474,
+    "lr": 3.227548365066013e-05,
+    "elapsed_sec": 1835.3491158485413
+  },
+  {
+    "step": 6500,
+    "train_shape": 0.7345336675643921,
+    "lr": 3.0935451262372706e-05,
+    "elapsed_sec": 1863.7828030586243
+  },
+  {
+    "step": 6500,
+    "val_shape": 0.832932710647583
+  },
+  {
+    "step": 6600,
+    "train_shape": 0.7596777677536011,
+    "lr": 2.9616700480169673e-05,
+    "elapsed_sec": 1893.1495413780212
+  },
+  {
+    "step": 6700,
+    "train_shape": 0.7917605638504028,
+    "lr": 2.8320532751858457e-05,
+    "elapsed_sec": 1921.571840763092
+  },
+  {
+    "step": 6800,
+    "train_shape": 0.7722147703170776,
+    "lr": 2.7048227238498715e-05,
+    "elapsed_sec": 1950.001941204071
+  },
+  {
+    "step": 6900,
+    "train_shape": 0.7766357660293579,
+    "lr": 2.5801039552024805e-05,
+    "elapsed_sec": 1978.4287514686584
+  },
+  {
+    "step": 7000,
+    "train_shape": 0.8071856498718262,
+    "lr": 2.458020051610849e-05,
+    "elapsed_sec": 2006.971553325653
+  },
+  {
+    "step": 7000,
+    "val_shape": 0.8302822709083557
+  },
+  {
+    "step": 7100,
+    "train_shape": 0.7794568538665771,
+    "lr": 2.3386914951484544e-05,
+    "elapsed_sec": 2036.3444521427155
+  },
+  {
+    "step": 7200,
+    "train_shape": 0.7872241735458374,
+    "lr": 2.222236048693809e-05,
+    "elapsed_sec": 2064.9168798923492
+  },
+  {
+    "step": 7300,
+    "train_shape": 0.7912191152572632,
+    "lr": 2.1087686397127328e-05,
+    "elapsed_sec": 2093.56316781044
+  },
+  {
+    "step": 7400,
+    "train_shape": 0.8204686045646667,
+    "lr": 1.998401246838802e-05,
+    "elapsed_sec": 2121.993660926819
+  },
+  {
+    "step": 7500,
+    "train_shape": 0.7824258804321289,
+    "lr": 1.891242789363976e-05,
+    "elapsed_sec": 2150.4096460342407
+  },
+  {
+    "step": 7500,
+    "val_shape": 0.8290730118751526
+  },
+  {
+    "step": 7600,
+    "train_shape": 0.7639555931091309,
+    "lr": 1.787399019748373e-05,
+    "elapsed_sec": 2179.7623631954193
+  },
+  {
+    "step": 7700,
+    "train_shape": 0.7192654013633728,
+    "lr": 1.6869724192553968e-05,
+    "elapsed_sec": 2208.1892786026
+  },
+  {
+    "step": 7800,
+    "train_shape": 0.7758867144584656,
+    "lr": 1.5900620968150808e-05,
+    "elapsed_sec": 2236.617487668991
+  },
+  {
+    "step": 7900,
+    "train_shape": 0.8120406866073608,
+    "lr": 1.496763691215551e-05,
+    "elapsed_sec": 2265.0407559871674
+  },
+  {
+    "step": 8000,
+    "train_shape": 0.818740725517273,
+    "lr": 1.4071692767190806e-05,
+    "elapsed_sec": 2293.46453166008
+  },
+  {
+    "step": 8000,
+    "val_shape": 0.8271117806434631
+  },
+  {
+    "step": 8100,
+    "train_shape": 0.7975656986236572,
+    "lr": 1.3213672721959122e-05,
+    "elapsed_sec": 2322.8237578868866
+  },
+  {
+    "step": 8200,
+    "train_shape": 0.7881734371185303,
+    "lr": 1.2394423538655062e-05,
+    "elapsed_sec": 2351.2517495155334
+  },
+  {
+    "step": 8300,
+    "train_shape": 0.8042082190513611,
+    "lr": 1.1614753717313381e-05,
+    "elapsed_sec": 2379.68536734581
+  },
+  {
+    "step": 8400,
+    "train_shape": 0.7898544073104858,
+    "lr": 1.0875432697917097e-05,
+    "elapsed_sec": 2408.103704214096
+  },
+  {
+    "step": 8500,
+    "train_shape": 0.7417004108428955,
+    "lr": 1.0177190101053074e-05,
+    "elapsed_sec": 2436.811436891556
+  },
+  {
+    "step": 8500,
+    "val_shape": 0.8267183303833008
+  },
+  {
+    "step": 8600,
+    "train_shape": 0.7516779899597168,
+    "lr": 9.520715007864551e-06,
+    "elapsed_sec": 2466.1670911312103
+  },
+  {
+    "step": 8700,
+    "train_shape": 0.7402617335319519,
+    "lr": 8.906655280011289e-06,
+    "elapsed_sec": 2494.5950319767
+  },
+  {
+    "step": 8800,
+    "train_shape": 0.7777749300003052,
+    "lr": 8.335616920308386e-06,
+    "elapsed_sec": 2523.017970561981
+  },
+  {
+    "step": 8900,
+    "train_shape": 0.734040379524231,
+    "lr": 7.808163474674577e-06,
+    "elapsed_sec": 2551.4431574344635
+  },
+  {
+    "step": 9000,
+    "train_shape": 0.78995680809021,
+    "lr": 7.32481547598044e-06,
+    "elapsed_sec": 2579.864238500595
+  },
+  {
+    "step": 9000,
+    "val_shape": 0.8270562887191772
+  },
+  {
+    "step": 9100,
+    "train_shape": 0.7355107665061951,
+    "lr": 6.886049930345396e-06,
+    "elapsed_sec": 2608.473929166794
+  },
+  {
+    "step": 9200,
+    "train_shape": 0.7915401458740234,
+    "lr": 6.492299846390179e-06,
+    "elapsed_sec": 2636.9158411026
+  },
+  {
+    "step": 9300,
+    "train_shape": 0.7653471231460571,
+    "lr": 6.143953807909618e-06,
+    "elapsed_sec": 2665.4597051143646
+  },
+  {
+    "step": 9400,
+    "train_shape": 0.7720314264297485,
+    "lr": 5.841355590387376e-06,
+    "elapsed_sec": 2693.889242172241
+  },
+  {
+    "step": 9500,
+    "train_shape": 0.7706986665725708,
+    "lr": 5.5848038217310204e-06,
+    "elapsed_sec": 2722.3175752162933
+  },
+  {
+    "step": 9500,
+    "val_shape": 0.8254305720329285
+  },
+  {
+    "step": 9600,
+    "train_shape": 0.8390926718711853,
+    "lr": 5.374551687562344e-06,
+    "elapsed_sec": 2751.683788061142
+  },
+  {
+    "step": 9700,
+    "train_shape": 0.7521244287490845,
+    "lr": 5.2108066813537186e-06,
+    "elapsed_sec": 2780.108333349228
+  },
+  {
+    "step": 9800,
+    "train_shape": 0.7757055759429932,
+    "lr": 5.093730399657109e-06,
+    "elapsed_sec": 2808.723040819168
+  },
+  {
+    "step": 9900,
+    "train_shape": 0.7241287231445312,
+    "lr": 5.023438382627753e-06,
+    "elapsed_sec": 2837.151882648468
+  },
+  {
+    "step": 10000,
+    "train_shape": 0.7819746136665344,
+    "lr": 5e-06,
+    "elapsed_sec": 2865.5812153816223
+  },
+  {
+    "step": 10000,
+    "val_shape": 0.825316309928894
+  }
+]
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..4c1fb2f4b859968c8461b1551ef93105beedde5d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:48.411806901Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:48.72611076Z","level":"INFO","msg":"stream: created new stream","id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.726197471Z","level":"INFO","msg":"handler: started","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.728531908Z","level":"INFO","msg":"stream: started","id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.728553418Z","level":"INFO","msg":"writer: started","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.728558017Z","level":"INFO","msg":"sender: started","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.524371942Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:40.773198116Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:40.777810105Z","level":"INFO","msg":"stream: closing","id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.777820735Z","level":"INFO","msg":"handler: closed","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.778978481Z","level":"INFO","msg":"sender: closed","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.778989892Z","level":"INFO","msg":"stream: closed","id":"1jk61ovk"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/debug.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..38c5a9db202e8aaef41a3a418f4575ce9e408d86
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug.log
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug-internal.log
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz', 'image_view': 'front', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.1, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-cfg-front-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,front', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:48,402 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:48,408 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:48,410 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:48,423 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:49,184 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:49,647 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:39,803 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/1jk61ovk
+2026-08-16 20:03:39,804 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:39,804 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:39,804 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:40,776 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/config.yaml b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..522aa2a64ee7d4761e4e0254fa58bda5678287de
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/config.yaml
@@ -0,0 +1,172 @@
+_wandb:
+    value:
+        cli_version: 0.22.3
+        e:
+            dkcl00gx0e611avqtiael2z2w1ea42od:
+                args:
+                    - --mode
+                    - train
+                    - --image-view
+                    - front
+                    - --paired-latents
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz
+                    - --out-dir
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front
+                    - --steps
+                    - "10000"
+                    - --batch-size
+                    - "128"
+                    - --lr
+                    - "1e-4"
+                    - --weight-decay
+                    - "0.01"
+                    - --seed
+                    - "20260526"
+                    - --pos-embed
+                    - 2d
+                    - --cond-dropout
+                    - "0.1"
+                    - --resume
+                    - --wandb
+                    - --wandb-run-name
+                    - drivaer3v-inputcond-cfg-front-0816c
+                    - --wandb-tags
+                    - drivaer,qwen3v,input-conditional,channel-concat,front
+                codePath: code/ICLR/customized-design/input_cond/train_cond_flow.py
+                codePathLocal: train_cond_flow.py
+                cpu_count: 32
+                cpu_count_logical: 64
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "7679361679360"
+                        used: "2855216566272"
+                email: 1193414549@qq.com
+                executable: /mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python
+                git:
+                    commit: 81d95989e1279e432c4145933bf9ce02c3f70f77
+                gpu: NVIDIA A100-SXM4-40GB
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "42949672960"
+                      name: NVIDIA A100-SXM4-40GB
+                      uuid: GPU-90650bb1-0aa9-a3b6-b9b1-25da163206eb
+                host: inputcond-cfg-front-a100-40g-0816c-0-0
+                memory:
+                    total: "540742438912"
+                os: Linux-6.8.0-124-generic-x86_64-with-glibc2.35
+                program: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py
+                python: CPython 3.10.18
+                root: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front/wandb
+                startedAt: "2026-08-16T19:15:48.172204Z"
+                writerId: dkcl00gx0e611avqtiael2z2w1ea42od
+        m: []
+        python_version: 3.10.18
+        t:
+            "1":
+                - 1
+                - 105
+            "2":
+                - 1
+                - 105
+            "3":
+                - 2
+                - 13
+                - 15
+                - 16
+                - 61
+                - 62
+            "4": 3.10.18
+            "5": 0.22.3
+            "12": 0.22.3
+            "13": linux-x86_64
+attention_backend:
+    value: sdpa
+batch_size:
+    value: 128
+cond_dropout:
+    value: 0.1
+decode_mesh:
+    value: false
+depth:
+    value: 12
+dropout:
+    value: 0
+eval_every:
+    value: 500
+eval_noise_seed:
+    value: 20260712
+gradient_checkpointing:
+    value: false
+guidance_scale:
+    value: 1
+head_dim:
+    value: 64
+hidden_size:
+    value: 512
+image_scale:
+    value: 0
+image_view:
+    value: front
+log_every:
+    value: 100
+lr:
+    value: 0.0001
+max_val:
+    value: 128
+mlp_ratio:
+    value: 4
+mode:
+    value: train
+num_heads:
+    value: 8
+num_workers:
+    value: 2
+out_dir:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front
+paired_latents:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz
+pos_embed:
+    value: 2d
+resume:
+    value: true
+sample_checkpoint:
+    value: model_best.pt
+sample_count:
+    value: 8
+sample_seed:
+    value: 20260712
+sample_steps:
+    value: 50
+save_every:
+    value: 5000
+seed:
+    value: 20260526
+shape_max_batch:
+    value: 32768
+shape_resolution:
+    value: 512
+shape_scale:
+    value: 0
+steps:
+    value: 10000
+val_fraction:
+    value: 0.05
+val_overlap_train:
+    value: true
+wandb:
+    value: true
+wandb_entity:
+    value: kanydao-epfl
+wandb_mode:
+    value: online
+wandb_project:
+    value: drivaer-qwen3v-input-cond
+wandb_run_name:
+    value: drivaer3v-inputcond-cfg-front-0816c
+wandb_tags:
+    value: drivaer,qwen3v,input-conditional,channel-concat,front
+weight_decay:
+    value: 0.01
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/output.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..a2a9cc2171915826dd18b32cfbd86db85c0d5fe1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/output.log
@@ -0,0 +1,123 @@
+[wandb] project=drivaer-qwen3v-input-cond run=drivaer3v-inputcond-cfg-front-0816c mode=online
+[train] step=000001 loss=2.009530 lr=1.00e-04
+[val] step=000001 shape=2.006061
+[train] step=000100 loss=1.235969 lr=1.00e-04
+[train] step=000200 loss=0.971956 lr=9.99e-05
+[train] step=000300 loss=0.898020 lr=9.98e-05
+[train] step=000400 loss=0.900363 lr=9.96e-05
+[train] step=000500 loss=0.876308 lr=9.94e-05
+[val] step=000500 shape=0.930115
+[train] step=000600 loss=0.927139 lr=9.92e-05
+[train] step=000700 loss=0.874795 lr=9.89e-05
+[train] step=000800 loss=0.842224 lr=9.85e-05
+[train] step=000900 loss=0.833016 lr=9.81e-05
+[train] step=001000 loss=0.832109 lr=9.77e-05
+[val] step=001000 shape=0.893551
+[train] step=001100 loss=0.879554 lr=9.72e-05
+[train] step=001200 loss=0.863432 lr=9.67e-05
+[train] step=001300 loss=0.822739 lr=9.61e-05
+[train] step=001400 loss=0.836679 lr=9.55e-05
+[train] step=001500 loss=0.875980 lr=9.48e-05
+[val] step=001500 shape=0.880902
+[train] step=001600 loss=0.868630 lr=9.41e-05
+[train] step=001700 loss=0.804403 lr=9.34e-05
+[train] step=001800 loss=0.857974 lr=9.26e-05
+[train] step=001900 loss=0.820348 lr=9.18e-05
+[train] step=002000 loss=0.776651 lr=9.09e-05
+[val] step=002000 shape=0.868111
+[train] step=002100 loss=0.873917 lr=9.00e-05
+[train] step=002200 loss=0.862098 lr=8.91e-05
+[train] step=002300 loss=0.817925 lr=8.81e-05
+[train] step=002400 loss=0.795073 lr=8.71e-05
+[train] step=002500 loss=0.920863 lr=8.61e-05
+[val] step=002500 shape=0.862085
+[train] step=002600 loss=0.854342 lr=8.50e-05
+[train] step=002700 loss=0.807060 lr=8.39e-05
+[train] step=002800 loss=0.809581 lr=8.28e-05
+[train] step=002900 loss=0.755431 lr=8.16e-05
+[train] step=003000 loss=0.786192 lr=8.04e-05
+[val] step=003000 shape=0.856758
+[train] step=003100 loss=0.834991 lr=7.92e-05
+[train] step=003200 loss=0.749875 lr=7.80e-05
+[train] step=003300 loss=0.805913 lr=7.67e-05
+[train] step=003400 loss=0.807128 lr=7.54e-05
+[train] step=003500 loss=0.816043 lr=7.41e-05
+[val] step=003500 shape=0.849240
+[train] step=003600 loss=0.751984 lr=7.27e-05
+[train] step=003700 loss=0.797446 lr=7.14e-05
+[train] step=003800 loss=0.806159 lr=7.00e-05
+[train] step=003900 loss=0.796107 lr=6.86e-05
+[train] step=004000 loss=0.807792 lr=6.72e-05
+[val] step=004000 shape=0.846254
+[train] step=004100 loss=0.783573 lr=6.58e-05
+[train] step=004200 loss=0.785509 lr=6.43e-05
+[train] step=004300 loss=0.771631 lr=6.29e-05
+[train] step=004400 loss=0.775893 lr=6.14e-05
+[train] step=004500 loss=0.775474 lr=5.99e-05
+[val] step=004500 shape=0.842167
+[train] step=004600 loss=0.808547 lr=5.85e-05
+[train] step=004700 loss=0.696148 lr=5.70e-05
+[train] step=004800 loss=0.794153 lr=5.55e-05
+[train] step=004900 loss=0.797892 lr=5.40e-05
+[train] step=005000 loss=0.763937 lr=5.25e-05
+[val] step=005000 shape=0.840154
+[train] step=005100 loss=0.792757 lr=5.10e-05
+[train] step=005200 loss=0.786533 lr=4.95e-05
+[train] step=005300 loss=0.740882 lr=4.80e-05
+[train] step=005400 loss=0.823352 lr=4.65e-05
+[train] step=005500 loss=0.761630 lr=4.51e-05
+[val] step=005500 shape=0.837293
+[train] step=005600 loss=0.809541 lr=4.36e-05
+[train] step=005700 loss=0.765162 lr=4.21e-05
+[train] step=005800 loss=0.831950 lr=4.07e-05
+[train] step=005900 loss=0.816566 lr=3.92e-05
+[train] step=006000 loss=0.738667 lr=3.78e-05
+[val] step=006000 shape=0.834582
+[train] step=006100 loss=0.743925 lr=3.64e-05
+[train] step=006200 loss=0.795039 lr=3.50e-05
+[train] step=006300 loss=0.753683 lr=3.36e-05
+[train] step=006400 loss=0.776034 lr=3.23e-05
+[train] step=006500 loss=0.734534 lr=3.09e-05
+[val] step=006500 shape=0.832933
+[train] step=006600 loss=0.759678 lr=2.96e-05
+[train] step=006700 loss=0.791761 lr=2.83e-05
+[train] step=006800 loss=0.772215 lr=2.70e-05
+[train] step=006900 loss=0.776636 lr=2.58e-05
+[train] step=007000 loss=0.807186 lr=2.46e-05
+[val] step=007000 shape=0.830282
+[train] step=007100 loss=0.779457 lr=2.34e-05
+[train] step=007200 loss=0.787224 lr=2.22e-05
+[train] step=007300 loss=0.791219 lr=2.11e-05
+[train] step=007400 loss=0.820469 lr=2.00e-05
+[train] step=007500 loss=0.782426 lr=1.89e-05
+[val] step=007500 shape=0.829073
+[train] step=007600 loss=0.763956 lr=1.79e-05
+[train] step=007700 loss=0.719265 lr=1.69e-05
+[train] step=007800 loss=0.775887 lr=1.59e-05
+[train] step=007900 loss=0.812041 lr=1.50e-05
+[train] step=008000 loss=0.818741 lr=1.41e-05
+[val] step=008000 shape=0.827112
+[train] step=008100 loss=0.797566 lr=1.32e-05
+[train] step=008200 loss=0.788173 lr=1.24e-05
+[train] step=008300 loss=0.804208 lr=1.16e-05
+[train] step=008400 loss=0.789854 lr=1.09e-05
+[train] step=008500 loss=0.741700 lr=1.02e-05
+[val] step=008500 shape=0.826718
+[train] step=008600 loss=0.751678 lr=9.52e-06
+[train] step=008700 loss=0.740262 lr=8.91e-06
+[train] step=008800 loss=0.777775 lr=8.34e-06
+[train] step=008900 loss=0.734040 lr=7.81e-06
+[train] step=009000 loss=0.789957 lr=7.32e-06
+[val] step=009000 shape=0.827056
+[train] step=009100 loss=0.735511 lr=6.89e-06
+[train] step=009200 loss=0.791540 lr=6.49e-06
+[train] step=009300 loss=0.765347 lr=6.14e-06
+[train] step=009400 loss=0.772031 lr=5.84e-06
+[train] step=009500 loss=0.770699 lr=5.58e-06
+[val] step=009500 shape=0.825431
+[train] step=009600 loss=0.839093 lr=5.37e-06
+[train] step=009700 loss=0.752124 lr=5.21e-06
+[train] step=009800 loss=0.775706 lr=5.09e-06
+[train] step=009900 loss=0.724129 lr=5.02e-06
+[train] step=010000 loss=0.781975 lr=5.00e-06
+[val] step=010000 shape=0.825316
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/requirements.txt b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..2e10c9ab6bc99041706780299f9e3e5035f34676
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/requirements.txt
@@ -0,0 +1,302 @@
+shellingham==1.5.4
+importlib_metadata==8.7.0
+websocket-client==1.8.0
+rfc3986-validator==0.1.1
+aiohappyeyeballs==2.6.1
+tensorboardX==2.6.2.2
+matplotlib==3.8.4
+zstandard==0.23.0
+gradio_client==1.4.2
+nbclient==0.10.2
+pyaml==25.7.0
+argparse==1.4.0
+h11==0.16.0
+scooby==0.10.1
+pexpect==4.9.0
+jedi==0.19.2
+scikit-optimize==0.10.2
+omegaconf==2.3.0
+tomlkit==0.12.0
+yarl==1.20.1
+point-cloud-utils==0.34.0
+cffi==1.17.1
+asttokens==3.0.0
+webencodings==0.5.1
+imageio==2.34.1
+attrs==25.3.0
+hydra-core==1.3.2
+datasets==2.19.0
+pyquaternion==0.9.9
+trame-common==1.0.0
+wslink==2.3.4
+Pygments==2.19.2
+pydantic==2.11.7
+regex==2024.11.6
+timm==0.9.16
+ipykernel==6.29.5
+prompt_toolkit==3.0.51
+torchdiffeq==0.2.5
+pybind11==2.13.6
+debugpy==1.8.14
+pytz==2025.2
+transformers==4.40.1
+jupyterlab_pygments==0.3.0
+dash==3.1.1
+psutil==5.9.8
+tqdm==4.66.2
+multidict==6.6.3
+setproctitle==1.3.6
+nbformat==5.10.4
+prometheus_client==0.22.1
+jupyter_core==5.8.1
+gradio==5.5.0
+multiprocess==0.70.16
+fastapi==0.115.14
+simpervisor==1.0.0
+comm==0.2.2
+widgetsnbextension==4.0.14
+webcolors==24.11.1
+sentry-sdk==2.32.0
+tornado==6.5.1
+defusedxml==0.7.1
+Markdown==3.8.2
+numba==0.61.2
+pyarrow-hotfix==0.7
+trame-server==3.5.0
+nbconvert==7.16.6
+websockets==12.0
+semantic-version==2.10.0
+bpy==4.0.0
+smmap==5.0.2
+grpcio==1.73.1
+referencing==0.36.2
+annotated-types==0.7.0
+stack-data==0.6.3
+vtk==9.4.2
+safehttpx==0.1.6
+GitPython==3.1.44
+ipywidgets==8.1.7
+platformdirs==4.3.8
+tzdata==2025.2
+typer==0.16.0
+click==8.2.1
+networkx==3.3
+fsspec==2024.3.1
+plotly==6.2.0
+isoduration==20.11.0
+joblib==1.4.0
+aiosignal==1.4.0
+jsonschema-specifications==2025.4.1
+six==1.17.0
+pyzmq==27.0.0
+lazy_loader==0.4
+typeguard==2.13.3
+markdown-it-py==3.0.0
+httpx==0.28.1
+fqdn==1.5.1
+threadpoolctl==3.6.0
+orjson==3.10.18
+docker-pycreds==0.4.0
+jsonschema==4.24.0
+tensorboard-data-server==0.7.2
+nvidia-cufft-cu12==11.0.2.54
+aiofiles==23.2.1
+rpds-py==0.26.0
+Cython==3.1.2
+python-multipart==0.0.12
+nvidia-cuda-runtime-cu12==12.1.105
+huggingface-hub==0.26.2
+frozenlist==1.7.0
+onnxruntime==1.22.0
+cycler==0.12.1
+tifffile==2025.5.10
+safetensors==0.4.3
+nest-asyncio==1.6.0
+parso==0.8.4
+PyMCubes==0.1.4
+nvidia-nvtx-cu12==12.1.105
+pydub==0.25.1
+packaging==25.0
+ipython==8.37.0
+uvicorn==0.35.0
+pyarrow==20.0.0
+rich==14.0.0
+executing==2.2.0
+jupyter_server_terminals==0.5.3
+typing_extensions==4.12.2
+typing_extensions==4.15.0
+tensorboard==2.16.2
+argon2-cffi-bindings==21.2.0
+wcwidth==0.2.13
+typing-inspection==0.4.1
+pure_eval==0.2.3
+trimesh==4.3.2
+nvidia-cudnn-cu12==9.1.0.70
+humanfriendly==10.0
+Werkzeug==3.1.3
+decorator==5.2.1
+nvidia-cuda-nvrtc-cu12==12.1.105
+aiohttp==3.12.13
+uri-template==1.3.0
+anyio==4.9.0
+fastjsonschema==2.21.1
+nvidia-curand-cu12==10.3.2.106
+zipp==3.23.0
+rtree==1.4.1
+kiwisolver==1.4.8
+Flask==3.1.1
+nvidia-cuda-cupti-cu12==12.1.105
+pysdf==0.1.9
+sniffio==1.3.1
+itsdangerous==2.2.0
+rfc3339-validator==0.1.4
+trame==3.10.2
+open3d==0.18.0
+scikit-learn==1.7.0
+mistune==3.1.3
+diffusers==0.31.0
+absl-py==2.3.1
+cubvh==0.1.0
+trame-client==3.9.1
+einops==0.8.0
+plyfile==1.0.3
+bleach==6.2.0
+ptyprocess==0.7.0
+pycparser==2.22
+llvmlite==0.44.0
+addict==2.4.0
+diso==0.1.4
+Send2Trash==1.8.3
+more-itertools==10.7.0
+scikit-image==0.23.2
+propcache==0.3.2
+pandas==2.2.2
+flatbuffers==25.2.10
+gitdb==4.0.12
+jaxtyping==0.2.28
+soupsieve==2.7
+mdurl==0.1.2
+protobuf==5.29.5
+fonttools==4.58.5
+coloredlogs==15.0.1
+spaces==0.28.3
+jupyter_server==2.16.0
+trame-vtk==2.9.0
+terminado==0.18.1
+overrides==7.7.0
+pandocfilters==1.5.1
+nvidia-cusolver-cu12==11.4.5.107
+trame-vuetify==3.0.1
+async-timeout==5.0.1
+blinker==1.9.0
+exceptiongroup==1.3.0
+xxhash==3.5.0
+python-json-logger==3.3.0
+pyvista==0.45.2
+python-dateutil==2.9.0.post0
+ConfigArgParse==1.7.1
+PyMatting==1.1.14
+fpsample==0.3.3
+traitlets==5.14.3
+msgpack==1.1.1
+argon2-cffi==25.1.0
+tinycss2==1.4.0
+retrying==1.4.0
+jsonpointer==3.0.0
+matplotlib-inline==0.1.7
+jupyter_client==8.6.3
+contourpy==1.3.2
+lightning-utilities==0.11.2
+pydantic_core==2.33.2
+httpcore==1.0.9
+narwhals==1.46.0
+nvidia-cublas-cu12==12.1.3.1
+jupyterlab_widgets==3.0.15
+antlr4-python3-runtime==4.9.3
+pyparsing==3.1.2
+rembg==2.0.59
+ffmpy==0.6.0
+tokenizers==0.19.1
+beautifulsoup4==4.13.4
+pooch==1.8.2
+ninja==1.11.1.4
+jupyter_server_proxy==4.4.0
+ruff==0.12.2
+types-python-dateutil==2.9.0.20250708
+dill==0.3.8
+pytorch-lightning==2.2.4
+jupyter-events==0.12.0
+starlette==0.46.2
+arrow==1.3.0
+setuptools==78.1.1
+wheel==0.45.1
+pip==25.1
+mkl-service==2.5.2
+mpi4py==4.0.3
+h5py==3.14.0
+mkl_fft==2.1.1
+mkl_random==1.3.0
+mcubes==0.1.6
+wandb==0.22.3
+certifi==2026.1.4
+mpmath==1.3.0
+triton==3.1.0
+sympy==1.13.1
+nvidia-nccl-cu12==2.21.5
+nvidia-nvjitlink-cu12==12.9.86
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nvtx-cu11==11.8.86
+nvidia-nccl-cu11==2.21.5
+nvidia-cusparse-cu11==11.7.5.86
+nvidia-curand-cu11==10.3.0.86
+nvidia-cufft-cu11==10.9.0.58
+nvidia-cuda-runtime-cu11==11.8.89
+nvidia-cuda-nvrtc-cu11==11.8.89
+nvidia-cuda-cupti-cu11==11.8.87
+nvidia-cublas-cu11==11.11.3.6
+nvidia-cusolver-cu11==11.4.1.48
+nvidia-cudnn-cu11==9.1.0.70
+torch==2.5.1+cu118
+torchvision==0.20.1+cu118
+torchaudio==2.5.1+cu118
+PyYAML==6.0.3
+pillow==12.1.0
+torchmetrics==1.8.2
+accelerate==1.12.0
+urllib3==2.6.3
+idna==3.11
+charset-normalizer==3.4.4
+requests==2.32.5
+filelock==3.20.3
+torch_cluster==1.6.3+pt25cu118
+numpy==1.26.4
+opencv-python-headless==4.9.0.80
+PyOpenGL-accelerate==3.1.10
+PyOpenGL==3.1.0
+pyglet==2.1.14
+freetype-py==2.5.1
+pyrender==0.1.45
+scipy==1.15.3
+memory-profiler==0.61.0
+pyamg==5.3.0
+meshio==5.3.5
+torch-fem==0.6.2
+torch-geometric==2.7.0
+MarkupSafe==3.0.3
+Jinja2==3.1.6
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/wandb-metadata.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..4dccb554b1ff820d6f1943db90cda20cb8228ef1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/wandb-metadata.json
@@ -0,0 +1,69 @@
+{
+  "os":  "Linux-6.8.0-124-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.10.18",
+  "startedAt":  "2026-08-16T19:15:48.172204Z",
+  "args":  [
+    "--mode",
+    "train",
+    "--image-view",
+    "front",
+    "--paired-latents",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz",
+    "--out-dir",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front",
+    "--steps",
+    "10000",
+    "--batch-size",
+    "128",
+    "--lr",
+    "1e-4",
+    "--weight-decay",
+    "0.01",
+    "--seed",
+    "20260526",
+    "--pos-embed",
+    "2d",
+    "--cond-dropout",
+    "0.1",
+    "--resume",
+    "--wandb",
+    "--wandb-run-name",
+    "drivaer3v-inputcond-cfg-front-0816c",
+    "--wandb-tags",
+    "drivaer,qwen3v,input-conditional,channel-concat,front"
+  ],
+  "program":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePath":  "code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePathLocal":  "train_cond_flow.py",
+  "git":  {
+    "commit":  "81d95989e1279e432c4145933bf9ce02c3f70f77"
+  },
+  "email":  "1193414549@qq.com",
+  "root":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front/wandb",
+  "host":  "inputcond-cfg-front-a100-40g-0816c-0-0",
+  "executable":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python",
+  "cpu_count":  32,
+  "cpu_count_logical":  64,
+  "gpu":  "NVIDIA A100-SXM4-40GB",
+  "gpu_count":  1,
+  "disk":  {
+    "/":  {
+      "total":  "7679361679360",
+      "used":  "2855216566272"
+    }
+  },
+  "memory":  {
+    "total":  "540742438912"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A100-SXM4-40GB",
+      "memoryTotal":  "42949672960",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-90650bb1-0aa9-a3b6-b9b1-25da163206eb"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "dkcl00gx0e611avqtiael2z2w1ea42od"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/wandb-summary.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..250c4ac24f2b0a1e136c0b148a830a39f7cdcd96
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_runtime":2870.619282072,"_wandb":{"runtime":2870},"_step":10000,"shape_scale":1.7132293052897125,"_timestamp":1.786910615419891e+09,"image_scale":1.6039202568304356,"train/shape":0.7819746136665344,"image_view":"front","steps":10000,"time/elapsed_sec":2865.5812153816223,"train_samples":3051,"val/shape":0.825316309928894,"conditioning":"sketch_latent_channel_concat","best_val_shape":0.825316309928894,"val_samples":128,"train/lr":5e-06}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..4c1fb2f4b859968c8461b1551ef93105beedde5d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:48.411806901Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:48.72611076Z","level":"INFO","msg":"stream: created new stream","id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.726197471Z","level":"INFO","msg":"handler: started","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.728531908Z","level":"INFO","msg":"stream: started","id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.728553418Z","level":"INFO","msg":"writer: started","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T19:15:48.728558017Z","level":"INFO","msg":"sender: started","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.524371942Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:40.773198116Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:40.777810105Z","level":"INFO","msg":"stream: closing","id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.777820735Z","level":"INFO","msg":"handler: closed","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.778978481Z","level":"INFO","msg":"sender: closed","stream_id":"1jk61ovk"}
+{"time":"2026-08-16T20:03:40.778989892Z","level":"INFO","msg":"stream: closed","id":"1jk61ovk"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..38c5a9db202e8aaef41a3a418f4575ce9e408d86
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug.log
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/logs/debug-internal.log
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/front', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz', 'image_view': 'front', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.1, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-cfg-front-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,front', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:48,188 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:48,402 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:48,408 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:48,410 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:48,423 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:49,184 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:49,647 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:39,803 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/1jk61ovk
+2026-08-16 20:03:39,804 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:39,804 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:39,804 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:40,776 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/run-1jk61ovk.wandb b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/run-1jk61ovk.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..040b3a3c5286349a4db9d72d5088e32a4baf28f9
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/front/wandb/wandb/run-20260816_191548-1jk61ovk/run-1jk61ovk.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8610a2cfe9e950dd6b3b39b908230132152093d2e8bd3a2d881a6945285b98a8
+size 237257
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/code_provenance.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/code_provenance.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d65c77fa6bb7025a1484f4ad82c6b647f7a2315
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/code_provenance.json
@@ -0,0 +1,7 @@
+{
+  "file": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "commit": "81d95989e1279e432c4145933bf9ce02c3f70f77",
+  "branch": "flow-field",
+  "train_cond_flow_sha256": "095053598f77923a8c8fbdd63bb77280bd641d5f8ca004bc100e63b82bae7616",
+  "seiler_cond_backbone_sha256": "908f1f3b3756ac054e640b4e6e9fe7a40e2130adcf7ba0ce09432f74783dd97c"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_best.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_best.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3131d88b016a27dfea68a085ac21de49d480376b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_best.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e37becec5f064df95ca13472b638df0818fc847cc39be3486cd701cd65eec8e0
+size 329404702
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_last.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_last.pt
new file mode 100644
index 0000000000000000000000000000000000000000..954a6d494419c1806b604324e60715e3af8804d7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_last.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3615c344f67fa69b406b822a1f655eebe12f9712e2ed94ff04390106a41bd26
+size 988212490
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_step_005000.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_step_005000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..478451e5f49592f156a6987cfbd704e146b3184e
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_step_005000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a475c8c13fd98d484a2805fa900e1175a30574e65f5d99c17f9a25ca8b7488f3
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_step_010000.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_step_010000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..dfc621f51c0e9f29e0be2fa32a8015fbfe9343f1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/model_step_010000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5349d5873ff6027ca5c3cd25d8b1ad65bbd5771e09214f1f92b02ff2cf8eb1ab
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/run_config.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/run_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..e6ec66c53eec7b09c75585a2b504ad1c033c94dc
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/run_config.json
@@ -0,0 +1,46 @@
+{
+  "mode": "train",
+  "out_dir": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear",
+  "paired_latents": "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz",
+  "image_view": "rear",
+  "seed": 20260526,
+  "val_fraction": 0.05,
+  "max_val": 128,
+  "val_overlap_train": true,
+  "shape_scale": 0.0,
+  "image_scale": 0.0,
+  "hidden_size": 512,
+  "depth": 12,
+  "num_heads": 8,
+  "head_dim": 64,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "pos_embed": "2d",
+  "attention_backend": "sdpa",
+  "gradient_checkpointing": false,
+  "cond_dropout": 0.1,
+  "batch_size": 128,
+  "num_workers": 2,
+  "steps": 10000,
+  "lr": 0.0001,
+  "weight_decay": 0.01,
+  "log_every": 100,
+  "eval_every": 500,
+  "eval_noise_seed": 20260712,
+  "save_every": 5000,
+  "resume": true,
+  "sample_count": 8,
+  "sample_steps": 50,
+  "sample_seed": 20260712,
+  "sample_checkpoint": "model_best.pt",
+  "guidance_scale": 1.0,
+  "decode_mesh": false,
+  "shape_resolution": 512,
+  "shape_max_batch": 32768,
+  "wandb": true,
+  "wandb_project": "drivaer-qwen3v-input-cond",
+  "wandb_entity": "kanydao-epfl",
+  "wandb_run_name": "drivaer3v-inputcond-cfg-rear-0816c",
+  "wandb_tags": "drivaer,qwen3v,input-conditional,channel-concat,rear",
+  "wandb_mode": "online"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/summary.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..ab4226e84e6d0bf87f8697e5d79643fee3ba0c59
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/summary.json
@@ -0,0 +1,10 @@
+{
+  "image_view": "rear",
+  "best_val_shape": 0.827269971370697,
+  "steps": 10000,
+  "train_samples": 3051,
+  "val_samples": 128,
+  "shape_scale": 1.7132293052897125,
+  "image_scale": 1.6045079043256218,
+  "conditioning": "sketch_latent_channel_concat"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/train_history.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/train_history.json
new file mode 100644
index 0000000000000000000000000000000000000000..fe8c563bab4d81873b61acae1a8bd21f9c1c0127
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/train_history.json
@@ -0,0 +1,692 @@
+[
+  {
+    "step": 1,
+    "train_shape": 2.0095300674438477,
+    "lr": 9.999999765596898e-05,
+    "elapsed_sec": 1.1804468631744385
+  },
+  {
+    "step": 1,
+    "val_shape": 2.0060465335845947
+  },
+  {
+    "step": 100,
+    "train_shape": 1.228218674659729,
+    "lr": 9.997656161737219e-05,
+    "elapsed_sec": 29.855958461761475
+  },
+  {
+    "step": 200,
+    "train_shape": 0.9723949432373047,
+    "lr": 9.990626960034288e-05,
+    "elapsed_sec": 57.96175718307495
+  },
+  {
+    "step": 300,
+    "train_shape": 0.893415093421936,
+    "lr": 9.978919331864633e-05,
+    "elapsed_sec": 86.09346628189087
+  },
+  {
+    "step": 400,
+    "train_shape": 0.9012304544448853,
+    "lr": 9.962544831243784e-05,
+    "elapsed_sec": 114.24715280532837
+  },
+  {
+    "step": 500,
+    "train_shape": 0.8704257607460022,
+    "lr": 9.94151961782692e-05,
+    "elapsed_sec": 142.41923069953918
+  },
+  {
+    "step": 500,
+    "val_shape": 0.9327939748764038
+  },
+  {
+    "step": 600,
+    "train_shape": 0.9292115569114685,
+    "lr": 9.915864440961283e-05,
+    "elapsed_sec": 171.60855150222778
+  },
+  {
+    "step": 700,
+    "train_shape": 0.8755108118057251,
+    "lr": 9.885604619209063e-05,
+    "elapsed_sec": 199.80450797080994
+  },
+  {
+    "step": 800,
+    "train_shape": 0.842117428779602,
+    "lr": 9.85077001536101e-05,
+    "elapsed_sec": 228.00592422485352
+  },
+  {
+    "step": 900,
+    "train_shape": 0.8331332206726074,
+    "lr": 9.811395006965495e-05,
+    "elapsed_sec": 256.21257495880127
+  },
+  {
+    "step": 1000,
+    "train_shape": 0.8340949416160583,
+    "lr": 9.767518452401991e-05,
+    "elapsed_sec": 284.43127179145813
+  },
+  {
+    "step": 1000,
+    "val_shape": 0.8968254327774048
+  },
+  {
+    "step": 1100,
+    "train_shape": 0.8789253830909729,
+    "lr": 9.71918365253258e-05,
+    "elapsed_sec": 313.56480169296265
+  },
+  {
+    "step": 1200,
+    "train_shape": 0.8632665872573853,
+    "lr": 9.666438307969207e-05,
+    "elapsed_sec": 341.79078125953674
+  },
+  {
+    "step": 1300,
+    "train_shape": 0.8227857351303101,
+    "lr": 9.60933447199893e-05,
+    "elapsed_sec": 370.00519037246704
+  },
+  {
+    "step": 1400,
+    "train_shape": 0.8397230505943298,
+    "lr": 9.547928499213621e-05,
+    "elapsed_sec": 398.4198133945465
+  },
+  {
+    "step": 1500,
+    "train_shape": 0.8768038749694824,
+    "lr": 9.482280989894765e-05,
+    "elapsed_sec": 426.63349294662476
+  },
+  {
+    "step": 1500,
+    "val_shape": 0.8808047771453857
+  },
+  {
+    "step": 1600,
+    "train_shape": 0.872372031211853,
+    "lr": 9.412456730208378e-05,
+    "elapsed_sec": 455.8579168319702
+  },
+  {
+    "step": 1700,
+    "train_shape": 0.8043794631958008,
+    "lr": 9.338524628268747e-05,
+    "elapsed_sec": 484.0735251903534
+  },
+  {
+    "step": 1800,
+    "train_shape": 0.8579391837120056,
+    "lr": 9.260557646134585e-05,
+    "elapsed_sec": 512.3045823574066
+  },
+  {
+    "step": 1900,
+    "train_shape": 0.8212283849716187,
+    "lr": 9.178632727804172e-05,
+    "elapsed_sec": 540.5643427371979
+  },
+  {
+    "step": 2000,
+    "train_shape": 0.7772009372711182,
+    "lr": 9.092830723281015e-05,
+    "elapsed_sec": 568.7865173816681
+  },
+  {
+    "step": 2000,
+    "val_shape": 0.8706642389297485
+  },
+  {
+    "step": 2100,
+    "train_shape": 0.8735413551330566,
+    "lr": 9.003236308784539e-05,
+    "elapsed_sec": 598.1678133010864
+  },
+  {
+    "step": 2200,
+    "train_shape": 0.8624672889709473,
+    "lr": 8.90993790318501e-05,
+    "elapsed_sec": 626.3989861011505
+  },
+  {
+    "step": 2300,
+    "train_shape": 0.8202571868896484,
+    "lr": 8.813027580744691e-05,
+    "elapsed_sec": 654.634122133255
+  },
+  {
+    "step": 2400,
+    "train_shape": 0.7983925342559814,
+    "lr": 8.712600980251722e-05,
+    "elapsed_sec": 682.9975924491882
+  },
+  {
+    "step": 2500,
+    "train_shape": 0.9186829328536987,
+    "lr": 8.60875721063612e-05,
+    "elapsed_sec": 711.2335669994354
+  },
+  {
+    "step": 2500,
+    "val_shape": 0.8638014197349548
+  },
+  {
+    "step": 2600,
+    "train_shape": 0.8556409478187561,
+    "lr": 8.501598753161292e-05,
+    "elapsed_sec": 740.3820531368256
+  },
+  {
+    "step": 2700,
+    "train_shape": 0.8102497458457947,
+    "lr": 8.391231360287367e-05,
+    "elapsed_sec": 768.6037967205048
+  },
+  {
+    "step": 2800,
+    "train_shape": 0.8117203712463379,
+    "lr": 8.277763951306297e-05,
+    "elapsed_sec": 796.8821601867676
+  },
+  {
+    "step": 2900,
+    "train_shape": 0.7561269402503967,
+    "lr": 8.161308504851662e-05,
+    "elapsed_sec": 825.1238126754761
+  },
+  {
+    "step": 3000,
+    "train_shape": 0.7873923778533936,
+    "lr": 8.04197994838928e-05,
+    "elapsed_sec": 853.6183919906616
+  },
+  {
+    "step": 3000,
+    "val_shape": 0.8585502505302429
+  },
+  {
+    "step": 3100,
+    "train_shape": 0.8356949090957642,
+    "lr": 7.919896044797652e-05,
+    "elapsed_sec": 882.8136887550354
+  },
+  {
+    "step": 3200,
+    "train_shape": 0.7530339360237122,
+    "lr": 7.79517727615026e-05,
+    "elapsed_sec": 911.0472450256348
+  },
+  {
+    "step": 3300,
+    "train_shape": 0.806694746017456,
+    "lr": 7.667946724814286e-05,
+    "elapsed_sec": 939.303288936615
+  },
+  {
+    "step": 3400,
+    "train_shape": 0.8085853457450867,
+    "lr": 7.538329951983173e-05,
+    "elapsed_sec": 967.5274791717529
+  },
+  {
+    "step": 3500,
+    "train_shape": 0.8171709179878235,
+    "lr": 7.406454873762866e-05,
+    "elapsed_sec": 995.748250246048
+  },
+  {
+    "step": 3500,
+    "val_shape": 0.8528761863708496
+  },
+  {
+    "step": 3600,
+    "train_shape": 0.7547885179519653,
+    "lr": 7.272451634934115e-05,
+    "elapsed_sec": 1024.8681826591492
+  },
+  {
+    "step": 3700,
+    "train_shape": 0.800256073474884,
+    "lr": 7.136452480515233e-05,
+    "elapsed_sec": 1053.1148891448975
+  },
+  {
+    "step": 3800,
+    "train_shape": 0.808086097240448,
+    "lr": 6.99859162525224e-05,
+    "elapsed_sec": 1081.351024389267
+  },
+  {
+    "step": 3900,
+    "train_shape": 0.79875248670578,
+    "lr": 6.859005121165148e-05,
+    "elapsed_sec": 1109.5713891983032
+  },
+  {
+    "step": 4000,
+    "train_shape": 0.8082075119018555,
+    "lr": 6.71783072328101e-05,
+    "elapsed_sec": 1137.7910528182983
+  },
+  {
+    "step": 4000,
+    "val_shape": 0.8492899537086487
+  },
+  {
+    "step": 4100,
+    "train_shape": 0.7862443923950195,
+    "lr": 6.575207753686356e-05,
+    "elapsed_sec": 1166.9274318218231
+  },
+  {
+    "step": 4200,
+    "train_shape": 0.7889789938926697,
+    "lr": 6.431276964033111e-05,
+    "elapsed_sec": 1195.1438219547272
+  },
+  {
+    "step": 4300,
+    "train_shape": 0.7743031978607178,
+    "lr": 6.286180396633663e-05,
+    "elapsed_sec": 1223.3968896865845
+  },
+  {
+    "step": 4400,
+    "train_shape": 0.7786076068878174,
+    "lr": 6.140061244282311e-05,
+    "elapsed_sec": 1251.6361725330353
+  },
+  {
+    "step": 4500,
+    "train_shape": 0.7773829102516174,
+    "lr": 5.9930637089412255e-05,
+    "elapsed_sec": 1280.0734293460846
+  },
+  {
+    "step": 4500,
+    "val_shape": 0.8454297780990601
+  },
+  {
+    "step": 4600,
+    "train_shape": 0.8091238737106323,
+    "lr": 5.845332859430572e-05,
+    "elapsed_sec": 1309.250943183899
+  },
+  {
+    "step": 4700,
+    "train_shape": 0.6957094669342041,
+    "lr": 5.697014488263067e-05,
+    "elapsed_sec": 1337.6122410297394
+  },
+  {
+    "step": 4800,
+    "train_shape": 0.795627236366272,
+    "lr": 5.548254967764354e-05,
+    "elapsed_sec": 1365.8372762203217
+  },
+  {
+    "step": 4900,
+    "train_shape": 0.799580991268158,
+    "lr": 5.3992011056212214e-05,
+    "elapsed_sec": 1394.0621004104614
+  },
+  {
+    "step": 5000,
+    "train_shape": 0.7656121850013733,
+    "lr": 5.250000000000107e-05,
+    "elapsed_sec": 1422.285757780075
+  },
+  {
+    "step": 5000,
+    "val_shape": 0.8414366245269775
+  },
+  {
+    "step": 5100,
+    "train_shape": 0.7967846393585205,
+    "lr": 5.100798894378995e-05,
+    "elapsed_sec": 1453.1616003513336
+  },
+  {
+    "step": 5200,
+    "train_shape": 0.789453387260437,
+    "lr": 4.951745032235864e-05,
+    "elapsed_sec": 1481.3819515705109
+  },
+  {
+    "step": 5300,
+    "train_shape": 0.7423689961433411,
+    "lr": 4.8029855117371535e-05,
+    "elapsed_sec": 1509.6020872592926
+  },
+  {
+    "step": 5400,
+    "train_shape": 0.8243147730827332,
+    "lr": 4.654667140569645e-05,
+    "elapsed_sec": 1537.8271026611328
+  },
+  {
+    "step": 5500,
+    "train_shape": 0.7648144364356995,
+    "lr": 4.506936291058995e-05,
+    "elapsed_sec": 1566.0781779289246
+  },
+  {
+    "step": 5500,
+    "val_shape": 0.8390533924102783
+  },
+  {
+    "step": 5600,
+    "train_shape": 0.8123018145561218,
+    "lr": 4.359938755717899e-05,
+    "elapsed_sec": 1595.2151091098785
+  },
+  {
+    "step": 5700,
+    "train_shape": 0.7666544914245605,
+    "lr": 4.213819603366514e-05,
+    "elapsed_sec": 1623.4312975406647
+  },
+  {
+    "step": 5800,
+    "train_shape": 0.8347450494766235,
+    "lr": 4.0687230359670296e-05,
+    "elapsed_sec": 1651.675840139389
+  },
+  {
+    "step": 5900,
+    "train_shape": 0.8190369009971619,
+    "lr": 3.924792246313749e-05,
+    "elapsed_sec": 1680.05211353302
+  },
+  {
+    "step": 6000,
+    "train_shape": 0.7392253875732422,
+    "lr": 3.7821692767190845e-05,
+    "elapsed_sec": 1708.2815330028534
+  },
+  {
+    "step": 6000,
+    "val_shape": 0.8362010717391968
+  },
+  {
+    "step": 6100,
+    "train_shape": 0.7455916404724121,
+    "lr": 3.640994878834947e-05,
+    "elapsed_sec": 1737.4430146217346
+  },
+  {
+    "step": 6200,
+    "train_shape": 0.7954226136207581,
+    "lr": 3.501408374747862e-05,
+    "elapsed_sec": 1765.6671755313873
+  },
+  {
+    "step": 6300,
+    "train_shape": 0.754315197467804,
+    "lr": 3.363547519484887e-05,
+    "elapsed_sec": 1793.89310836792
+  },
+  {
+    "step": 6400,
+    "train_shape": 0.7783094644546509,
+    "lr": 3.227548365066013e-05,
+    "elapsed_sec": 1822.1562638282776
+  },
+  {
+    "step": 6500,
+    "train_shape": 0.7365042567253113,
+    "lr": 3.0935451262372706e-05,
+    "elapsed_sec": 1850.3798377513885
+  },
+  {
+    "step": 6500,
+    "val_shape": 0.8343663215637207
+  },
+  {
+    "step": 6600,
+    "train_shape": 0.7614246606826782,
+    "lr": 2.9616700480169673e-05,
+    "elapsed_sec": 1879.5329496860504
+  },
+  {
+    "step": 6700,
+    "train_shape": 0.793758749961853,
+    "lr": 2.8320532751858457e-05,
+    "elapsed_sec": 1907.770367383957
+  },
+  {
+    "step": 6800,
+    "train_shape": 0.7742476463317871,
+    "lr": 2.7048227238498715e-05,
+    "elapsed_sec": 1935.994249343872
+  },
+  {
+    "step": 6900,
+    "train_shape": 0.7787187099456787,
+    "lr": 2.5801039552024805e-05,
+    "elapsed_sec": 1964.218480348587
+  },
+  {
+    "step": 7000,
+    "train_shape": 0.8083285689353943,
+    "lr": 2.458020051610849e-05,
+    "elapsed_sec": 1992.5605449676514
+  },
+  {
+    "step": 7000,
+    "val_shape": 0.8320755958557129
+  },
+  {
+    "step": 7100,
+    "train_shape": 0.7809754610061646,
+    "lr": 2.3386914951484544e-05,
+    "elapsed_sec": 2021.776950597763
+  },
+  {
+    "step": 7200,
+    "train_shape": 0.7889120578765869,
+    "lr": 2.222236048693809e-05,
+    "elapsed_sec": 2050.0092573165894
+  },
+  {
+    "step": 7300,
+    "train_shape": 0.7944762706756592,
+    "lr": 2.1087686397127328e-05,
+    "elapsed_sec": 2078.438346862793
+  },
+  {
+    "step": 7400,
+    "train_shape": 0.8216014504432678,
+    "lr": 1.998401246838802e-05,
+    "elapsed_sec": 2106.806799888611
+  },
+  {
+    "step": 7500,
+    "train_shape": 0.7835626602172852,
+    "lr": 1.891242789363976e-05,
+    "elapsed_sec": 2135.045908689499
+  },
+  {
+    "step": 7500,
+    "val_shape": 0.8311185836791992
+  },
+  {
+    "step": 7600,
+    "train_shape": 0.767005205154419,
+    "lr": 1.787399019748373e-05,
+    "elapsed_sec": 2164.1872577667236
+  },
+  {
+    "step": 7700,
+    "train_shape": 0.7195855975151062,
+    "lr": 1.6869724192553968e-05,
+    "elapsed_sec": 2192.4242191314697
+  },
+  {
+    "step": 7800,
+    "train_shape": 0.7767179012298584,
+    "lr": 1.5900620968150808e-05,
+    "elapsed_sec": 2220.643128633499
+  },
+  {
+    "step": 7900,
+    "train_shape": 0.8137301206588745,
+    "lr": 1.496763691215551e-05,
+    "elapsed_sec": 2248.8795461654663
+  },
+  {
+    "step": 8000,
+    "train_shape": 0.819430947303772,
+    "lr": 1.4071692767190806e-05,
+    "elapsed_sec": 2277.100219964981
+  },
+  {
+    "step": 8000,
+    "val_shape": 0.8293596506118774
+  },
+  {
+    "step": 8100,
+    "train_shape": 0.7993512153625488,
+    "lr": 1.3213672721959122e-05,
+    "elapsed_sec": 2306.2877933979034
+  },
+  {
+    "step": 8200,
+    "train_shape": 0.7903712391853333,
+    "lr": 1.2394423538655062e-05,
+    "elapsed_sec": 2334.505030155182
+  },
+  {
+    "step": 8300,
+    "train_shape": 0.8048752546310425,
+    "lr": 1.1614753717313381e-05,
+    "elapsed_sec": 2362.7529957294464
+  },
+  {
+    "step": 8400,
+    "train_shape": 0.789883017539978,
+    "lr": 1.0875432697917097e-05,
+    "elapsed_sec": 2390.9953083992004
+  },
+  {
+    "step": 8500,
+    "train_shape": 0.741385817527771,
+    "lr": 1.0177190101053074e-05,
+    "elapsed_sec": 2419.232983827591
+  },
+  {
+    "step": 8500,
+    "val_shape": 0.8288300633430481
+  },
+  {
+    "step": 8600,
+    "train_shape": 0.7536501884460449,
+    "lr": 9.520715007864551e-06,
+    "elapsed_sec": 2448.4082369804382
+  },
+  {
+    "step": 8700,
+    "train_shape": 0.7417742013931274,
+    "lr": 8.906655280011289e-06,
+    "elapsed_sec": 2476.6214344501495
+  },
+  {
+    "step": 8800,
+    "train_shape": 0.779025673866272,
+    "lr": 8.335616920308386e-06,
+    "elapsed_sec": 2504.835286140442
+  },
+  {
+    "step": 8900,
+    "train_shape": 0.7363979816436768,
+    "lr": 7.808163474674577e-06,
+    "elapsed_sec": 2533.050084590912
+  },
+  {
+    "step": 9000,
+    "train_shape": 0.790111780166626,
+    "lr": 7.32481547598044e-06,
+    "elapsed_sec": 2561.3947768211365
+  },
+  {
+    "step": 9000,
+    "val_shape": 0.8290339708328247
+  },
+  {
+    "step": 9100,
+    "train_shape": 0.7343631982803345,
+    "lr": 6.886049930345396e-06,
+    "elapsed_sec": 2589.8126831054688
+  },
+  {
+    "step": 9200,
+    "train_shape": 0.7906165719032288,
+    "lr": 6.492299846390179e-06,
+    "elapsed_sec": 2618.259387254715
+  },
+  {
+    "step": 9300,
+    "train_shape": 0.7655789256095886,
+    "lr": 6.143953807909618e-06,
+    "elapsed_sec": 2646.588972091675
+  },
+  {
+    "step": 9400,
+    "train_shape": 0.7721428275108337,
+    "lr": 5.841355590387376e-06,
+    "elapsed_sec": 2674.812264919281
+  },
+  {
+    "step": 9500,
+    "train_shape": 0.7720650434494019,
+    "lr": 5.5848038217310204e-06,
+    "elapsed_sec": 2703.029899120331
+  },
+  {
+    "step": 9500,
+    "val_shape": 0.8274837136268616
+  },
+  {
+    "step": 9600,
+    "train_shape": 0.8399871587753296,
+    "lr": 5.374551687562344e-06,
+    "elapsed_sec": 2732.179806947708
+  },
+  {
+    "step": 9700,
+    "train_shape": 0.7529326677322388,
+    "lr": 5.2108066813537186e-06,
+    "elapsed_sec": 2760.3982021808624
+  },
+  {
+    "step": 9800,
+    "train_shape": 0.777104377746582,
+    "lr": 5.093730399657109e-06,
+    "elapsed_sec": 2788.60946393013
+  },
+  {
+    "step": 9900,
+    "train_shape": 0.7243473529815674,
+    "lr": 5.023438382627753e-06,
+    "elapsed_sec": 2816.8275718688965
+  },
+  {
+    "step": 10000,
+    "train_shape": 0.7834863662719727,
+    "lr": 5e-06,
+    "elapsed_sec": 2845.0473029613495
+  },
+  {
+    "step": 10000,
+    "val_shape": 0.827269971370697
+  }
+]
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..47bafd4a37461cf4c72d1976fc245959f3d7774d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:48.411910312Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:48.732046223Z","level":"INFO","msg":"stream: created new stream","id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.732109961Z","level":"INFO","msg":"handler: started","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.758606289Z","level":"INFO","msg":"stream: started","id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.758626276Z","level":"INFO","msg":"writer: started","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.758627478Z","level":"INFO","msg":"sender: started","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.301769911Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:20.54314005Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:20.547726522Z","level":"INFO","msg":"stream: closing","id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.547738915Z","level":"INFO","msg":"handler: closed","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.549211043Z","level":"INFO","msg":"sender: closed","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.549222915Z","level":"INFO","msg":"stream: closed","id":"e6xhvf8a"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/debug.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..6d0b1b75735e2958246f328b9cd9f323978659d1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug.log
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug-internal.log
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz', 'image_view': 'rear', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.1, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-cfg-rear-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,rear', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:48,403 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:48,409 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:48,411 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:48,423 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:49,171 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:49,642 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:49,642 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:49,642 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:49,646 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:19,374 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/e6xhvf8a
+2026-08-16 20:03:19,375 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:19,375 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:19,375 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:20,545 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/config.yaml b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..4adf4983333c4b9e5d66907fc02cc7bd6d28dbf9
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/config.yaml
@@ -0,0 +1,172 @@
+_wandb:
+    value:
+        cli_version: 0.22.3
+        e:
+            qi7e64wti7rpr52lymi321pqfm8hydxy:
+                args:
+                    - --mode
+                    - train
+                    - --image-view
+                    - rear
+                    - --paired-latents
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz
+                    - --out-dir
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear
+                    - --steps
+                    - "10000"
+                    - --batch-size
+                    - "128"
+                    - --lr
+                    - "1e-4"
+                    - --weight-decay
+                    - "0.01"
+                    - --seed
+                    - "20260526"
+                    - --pos-embed
+                    - 2d
+                    - --cond-dropout
+                    - "0.1"
+                    - --resume
+                    - --wandb
+                    - --wandb-run-name
+                    - drivaer3v-inputcond-cfg-rear-0816c
+                    - --wandb-tags
+                    - drivaer,qwen3v,input-conditional,channel-concat,rear
+                codePath: code/ICLR/customized-design/input_cond/train_cond_flow.py
+                codePathLocal: train_cond_flow.py
+                cpu_count: 32
+                cpu_count_logical: 64
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "7679361679360"
+                        used: "2855216566272"
+                email: 1193414549@qq.com
+                executable: /mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python
+                git:
+                    commit: 81d95989e1279e432c4145933bf9ce02c3f70f77
+                gpu: NVIDIA A100-SXM4-40GB
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "42949672960"
+                      name: NVIDIA A100-SXM4-40GB
+                      uuid: GPU-d1fb9f59-34c2-ed7f-8e2f-be250e503c95
+                host: inputcond-cfg-rear-a100-40g-0816c-0-0
+                memory:
+                    total: "540742438912"
+                os: Linux-6.8.0-124-generic-x86_64-with-glibc2.35
+                program: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py
+                python: CPython 3.10.18
+                root: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear/wandb
+                startedAt: "2026-08-16T19:15:48.178344Z"
+                writerId: qi7e64wti7rpr52lymi321pqfm8hydxy
+        m: []
+        python_version: 3.10.18
+        t:
+            "1":
+                - 1
+                - 105
+            "2":
+                - 1
+                - 105
+            "3":
+                - 2
+                - 13
+                - 15
+                - 16
+                - 61
+                - 62
+            "4": 3.10.18
+            "5": 0.22.3
+            "12": 0.22.3
+            "13": linux-x86_64
+attention_backend:
+    value: sdpa
+batch_size:
+    value: 128
+cond_dropout:
+    value: 0.1
+decode_mesh:
+    value: false
+depth:
+    value: 12
+dropout:
+    value: 0
+eval_every:
+    value: 500
+eval_noise_seed:
+    value: 20260712
+gradient_checkpointing:
+    value: false
+guidance_scale:
+    value: 1
+head_dim:
+    value: 64
+hidden_size:
+    value: 512
+image_scale:
+    value: 0
+image_view:
+    value: rear
+log_every:
+    value: 100
+lr:
+    value: 0.0001
+max_val:
+    value: 128
+mlp_ratio:
+    value: 4
+mode:
+    value: train
+num_heads:
+    value: 8
+num_workers:
+    value: 2
+out_dir:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear
+paired_latents:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz
+pos_embed:
+    value: 2d
+resume:
+    value: true
+sample_checkpoint:
+    value: model_best.pt
+sample_count:
+    value: 8
+sample_seed:
+    value: 20260712
+sample_steps:
+    value: 50
+save_every:
+    value: 5000
+seed:
+    value: 20260526
+shape_max_batch:
+    value: 32768
+shape_resolution:
+    value: 512
+shape_scale:
+    value: 0
+steps:
+    value: 10000
+val_fraction:
+    value: 0.05
+val_overlap_train:
+    value: true
+wandb:
+    value: true
+wandb_entity:
+    value: kanydao-epfl
+wandb_mode:
+    value: online
+wandb_project:
+    value: drivaer-qwen3v-input-cond
+wandb_run_name:
+    value: drivaer3v-inputcond-cfg-rear-0816c
+wandb_tags:
+    value: drivaer,qwen3v,input-conditional,channel-concat,rear
+weight_decay:
+    value: 0.01
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/output.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..312994d64cc77fb7a34c25c4257963512da8e395
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/output.log
@@ -0,0 +1,123 @@
+[wandb] project=drivaer-qwen3v-input-cond run=drivaer3v-inputcond-cfg-rear-0816c mode=online
+[train] step=000001 loss=2.009530 lr=1.00e-04
+[val] step=000001 shape=2.006047
+[train] step=000100 loss=1.228219 lr=1.00e-04
+[train] step=000200 loss=0.972395 lr=9.99e-05
+[train] step=000300 loss=0.893415 lr=9.98e-05
+[train] step=000400 loss=0.901230 lr=9.96e-05
+[train] step=000500 loss=0.870426 lr=9.94e-05
+[val] step=000500 shape=0.932794
+[train] step=000600 loss=0.929212 lr=9.92e-05
+[train] step=000700 loss=0.875511 lr=9.89e-05
+[train] step=000800 loss=0.842117 lr=9.85e-05
+[train] step=000900 loss=0.833133 lr=9.81e-05
+[train] step=001000 loss=0.834095 lr=9.77e-05
+[val] step=001000 shape=0.896825
+[train] step=001100 loss=0.878925 lr=9.72e-05
+[train] step=001200 loss=0.863267 lr=9.67e-05
+[train] step=001300 loss=0.822786 lr=9.61e-05
+[train] step=001400 loss=0.839723 lr=9.55e-05
+[train] step=001500 loss=0.876804 lr=9.48e-05
+[val] step=001500 shape=0.880805
+[train] step=001600 loss=0.872372 lr=9.41e-05
+[train] step=001700 loss=0.804379 lr=9.34e-05
+[train] step=001800 loss=0.857939 lr=9.26e-05
+[train] step=001900 loss=0.821228 lr=9.18e-05
+[train] step=002000 loss=0.777201 lr=9.09e-05
+[val] step=002000 shape=0.870664
+[train] step=002100 loss=0.873541 lr=9.00e-05
+[train] step=002200 loss=0.862467 lr=8.91e-05
+[train] step=002300 loss=0.820257 lr=8.81e-05
+[train] step=002400 loss=0.798393 lr=8.71e-05
+[train] step=002500 loss=0.918683 lr=8.61e-05
+[val] step=002500 shape=0.863801
+[train] step=002600 loss=0.855641 lr=8.50e-05
+[train] step=002700 loss=0.810250 lr=8.39e-05
+[train] step=002800 loss=0.811720 lr=8.28e-05
+[train] step=002900 loss=0.756127 lr=8.16e-05
+[train] step=003000 loss=0.787392 lr=8.04e-05
+[val] step=003000 shape=0.858550
+[train] step=003100 loss=0.835695 lr=7.92e-05
+[train] step=003200 loss=0.753034 lr=7.80e-05
+[train] step=003300 loss=0.806695 lr=7.67e-05
+[train] step=003400 loss=0.808585 lr=7.54e-05
+[train] step=003500 loss=0.817171 lr=7.41e-05
+[val] step=003500 shape=0.852876
+[train] step=003600 loss=0.754789 lr=7.27e-05
+[train] step=003700 loss=0.800256 lr=7.14e-05
+[train] step=003800 loss=0.808086 lr=7.00e-05
+[train] step=003900 loss=0.798752 lr=6.86e-05
+[train] step=004000 loss=0.808208 lr=6.72e-05
+[val] step=004000 shape=0.849290
+[train] step=004100 loss=0.786244 lr=6.58e-05
+[train] step=004200 loss=0.788979 lr=6.43e-05
+[train] step=004300 loss=0.774303 lr=6.29e-05
+[train] step=004400 loss=0.778608 lr=6.14e-05
+[train] step=004500 loss=0.777383 lr=5.99e-05
+[val] step=004500 shape=0.845430
+[train] step=004600 loss=0.809124 lr=5.85e-05
+[train] step=004700 loss=0.695709 lr=5.70e-05
+[train] step=004800 loss=0.795627 lr=5.55e-05
+[train] step=004900 loss=0.799581 lr=5.40e-05
+[train] step=005000 loss=0.765612 lr=5.25e-05
+[val] step=005000 shape=0.841437
+[train] step=005100 loss=0.796785 lr=5.10e-05
+[train] step=005200 loss=0.789453 lr=4.95e-05
+[train] step=005300 loss=0.742369 lr=4.80e-05
+[train] step=005400 loss=0.824315 lr=4.65e-05
+[train] step=005500 loss=0.764814 lr=4.51e-05
+[val] step=005500 shape=0.839053
+[train] step=005600 loss=0.812302 lr=4.36e-05
+[train] step=005700 loss=0.766654 lr=4.21e-05
+[train] step=005800 loss=0.834745 lr=4.07e-05
+[train] step=005900 loss=0.819037 lr=3.92e-05
+[train] step=006000 loss=0.739225 lr=3.78e-05
+[val] step=006000 shape=0.836201
+[train] step=006100 loss=0.745592 lr=3.64e-05
+[train] step=006200 loss=0.795423 lr=3.50e-05
+[train] step=006300 loss=0.754315 lr=3.36e-05
+[train] step=006400 loss=0.778309 lr=3.23e-05
+[train] step=006500 loss=0.736504 lr=3.09e-05
+[val] step=006500 shape=0.834366
+[train] step=006600 loss=0.761425 lr=2.96e-05
+[train] step=006700 loss=0.793759 lr=2.83e-05
+[train] step=006800 loss=0.774248 lr=2.70e-05
+[train] step=006900 loss=0.778719 lr=2.58e-05
+[train] step=007000 loss=0.808329 lr=2.46e-05
+[val] step=007000 shape=0.832076
+[train] step=007100 loss=0.780975 lr=2.34e-05
+[train] step=007200 loss=0.788912 lr=2.22e-05
+[train] step=007300 loss=0.794476 lr=2.11e-05
+[train] step=007400 loss=0.821601 lr=2.00e-05
+[train] step=007500 loss=0.783563 lr=1.89e-05
+[val] step=007500 shape=0.831119
+[train] step=007600 loss=0.767005 lr=1.79e-05
+[train] step=007700 loss=0.719586 lr=1.69e-05
+[train] step=007800 loss=0.776718 lr=1.59e-05
+[train] step=007900 loss=0.813730 lr=1.50e-05
+[train] step=008000 loss=0.819431 lr=1.41e-05
+[val] step=008000 shape=0.829360
+[train] step=008100 loss=0.799351 lr=1.32e-05
+[train] step=008200 loss=0.790371 lr=1.24e-05
+[train] step=008300 loss=0.804875 lr=1.16e-05
+[train] step=008400 loss=0.789883 lr=1.09e-05
+[train] step=008500 loss=0.741386 lr=1.02e-05
+[val] step=008500 shape=0.828830
+[train] step=008600 loss=0.753650 lr=9.52e-06
+[train] step=008700 loss=0.741774 lr=8.91e-06
+[train] step=008800 loss=0.779026 lr=8.34e-06
+[train] step=008900 loss=0.736398 lr=7.81e-06
+[train] step=009000 loss=0.790112 lr=7.32e-06
+[val] step=009000 shape=0.829034
+[train] step=009100 loss=0.734363 lr=6.89e-06
+[train] step=009200 loss=0.790617 lr=6.49e-06
+[train] step=009300 loss=0.765579 lr=6.14e-06
+[train] step=009400 loss=0.772143 lr=5.84e-06
+[train] step=009500 loss=0.772065 lr=5.58e-06
+[val] step=009500 shape=0.827484
+[train] step=009600 loss=0.839987 lr=5.37e-06
+[train] step=009700 loss=0.752933 lr=5.21e-06
+[train] step=009800 loss=0.777104 lr=5.09e-06
+[train] step=009900 loss=0.724347 lr=5.02e-06
+[train] step=010000 loss=0.783486 lr=5.00e-06
+[val] step=010000 shape=0.827270
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/requirements.txt b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..2e10c9ab6bc99041706780299f9e3e5035f34676
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/requirements.txt
@@ -0,0 +1,302 @@
+shellingham==1.5.4
+importlib_metadata==8.7.0
+websocket-client==1.8.0
+rfc3986-validator==0.1.1
+aiohappyeyeballs==2.6.1
+tensorboardX==2.6.2.2
+matplotlib==3.8.4
+zstandard==0.23.0
+gradio_client==1.4.2
+nbclient==0.10.2
+pyaml==25.7.0
+argparse==1.4.0
+h11==0.16.0
+scooby==0.10.1
+pexpect==4.9.0
+jedi==0.19.2
+scikit-optimize==0.10.2
+omegaconf==2.3.0
+tomlkit==0.12.0
+yarl==1.20.1
+point-cloud-utils==0.34.0
+cffi==1.17.1
+asttokens==3.0.0
+webencodings==0.5.1
+imageio==2.34.1
+attrs==25.3.0
+hydra-core==1.3.2
+datasets==2.19.0
+pyquaternion==0.9.9
+trame-common==1.0.0
+wslink==2.3.4
+Pygments==2.19.2
+pydantic==2.11.7
+regex==2024.11.6
+timm==0.9.16
+ipykernel==6.29.5
+prompt_toolkit==3.0.51
+torchdiffeq==0.2.5
+pybind11==2.13.6
+debugpy==1.8.14
+pytz==2025.2
+transformers==4.40.1
+jupyterlab_pygments==0.3.0
+dash==3.1.1
+psutil==5.9.8
+tqdm==4.66.2
+multidict==6.6.3
+setproctitle==1.3.6
+nbformat==5.10.4
+prometheus_client==0.22.1
+jupyter_core==5.8.1
+gradio==5.5.0
+multiprocess==0.70.16
+fastapi==0.115.14
+simpervisor==1.0.0
+comm==0.2.2
+widgetsnbextension==4.0.14
+webcolors==24.11.1
+sentry-sdk==2.32.0
+tornado==6.5.1
+defusedxml==0.7.1
+Markdown==3.8.2
+numba==0.61.2
+pyarrow-hotfix==0.7
+trame-server==3.5.0
+nbconvert==7.16.6
+websockets==12.0
+semantic-version==2.10.0
+bpy==4.0.0
+smmap==5.0.2
+grpcio==1.73.1
+referencing==0.36.2
+annotated-types==0.7.0
+stack-data==0.6.3
+vtk==9.4.2
+safehttpx==0.1.6
+GitPython==3.1.44
+ipywidgets==8.1.7
+platformdirs==4.3.8
+tzdata==2025.2
+typer==0.16.0
+click==8.2.1
+networkx==3.3
+fsspec==2024.3.1
+plotly==6.2.0
+isoduration==20.11.0
+joblib==1.4.0
+aiosignal==1.4.0
+jsonschema-specifications==2025.4.1
+six==1.17.0
+pyzmq==27.0.0
+lazy_loader==0.4
+typeguard==2.13.3
+markdown-it-py==3.0.0
+httpx==0.28.1
+fqdn==1.5.1
+threadpoolctl==3.6.0
+orjson==3.10.18
+docker-pycreds==0.4.0
+jsonschema==4.24.0
+tensorboard-data-server==0.7.2
+nvidia-cufft-cu12==11.0.2.54
+aiofiles==23.2.1
+rpds-py==0.26.0
+Cython==3.1.2
+python-multipart==0.0.12
+nvidia-cuda-runtime-cu12==12.1.105
+huggingface-hub==0.26.2
+frozenlist==1.7.0
+onnxruntime==1.22.0
+cycler==0.12.1
+tifffile==2025.5.10
+safetensors==0.4.3
+nest-asyncio==1.6.0
+parso==0.8.4
+PyMCubes==0.1.4
+nvidia-nvtx-cu12==12.1.105
+pydub==0.25.1
+packaging==25.0
+ipython==8.37.0
+uvicorn==0.35.0
+pyarrow==20.0.0
+rich==14.0.0
+executing==2.2.0
+jupyter_server_terminals==0.5.3
+typing_extensions==4.12.2
+typing_extensions==4.15.0
+tensorboard==2.16.2
+argon2-cffi-bindings==21.2.0
+wcwidth==0.2.13
+typing-inspection==0.4.1
+pure_eval==0.2.3
+trimesh==4.3.2
+nvidia-cudnn-cu12==9.1.0.70
+humanfriendly==10.0
+Werkzeug==3.1.3
+decorator==5.2.1
+nvidia-cuda-nvrtc-cu12==12.1.105
+aiohttp==3.12.13
+uri-template==1.3.0
+anyio==4.9.0
+fastjsonschema==2.21.1
+nvidia-curand-cu12==10.3.2.106
+zipp==3.23.0
+rtree==1.4.1
+kiwisolver==1.4.8
+Flask==3.1.1
+nvidia-cuda-cupti-cu12==12.1.105
+pysdf==0.1.9
+sniffio==1.3.1
+itsdangerous==2.2.0
+rfc3339-validator==0.1.4
+trame==3.10.2
+open3d==0.18.0
+scikit-learn==1.7.0
+mistune==3.1.3
+diffusers==0.31.0
+absl-py==2.3.1
+cubvh==0.1.0
+trame-client==3.9.1
+einops==0.8.0
+plyfile==1.0.3
+bleach==6.2.0
+ptyprocess==0.7.0
+pycparser==2.22
+llvmlite==0.44.0
+addict==2.4.0
+diso==0.1.4
+Send2Trash==1.8.3
+more-itertools==10.7.0
+scikit-image==0.23.2
+propcache==0.3.2
+pandas==2.2.2
+flatbuffers==25.2.10
+gitdb==4.0.12
+jaxtyping==0.2.28
+soupsieve==2.7
+mdurl==0.1.2
+protobuf==5.29.5
+fonttools==4.58.5
+coloredlogs==15.0.1
+spaces==0.28.3
+jupyter_server==2.16.0
+trame-vtk==2.9.0
+terminado==0.18.1
+overrides==7.7.0
+pandocfilters==1.5.1
+nvidia-cusolver-cu12==11.4.5.107
+trame-vuetify==3.0.1
+async-timeout==5.0.1
+blinker==1.9.0
+exceptiongroup==1.3.0
+xxhash==3.5.0
+python-json-logger==3.3.0
+pyvista==0.45.2
+python-dateutil==2.9.0.post0
+ConfigArgParse==1.7.1
+PyMatting==1.1.14
+fpsample==0.3.3
+traitlets==5.14.3
+msgpack==1.1.1
+argon2-cffi==25.1.0
+tinycss2==1.4.0
+retrying==1.4.0
+jsonpointer==3.0.0
+matplotlib-inline==0.1.7
+jupyter_client==8.6.3
+contourpy==1.3.2
+lightning-utilities==0.11.2
+pydantic_core==2.33.2
+httpcore==1.0.9
+narwhals==1.46.0
+nvidia-cublas-cu12==12.1.3.1
+jupyterlab_widgets==3.0.15
+antlr4-python3-runtime==4.9.3
+pyparsing==3.1.2
+rembg==2.0.59
+ffmpy==0.6.0
+tokenizers==0.19.1
+beautifulsoup4==4.13.4
+pooch==1.8.2
+ninja==1.11.1.4
+jupyter_server_proxy==4.4.0
+ruff==0.12.2
+types-python-dateutil==2.9.0.20250708
+dill==0.3.8
+pytorch-lightning==2.2.4
+jupyter-events==0.12.0
+starlette==0.46.2
+arrow==1.3.0
+setuptools==78.1.1
+wheel==0.45.1
+pip==25.1
+mkl-service==2.5.2
+mpi4py==4.0.3
+h5py==3.14.0
+mkl_fft==2.1.1
+mkl_random==1.3.0
+mcubes==0.1.6
+wandb==0.22.3
+certifi==2026.1.4
+mpmath==1.3.0
+triton==3.1.0
+sympy==1.13.1
+nvidia-nccl-cu12==2.21.5
+nvidia-nvjitlink-cu12==12.9.86
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nvtx-cu11==11.8.86
+nvidia-nccl-cu11==2.21.5
+nvidia-cusparse-cu11==11.7.5.86
+nvidia-curand-cu11==10.3.0.86
+nvidia-cufft-cu11==10.9.0.58
+nvidia-cuda-runtime-cu11==11.8.89
+nvidia-cuda-nvrtc-cu11==11.8.89
+nvidia-cuda-cupti-cu11==11.8.87
+nvidia-cublas-cu11==11.11.3.6
+nvidia-cusolver-cu11==11.4.1.48
+nvidia-cudnn-cu11==9.1.0.70
+torch==2.5.1+cu118
+torchvision==0.20.1+cu118
+torchaudio==2.5.1+cu118
+PyYAML==6.0.3
+pillow==12.1.0
+torchmetrics==1.8.2
+accelerate==1.12.0
+urllib3==2.6.3
+idna==3.11
+charset-normalizer==3.4.4
+requests==2.32.5
+filelock==3.20.3
+torch_cluster==1.6.3+pt25cu118
+numpy==1.26.4
+opencv-python-headless==4.9.0.80
+PyOpenGL-accelerate==3.1.10
+PyOpenGL==3.1.0
+pyglet==2.1.14
+freetype-py==2.5.1
+pyrender==0.1.45
+scipy==1.15.3
+memory-profiler==0.61.0
+pyamg==5.3.0
+meshio==5.3.5
+torch-fem==0.6.2
+torch-geometric==2.7.0
+MarkupSafe==3.0.3
+Jinja2==3.1.6
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/wandb-metadata.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..75dd6856f8878dbfcdbb5cb61fa43e0911a22e25
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/wandb-metadata.json
@@ -0,0 +1,69 @@
+{
+  "os":  "Linux-6.8.0-124-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.10.18",
+  "startedAt":  "2026-08-16T19:15:48.178344Z",
+  "args":  [
+    "--mode",
+    "train",
+    "--image-view",
+    "rear",
+    "--paired-latents",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz",
+    "--out-dir",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear",
+    "--steps",
+    "10000",
+    "--batch-size",
+    "128",
+    "--lr",
+    "1e-4",
+    "--weight-decay",
+    "0.01",
+    "--seed",
+    "20260526",
+    "--pos-embed",
+    "2d",
+    "--cond-dropout",
+    "0.1",
+    "--resume",
+    "--wandb",
+    "--wandb-run-name",
+    "drivaer3v-inputcond-cfg-rear-0816c",
+    "--wandb-tags",
+    "drivaer,qwen3v,input-conditional,channel-concat,rear"
+  ],
+  "program":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePath":  "code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePathLocal":  "train_cond_flow.py",
+  "git":  {
+    "commit":  "81d95989e1279e432c4145933bf9ce02c3f70f77"
+  },
+  "email":  "1193414549@qq.com",
+  "root":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear/wandb",
+  "host":  "inputcond-cfg-rear-a100-40g-0816c-0-0",
+  "executable":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python",
+  "cpu_count":  32,
+  "cpu_count_logical":  64,
+  "gpu":  "NVIDIA A100-SXM4-40GB",
+  "gpu_count":  1,
+  "disk":  {
+    "/":  {
+      "total":  "7679361679360",
+      "used":  "2855216566272"
+    }
+  },
+  "memory":  {
+    "total":  "540742438912"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A100-SXM4-40GB",
+      "memoryTotal":  "42949672960",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-d1fb9f59-34c2-ed7f-8e2f-be250e503c95"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "qi7e64wti7rpr52lymi321pqfm8hydxy"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/wandb-summary.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..6bbe02dae23d42aa4bd8355971a0e3963680f8c7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/files/wandb-summary.json
@@ -0,0 +1 @@
+{"best_val_shape":0.827269971370697,"shape_scale":1.7132293052897125,"_timestamp":1.7869105948920743e+09,"_step":10000,"time/elapsed_sec":2845.0473029613495,"train/lr":5e-06,"train/shape":0.7834863662719727,"val/shape":0.827269971370697,"_wandb":{"runtime":2850},"val_samples":128,"_runtime":2850.203776908,"image_view":"rear","steps":10000,"image_scale":1.6045079043256218,"train_samples":3051,"conditioning":"sketch_latent_channel_concat"}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..47bafd4a37461cf4c72d1976fc245959f3d7774d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:48.411910312Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:48.732046223Z","level":"INFO","msg":"stream: created new stream","id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.732109961Z","level":"INFO","msg":"handler: started","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.758606289Z","level":"INFO","msg":"stream: started","id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.758626276Z","level":"INFO","msg":"writer: started","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T19:15:48.758627478Z","level":"INFO","msg":"sender: started","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.301769911Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:20.54314005Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:20.547726522Z","level":"INFO","msg":"stream: closing","id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.547738915Z","level":"INFO","msg":"handler: closed","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.549211043Z","level":"INFO","msg":"sender: closed","stream_id":"e6xhvf8a"}
+{"time":"2026-08-16T20:03:20.549222915Z","level":"INFO","msg":"stream: closed","id":"e6xhvf8a"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..6d0b1b75735e2958246f328b9cd9f323978659d1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug.log
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/logs/debug-internal.log
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/rear', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz', 'image_view': 'rear', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.1, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-cfg-rear-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,rear', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:48,193 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:48,403 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:48,409 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:48,411 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:48,423 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:49,171 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:49,641 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:49,642 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:49,642 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:49,642 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:49,646 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:19,374 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/e6xhvf8a
+2026-08-16 20:03:19,375 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:19,375 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:19,375 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:20,545 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/run-e6xhvf8a.wandb b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/run-e6xhvf8a.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..4ff7522c50d2debdd225c605938d7f7289302d39
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/rear/wandb/wandb/run-20260816_191548-e6xhvf8a/run-e6xhvf8a.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9610f18b3eb276c71b34652f17bba468d23b0bf0361257978d0f214ae2f3e49
+size 236378
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/code_provenance.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/code_provenance.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d65c77fa6bb7025a1484f4ad82c6b647f7a2315
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/code_provenance.json
@@ -0,0 +1,7 @@
+{
+  "file": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "commit": "81d95989e1279e432c4145933bf9ce02c3f70f77",
+  "branch": "flow-field",
+  "train_cond_flow_sha256": "095053598f77923a8c8fbdd63bb77280bd641d5f8ca004bc100e63b82bae7616",
+  "seiler_cond_backbone_sha256": "908f1f3b3756ac054e640b4e6e9fe7a40e2130adcf7ba0ce09432f74783dd97c"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_best.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_best.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c97ce3f1659fbc82b8a29797f7f9b1a94a35c01a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_best.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da5cdec67eddfbdf848497b9a7a16083f74cb61dafb77dcac0928e0e5e68635f
+size 329404702
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_last.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_last.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4c29bd7ddec53024149eddf8cab45b0c026cad66
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_last.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88ad22f81535f67833592450c6e83a7ed6b914a861022ebdc1c9e5a4f00ba928
+size 988212490
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_step_005000.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_step_005000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6591c99fb4d2cae492e832754616f981bc15c1f9
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_step_005000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e77475c3523f26ed86f4bf177cd095169f90aba579e766a6879b1e07450eefb
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_step_010000.pt b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_step_010000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..79772e9c0b6291748f86c9357f62d9965830b90b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/model_step_010000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56c35ac8a41679234f47fb62a1ff9b9dc92a57eeab2af8da1ac66c2bcda96652
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/run_config.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/run_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..c53f5957b5c62b544de0bef357a7834698325e88
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/run_config.json
@@ -0,0 +1,46 @@
+{
+  "mode": "train",
+  "out_dir": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side",
+  "paired_latents": "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz",
+  "image_view": "side",
+  "seed": 20260526,
+  "val_fraction": 0.05,
+  "max_val": 128,
+  "val_overlap_train": true,
+  "shape_scale": 0.0,
+  "image_scale": 0.0,
+  "hidden_size": 512,
+  "depth": 12,
+  "num_heads": 8,
+  "head_dim": 64,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "pos_embed": "2d",
+  "attention_backend": "sdpa",
+  "gradient_checkpointing": false,
+  "cond_dropout": 0.1,
+  "batch_size": 128,
+  "num_workers": 2,
+  "steps": 10000,
+  "lr": 0.0001,
+  "weight_decay": 0.01,
+  "log_every": 100,
+  "eval_every": 500,
+  "eval_noise_seed": 20260712,
+  "save_every": 5000,
+  "resume": true,
+  "sample_count": 8,
+  "sample_steps": 50,
+  "sample_seed": 20260712,
+  "sample_checkpoint": "model_best.pt",
+  "guidance_scale": 1.0,
+  "decode_mesh": false,
+  "shape_resolution": 512,
+  "shape_max_batch": 32768,
+  "wandb": true,
+  "wandb_project": "drivaer-qwen3v-input-cond",
+  "wandb_entity": "kanydao-epfl",
+  "wandb_run_name": "drivaer3v-inputcond-cfg-side-0816c",
+  "wandb_tags": "drivaer,qwen3v,input-conditional,channel-concat,side",
+  "wandb_mode": "online"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/summary.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..93a0aed4646d1fb8f25e14c7ea1b0a2a754a47b5
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/summary.json
@@ -0,0 +1,10 @@
+{
+  "image_view": "side",
+  "best_val_shape": 0.8275899887084961,
+  "steps": 10000,
+  "train_samples": 3051,
+  "val_samples": 128,
+  "shape_scale": 1.7132293052897125,
+  "image_scale": 1.4969936040640752,
+  "conditioning": "sketch_latent_channel_concat"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/train_history.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/train_history.json
new file mode 100644
index 0000000000000000000000000000000000000000..2ca35ca389793c0e56ae0d36011fbe29d89cce5c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/train_history.json
@@ -0,0 +1,692 @@
+[
+  {
+    "step": 1,
+    "train_shape": 2.0095300674438477,
+    "lr": 9.999999765596898e-05,
+    "elapsed_sec": 1.0152995586395264
+  },
+  {
+    "step": 1,
+    "val_shape": 2.0061230659484863
+  },
+  {
+    "step": 100,
+    "train_shape": 1.230437159538269,
+    "lr": 9.997656161737219e-05,
+    "elapsed_sec": 29.062531232833862
+  },
+  {
+    "step": 200,
+    "train_shape": 0.9688507318496704,
+    "lr": 9.990626960034288e-05,
+    "elapsed_sec": 56.637324810028076
+  },
+  {
+    "step": 300,
+    "train_shape": 0.8945338726043701,
+    "lr": 9.978919331864633e-05,
+    "elapsed_sec": 84.2354826927185
+  },
+  {
+    "step": 400,
+    "train_shape": 0.903636634349823,
+    "lr": 9.962544831243784e-05,
+    "elapsed_sec": 111.84298896789551
+  },
+  {
+    "step": 500,
+    "train_shape": 0.8689815998077393,
+    "lr": 9.94151961782692e-05,
+    "elapsed_sec": 139.46273922920227
+  },
+  {
+    "step": 500,
+    "val_shape": 0.9321322441101074
+  },
+  {
+    "step": 600,
+    "train_shape": 0.928634524345398,
+    "lr": 9.915864440961283e-05,
+    "elapsed_sec": 168.01548767089844
+  },
+  {
+    "step": 700,
+    "train_shape": 0.8781734704971313,
+    "lr": 9.885604619209063e-05,
+    "elapsed_sec": 195.6612594127655
+  },
+  {
+    "step": 800,
+    "train_shape": 0.8472621440887451,
+    "lr": 9.85077001536101e-05,
+    "elapsed_sec": 223.31054401397705
+  },
+  {
+    "step": 900,
+    "train_shape": 0.8363358974456787,
+    "lr": 9.811395006965495e-05,
+    "elapsed_sec": 250.9572994709015
+  },
+  {
+    "step": 1000,
+    "train_shape": 0.8350677490234375,
+    "lr": 9.767518452401991e-05,
+    "elapsed_sec": 278.6107852458954
+  },
+  {
+    "step": 1000,
+    "val_shape": 0.8978338241577148
+  },
+  {
+    "step": 1100,
+    "train_shape": 0.882240891456604,
+    "lr": 9.71918365253258e-05,
+    "elapsed_sec": 307.1630344390869
+  },
+  {
+    "step": 1200,
+    "train_shape": 0.8673239350318909,
+    "lr": 9.666438307969207e-05,
+    "elapsed_sec": 334.8068337440491
+  },
+  {
+    "step": 1300,
+    "train_shape": 0.8272168636322021,
+    "lr": 9.60933447199893e-05,
+    "elapsed_sec": 362.4472620487213
+  },
+  {
+    "step": 1400,
+    "train_shape": 0.843940258026123,
+    "lr": 9.547928499213621e-05,
+    "elapsed_sec": 390.0825707912445
+  },
+  {
+    "step": 1500,
+    "train_shape": 0.8801677823066711,
+    "lr": 9.482280989894765e-05,
+    "elapsed_sec": 417.72292470932007
+  },
+  {
+    "step": 1500,
+    "val_shape": 0.8837422728538513
+  },
+  {
+    "step": 1600,
+    "train_shape": 0.8753843307495117,
+    "lr": 9.412456730208378e-05,
+    "elapsed_sec": 446.2963662147522
+  },
+  {
+    "step": 1700,
+    "train_shape": 0.809107780456543,
+    "lr": 9.338524628268747e-05,
+    "elapsed_sec": 473.93797492980957
+  },
+  {
+    "step": 1800,
+    "train_shape": 0.8599650859832764,
+    "lr": 9.260557646134585e-05,
+    "elapsed_sec": 501.575279712677
+  },
+  {
+    "step": 1900,
+    "train_shape": 0.8246607184410095,
+    "lr": 9.178632727804172e-05,
+    "elapsed_sec": 529.2274954319
+  },
+  {
+    "step": 2000,
+    "train_shape": 0.7819772958755493,
+    "lr": 9.092830723281015e-05,
+    "elapsed_sec": 556.8708121776581
+  },
+  {
+    "step": 2000,
+    "val_shape": 0.8747957348823547
+  },
+  {
+    "step": 2100,
+    "train_shape": 0.8825388550758362,
+    "lr": 9.003236308784539e-05,
+    "elapsed_sec": 585.4527173042297
+  },
+  {
+    "step": 2200,
+    "train_shape": 0.8644482493400574,
+    "lr": 8.90993790318501e-05,
+    "elapsed_sec": 613.112726688385
+  },
+  {
+    "step": 2300,
+    "train_shape": 0.824223518371582,
+    "lr": 8.813027580744691e-05,
+    "elapsed_sec": 640.7650475502014
+  },
+  {
+    "step": 2400,
+    "train_shape": 0.8007715940475464,
+    "lr": 8.712600980251722e-05,
+    "elapsed_sec": 668.5157489776611
+  },
+  {
+    "step": 2500,
+    "train_shape": 0.9252511858940125,
+    "lr": 8.60875721063612e-05,
+    "elapsed_sec": 696.1761031150818
+  },
+  {
+    "step": 2500,
+    "val_shape": 0.8689936399459839
+  },
+  {
+    "step": 2600,
+    "train_shape": 0.8613280057907104,
+    "lr": 8.501598753161292e-05,
+    "elapsed_sec": 724.8796186447144
+  },
+  {
+    "step": 2700,
+    "train_shape": 0.8130918741226196,
+    "lr": 8.391231360287367e-05,
+    "elapsed_sec": 752.5387854576111
+  },
+  {
+    "step": 2800,
+    "train_shape": 0.8184294104576111,
+    "lr": 8.277763951306297e-05,
+    "elapsed_sec": 780.1984882354736
+  },
+  {
+    "step": 2900,
+    "train_shape": 0.7601141929626465,
+    "lr": 8.161308504851662e-05,
+    "elapsed_sec": 808.0357782840729
+  },
+  {
+    "step": 3000,
+    "train_shape": 0.791191577911377,
+    "lr": 8.04197994838928e-05,
+    "elapsed_sec": 835.7075960636139
+  },
+  {
+    "step": 3000,
+    "val_shape": 0.8621280193328857
+  },
+  {
+    "step": 3100,
+    "train_shape": 0.8406544327735901,
+    "lr": 7.919896044797652e-05,
+    "elapsed_sec": 864.2567901611328
+  },
+  {
+    "step": 3200,
+    "train_shape": 0.7526840567588806,
+    "lr": 7.79517727615026e-05,
+    "elapsed_sec": 892.1400363445282
+  },
+  {
+    "step": 3300,
+    "train_shape": 0.8085770606994629,
+    "lr": 7.667946724814286e-05,
+    "elapsed_sec": 919.7933354377747
+  },
+  {
+    "step": 3400,
+    "train_shape": 0.8090295195579529,
+    "lr": 7.538329951983173e-05,
+    "elapsed_sec": 947.537921667099
+  },
+  {
+    "step": 3500,
+    "train_shape": 0.820824146270752,
+    "lr": 7.406454873762866e-05,
+    "elapsed_sec": 975.1980719566345
+  },
+  {
+    "step": 3500,
+    "val_shape": 0.8541576266288757
+  },
+  {
+    "step": 3600,
+    "train_shape": 0.7542186379432678,
+    "lr": 7.272451634934115e-05,
+    "elapsed_sec": 1003.7780251502991
+  },
+  {
+    "step": 3700,
+    "train_shape": 0.8020529747009277,
+    "lr": 7.136452480515233e-05,
+    "elapsed_sec": 1031.4249048233032
+  },
+  {
+    "step": 3800,
+    "train_shape": 0.8098368644714355,
+    "lr": 6.99859162525224e-05,
+    "elapsed_sec": 1059.0668776035309
+  },
+  {
+    "step": 3900,
+    "train_shape": 0.8001044988632202,
+    "lr": 6.859005121165148e-05,
+    "elapsed_sec": 1086.709549665451
+  },
+  {
+    "step": 4000,
+    "train_shape": 0.8102914094924927,
+    "lr": 6.71783072328101e-05,
+    "elapsed_sec": 1114.350176334381
+  },
+  {
+    "step": 4000,
+    "val_shape": 0.8493450284004211
+  },
+  {
+    "step": 4100,
+    "train_shape": 0.7861857414245605,
+    "lr": 6.575207753686356e-05,
+    "elapsed_sec": 1142.8732233047485
+  },
+  {
+    "step": 4200,
+    "train_shape": 0.7884088158607483,
+    "lr": 6.431276964033111e-05,
+    "elapsed_sec": 1170.5187573432922
+  },
+  {
+    "step": 4300,
+    "train_shape": 0.7744752168655396,
+    "lr": 6.286180396633663e-05,
+    "elapsed_sec": 1198.1503121852875
+  },
+  {
+    "step": 4400,
+    "train_shape": 0.7798285484313965,
+    "lr": 6.140061244282311e-05,
+    "elapsed_sec": 1225.7891006469727
+  },
+  {
+    "step": 4500,
+    "train_shape": 0.7777060270309448,
+    "lr": 5.9930637089412255e-05,
+    "elapsed_sec": 1253.429010629654
+  },
+  {
+    "step": 4500,
+    "val_shape": 0.8456041216850281
+  },
+  {
+    "step": 4600,
+    "train_shape": 0.811783492565155,
+    "lr": 5.845332859430572e-05,
+    "elapsed_sec": 1281.987181186676
+  },
+  {
+    "step": 4700,
+    "train_shape": 0.6973405480384827,
+    "lr": 5.697014488263067e-05,
+    "elapsed_sec": 1309.742538690567
+  },
+  {
+    "step": 4800,
+    "train_shape": 0.795626163482666,
+    "lr": 5.548254967764354e-05,
+    "elapsed_sec": 1337.39915060997
+  },
+  {
+    "step": 4900,
+    "train_shape": 0.8000576496124268,
+    "lr": 5.3992011056212214e-05,
+    "elapsed_sec": 1365.0616211891174
+  },
+  {
+    "step": 5000,
+    "train_shape": 0.7667695879936218,
+    "lr": 5.250000000000107e-05,
+    "elapsed_sec": 1392.7173030376434
+  },
+  {
+    "step": 5000,
+    "val_shape": 0.8429937362670898
+  },
+  {
+    "step": 5100,
+    "train_shape": 0.796161413192749,
+    "lr": 5.100798894378995e-05,
+    "elapsed_sec": 1423.3116273880005
+  },
+  {
+    "step": 5200,
+    "train_shape": 0.7902384996414185,
+    "lr": 4.951745032235864e-05,
+    "elapsed_sec": 1450.9636805057526
+  },
+  {
+    "step": 5300,
+    "train_shape": 0.7437206506729126,
+    "lr": 4.8029855117371535e-05,
+    "elapsed_sec": 1478.614720582962
+  },
+  {
+    "step": 5400,
+    "train_shape": 0.8234460949897766,
+    "lr": 4.654667140569645e-05,
+    "elapsed_sec": 1506.2726650238037
+  },
+  {
+    "step": 5500,
+    "train_shape": 0.7636897563934326,
+    "lr": 4.506936291058995e-05,
+    "elapsed_sec": 1533.9219110012054
+  },
+  {
+    "step": 5500,
+    "val_shape": 0.8392347693443298
+  },
+  {
+    "step": 5600,
+    "train_shape": 0.8129669427871704,
+    "lr": 4.359938755717899e-05,
+    "elapsed_sec": 1562.6412620544434
+  },
+  {
+    "step": 5700,
+    "train_shape": 0.7670145034790039,
+    "lr": 4.213819603366514e-05,
+    "elapsed_sec": 1590.2936761379242
+  },
+  {
+    "step": 5800,
+    "train_shape": 0.8358398079872131,
+    "lr": 4.0687230359670296e-05,
+    "elapsed_sec": 1617.9485573768616
+  },
+  {
+    "step": 5900,
+    "train_shape": 0.8181709051132202,
+    "lr": 3.924792246313749e-05,
+    "elapsed_sec": 1645.7995762825012
+  },
+  {
+    "step": 6000,
+    "train_shape": 0.7403063774108887,
+    "lr": 3.7821692767190845e-05,
+    "elapsed_sec": 1673.4534966945648
+  },
+  {
+    "step": 6000,
+    "val_shape": 0.8371540904045105
+  },
+  {
+    "step": 6100,
+    "train_shape": 0.7463539838790894,
+    "lr": 3.640994878834947e-05,
+    "elapsed_sec": 1702.175609588623
+  },
+  {
+    "step": 6200,
+    "train_shape": 0.7964744567871094,
+    "lr": 3.501408374747862e-05,
+    "elapsed_sec": 1729.8023500442505
+  },
+  {
+    "step": 6300,
+    "train_shape": 0.755229115486145,
+    "lr": 3.363547519484887e-05,
+    "elapsed_sec": 1757.4466133117676
+  },
+  {
+    "step": 6400,
+    "train_shape": 0.7779753804206848,
+    "lr": 3.227548365066013e-05,
+    "elapsed_sec": 1785.0818228721619
+  },
+  {
+    "step": 6500,
+    "train_shape": 0.7367839813232422,
+    "lr": 3.0935451262372706e-05,
+    "elapsed_sec": 1812.71892619133
+  },
+  {
+    "step": 6500,
+    "val_shape": 0.8348178863525391
+  },
+  {
+    "step": 6600,
+    "train_shape": 0.7599100470542908,
+    "lr": 2.9616700480169673e-05,
+    "elapsed_sec": 1841.2494912147522
+  },
+  {
+    "step": 6700,
+    "train_shape": 0.7929325103759766,
+    "lr": 2.8320532751858457e-05,
+    "elapsed_sec": 1868.882966518402
+  },
+  {
+    "step": 6800,
+    "train_shape": 0.7733491659164429,
+    "lr": 2.7048227238498715e-05,
+    "elapsed_sec": 1896.51606631279
+  },
+  {
+    "step": 6900,
+    "train_shape": 0.778327465057373,
+    "lr": 2.5801039552024805e-05,
+    "elapsed_sec": 1924.1498939990997
+  },
+  {
+    "step": 7000,
+    "train_shape": 0.8101559281349182,
+    "lr": 2.458020051610849e-05,
+    "elapsed_sec": 1951.8770596981049
+  },
+  {
+    "step": 7000,
+    "val_shape": 0.8326500058174133
+  },
+  {
+    "step": 7100,
+    "train_shape": 0.7814571261405945,
+    "lr": 2.3386914951484544e-05,
+    "elapsed_sec": 1980.4736804962158
+  },
+  {
+    "step": 7200,
+    "train_shape": 0.7910606265068054,
+    "lr": 2.222236048693809e-05,
+    "elapsed_sec": 2008.107592344284
+  },
+  {
+    "step": 7300,
+    "train_shape": 0.7953336834907532,
+    "lr": 2.1087686397127328e-05,
+    "elapsed_sec": 2035.7451848983765
+  },
+  {
+    "step": 7400,
+    "train_shape": 0.8225245475769043,
+    "lr": 1.998401246838802e-05,
+    "elapsed_sec": 2063.3748881816864
+  },
+  {
+    "step": 7500,
+    "train_shape": 0.7841835021972656,
+    "lr": 1.891242789363976e-05,
+    "elapsed_sec": 2091.014626979828
+  },
+  {
+    "step": 7500,
+    "val_shape": 0.8315709829330444
+  },
+  {
+    "step": 7600,
+    "train_shape": 0.7672529816627502,
+    "lr": 1.787399019748373e-05,
+    "elapsed_sec": 2119.618835926056
+  },
+  {
+    "step": 7700,
+    "train_shape": 0.71863853931427,
+    "lr": 1.6869724192553968e-05,
+    "elapsed_sec": 2147.2673654556274
+  },
+  {
+    "step": 7800,
+    "train_shape": 0.780920684337616,
+    "lr": 1.5900620968150808e-05,
+    "elapsed_sec": 2174.9229798316956
+  },
+  {
+    "step": 7900,
+    "train_shape": 0.8127443790435791,
+    "lr": 1.496763691215551e-05,
+    "elapsed_sec": 2202.5790224075317
+  },
+  {
+    "step": 8000,
+    "train_shape": 0.8237189054489136,
+    "lr": 1.4071692767190806e-05,
+    "elapsed_sec": 2230.23694562912
+  },
+  {
+    "step": 8000,
+    "val_shape": 0.829904317855835
+  },
+  {
+    "step": 8100,
+    "train_shape": 0.7979024648666382,
+    "lr": 1.3213672721959122e-05,
+    "elapsed_sec": 2258.8097693920135
+  },
+  {
+    "step": 8200,
+    "train_shape": 0.7921815514564514,
+    "lr": 1.2394423538655062e-05,
+    "elapsed_sec": 2286.460554122925
+  },
+  {
+    "step": 8300,
+    "train_shape": 0.8077646493911743,
+    "lr": 1.1614753717313381e-05,
+    "elapsed_sec": 2314.1202495098114
+  },
+  {
+    "step": 8400,
+    "train_shape": 0.7919397354125977,
+    "lr": 1.0875432697917097e-05,
+    "elapsed_sec": 2341.77104640007
+  },
+  {
+    "step": 8500,
+    "train_shape": 0.7434149384498596,
+    "lr": 1.0177190101053074e-05,
+    "elapsed_sec": 2369.4247636795044
+  },
+  {
+    "step": 8500,
+    "val_shape": 0.8288809061050415
+  },
+  {
+    "step": 8600,
+    "train_shape": 0.753404438495636,
+    "lr": 9.520715007864551e-06,
+    "elapsed_sec": 2398.0150430202484
+  },
+  {
+    "step": 8700,
+    "train_shape": 0.7433322668075562,
+    "lr": 8.906655280011289e-06,
+    "elapsed_sec": 2425.665712594986
+  },
+  {
+    "step": 8800,
+    "train_shape": 0.780956506729126,
+    "lr": 8.335616920308386e-06,
+    "elapsed_sec": 2453.32111120224
+  },
+  {
+    "step": 8900,
+    "train_shape": 0.7386068105697632,
+    "lr": 7.808163474674577e-06,
+    "elapsed_sec": 2480.969326734543
+  },
+  {
+    "step": 9000,
+    "train_shape": 0.7933535575866699,
+    "lr": 7.32481547598044e-06,
+    "elapsed_sec": 2508.6227786540985
+  },
+  {
+    "step": 9000,
+    "val_shape": 0.8290653824806213
+  },
+  {
+    "step": 9100,
+    "train_shape": 0.7377415895462036,
+    "lr": 6.886049930345396e-06,
+    "elapsed_sec": 2536.462882757187
+  },
+  {
+    "step": 9200,
+    "train_shape": 0.7918436527252197,
+    "lr": 6.492299846390179e-06,
+    "elapsed_sec": 2564.1170580387115
+  },
+  {
+    "step": 9300,
+    "train_shape": 0.7681806087493896,
+    "lr": 6.143953807909618e-06,
+    "elapsed_sec": 2591.865647792816
+  },
+  {
+    "step": 9400,
+    "train_shape": 0.7745444178581238,
+    "lr": 5.841355590387376e-06,
+    "elapsed_sec": 2619.5175914764404
+  },
+  {
+    "step": 9500,
+    "train_shape": 0.7733450531959534,
+    "lr": 5.5848038217310204e-06,
+    "elapsed_sec": 2647.384707927704
+  },
+  {
+    "step": 9500,
+    "val_shape": 0.8277757167816162
+  },
+  {
+    "step": 9600,
+    "train_shape": 0.8388252258300781,
+    "lr": 5.374551687562344e-06,
+    "elapsed_sec": 2675.9787826538086
+  },
+  {
+    "step": 9700,
+    "train_shape": 0.7549374103546143,
+    "lr": 5.2108066813537186e-06,
+    "elapsed_sec": 2703.7809925079346
+  },
+  {
+    "step": 9800,
+    "train_shape": 0.7799059152603149,
+    "lr": 5.093730399657109e-06,
+    "elapsed_sec": 2731.408934354782
+  },
+  {
+    "step": 9900,
+    "train_shape": 0.7245686650276184,
+    "lr": 5.023438382627753e-06,
+    "elapsed_sec": 2759.0419824123383
+  },
+  {
+    "step": 10000,
+    "train_shape": 0.7847051024436951,
+    "lr": 5e-06,
+    "elapsed_sec": 2786.8980469703674
+  },
+  {
+    "step": 10000,
+    "val_shape": 0.8275899887084961
+  }
+]
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..965f111e9b1bbf9bb621601c72a002dde641980e
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:48.977817566Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:49.293408985Z","level":"INFO","msg":"stream: created new stream","id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.29350293Z","level":"INFO","msg":"handler: started","stream_id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.296351578Z","level":"INFO","msg":"stream: started","id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.296386794Z","level":"INFO","msg":"writer: started","stream_id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.296391062Z","level":"INFO","msg":"sender: started","stream_id":"42v9byh1"}
+{"time":"2026-08-16T20:02:22.815923116Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:02:23.068725923Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:02:23.075143059Z","level":"INFO","msg":"stream: closing","id":"42v9byh1"}
+{"time":"2026-08-16T20:02:23.07515987Z","level":"INFO","msg":"handler: closed","stream_id":"42v9byh1"}
+{"time":"2026-08-16T20:02:23.076217748Z","level":"INFO","msg":"sender: closed","stream_id":"42v9byh1"}
+{"time":"2026-08-16T20:02:23.076241342Z","level":"INFO","msg":"stream: closed","id":"42v9byh1"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/debug.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..db5896eedb5ee02c98e4f5e623debfa54e0bfdfb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug.log
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug-internal.log
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:48,733 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz', 'image_view': 'side', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.1, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-cfg-side-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,side', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:48,733 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:48,959 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:48,970 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:48,972 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:48,994 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:49,617 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:50,207 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:02:21,876 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/42v9byh1
+2026-08-16 20:02:21,877 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:02:21,878 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:02:21,878 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:02:23,073 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/config.yaml b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..94810f077cbe3e12df1dcde991cba1fd51b54dea
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/config.yaml
@@ -0,0 +1,172 @@
+_wandb:
+    value:
+        cli_version: 0.22.3
+        e:
+            ft0m6ru8kwsntudviapr5e07h1ozshp2:
+                args:
+                    - --mode
+                    - train
+                    - --image-view
+                    - side
+                    - --paired-latents
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz
+                    - --out-dir
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side
+                    - --steps
+                    - "10000"
+                    - --batch-size
+                    - "128"
+                    - --lr
+                    - "1e-4"
+                    - --weight-decay
+                    - "0.01"
+                    - --seed
+                    - "20260526"
+                    - --pos-embed
+                    - 2d
+                    - --cond-dropout
+                    - "0.1"
+                    - --resume
+                    - --wandb
+                    - --wandb-run-name
+                    - drivaer3v-inputcond-cfg-side-0816c
+                    - --wandb-tags
+                    - drivaer,qwen3v,input-conditional,channel-concat,side
+                codePath: code/ICLR/customized-design/input_cond/train_cond_flow.py
+                codePathLocal: train_cond_flow.py
+                cpu_count: 32
+                cpu_count_logical: 64
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "7679361679360"
+                        used: "3024004001792"
+                email: 1193414549@qq.com
+                executable: /mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python
+                git:
+                    commit: 81d95989e1279e432c4145933bf9ce02c3f70f77
+                gpu: NVIDIA A100-SXM4-40GB
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "42949672960"
+                      name: NVIDIA A100-SXM4-40GB
+                      uuid: GPU-163dd541-2599-630b-f41a-fb118ab4205d
+                host: inputcond-cfg-side-a100-40g-0816c-0-0
+                memory:
+                    total: "540742455296"
+                os: Linux-6.8.0-124-generic-x86_64-with-glibc2.35
+                program: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py
+                python: CPython 3.10.18
+                root: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side/wandb
+                startedAt: "2026-08-16T19:15:48.715546Z"
+                writerId: ft0m6ru8kwsntudviapr5e07h1ozshp2
+        m: []
+        python_version: 3.10.18
+        t:
+            "1":
+                - 1
+                - 105
+            "2":
+                - 1
+                - 105
+            "3":
+                - 2
+                - 13
+                - 15
+                - 16
+                - 61
+                - 62
+            "4": 3.10.18
+            "5": 0.22.3
+            "12": 0.22.3
+            "13": linux-x86_64
+attention_backend:
+    value: sdpa
+batch_size:
+    value: 128
+cond_dropout:
+    value: 0.1
+decode_mesh:
+    value: false
+depth:
+    value: 12
+dropout:
+    value: 0
+eval_every:
+    value: 500
+eval_noise_seed:
+    value: 20260712
+gradient_checkpointing:
+    value: false
+guidance_scale:
+    value: 1
+head_dim:
+    value: 64
+hidden_size:
+    value: 512
+image_scale:
+    value: 0
+image_view:
+    value: side
+log_every:
+    value: 100
+lr:
+    value: 0.0001
+max_val:
+    value: 128
+mlp_ratio:
+    value: 4
+mode:
+    value: train
+num_heads:
+    value: 8
+num_workers:
+    value: 2
+out_dir:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side
+paired_latents:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz
+pos_embed:
+    value: 2d
+resume:
+    value: true
+sample_checkpoint:
+    value: model_best.pt
+sample_count:
+    value: 8
+sample_seed:
+    value: 20260712
+sample_steps:
+    value: 50
+save_every:
+    value: 5000
+seed:
+    value: 20260526
+shape_max_batch:
+    value: 32768
+shape_resolution:
+    value: 512
+shape_scale:
+    value: 0
+steps:
+    value: 10000
+val_fraction:
+    value: 0.05
+val_overlap_train:
+    value: true
+wandb:
+    value: true
+wandb_entity:
+    value: kanydao-epfl
+wandb_mode:
+    value: online
+wandb_project:
+    value: drivaer-qwen3v-input-cond
+wandb_run_name:
+    value: drivaer3v-inputcond-cfg-side-0816c
+wandb_tags:
+    value: drivaer,qwen3v,input-conditional,channel-concat,side
+weight_decay:
+    value: 0.01
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/output.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..e7916664db69d730b370473b7d8c278b33940e41
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/output.log
@@ -0,0 +1,123 @@
+[wandb] project=drivaer-qwen3v-input-cond run=drivaer3v-inputcond-cfg-side-0816c mode=online
+[train] step=000001 loss=2.009530 lr=1.00e-04
+[val] step=000001 shape=2.006123
+[train] step=000100 loss=1.230437 lr=1.00e-04
+[train] step=000200 loss=0.968851 lr=9.99e-05
+[train] step=000300 loss=0.894534 lr=9.98e-05
+[train] step=000400 loss=0.903637 lr=9.96e-05
+[train] step=000500 loss=0.868982 lr=9.94e-05
+[val] step=000500 shape=0.932132
+[train] step=000600 loss=0.928635 lr=9.92e-05
+[train] step=000700 loss=0.878173 lr=9.89e-05
+[train] step=000800 loss=0.847262 lr=9.85e-05
+[train] step=000900 loss=0.836336 lr=9.81e-05
+[train] step=001000 loss=0.835068 lr=9.77e-05
+[val] step=001000 shape=0.897834
+[train] step=001100 loss=0.882241 lr=9.72e-05
+[train] step=001200 loss=0.867324 lr=9.67e-05
+[train] step=001300 loss=0.827217 lr=9.61e-05
+[train] step=001400 loss=0.843940 lr=9.55e-05
+[train] step=001500 loss=0.880168 lr=9.48e-05
+[val] step=001500 shape=0.883742
+[train] step=001600 loss=0.875384 lr=9.41e-05
+[train] step=001700 loss=0.809108 lr=9.34e-05
+[train] step=001800 loss=0.859965 lr=9.26e-05
+[train] step=001900 loss=0.824661 lr=9.18e-05
+[train] step=002000 loss=0.781977 lr=9.09e-05
+[val] step=002000 shape=0.874796
+[train] step=002100 loss=0.882539 lr=9.00e-05
+[train] step=002200 loss=0.864448 lr=8.91e-05
+[train] step=002300 loss=0.824224 lr=8.81e-05
+[train] step=002400 loss=0.800772 lr=8.71e-05
+[train] step=002500 loss=0.925251 lr=8.61e-05
+[val] step=002500 shape=0.868994
+[train] step=002600 loss=0.861328 lr=8.50e-05
+[train] step=002700 loss=0.813092 lr=8.39e-05
+[train] step=002800 loss=0.818429 lr=8.28e-05
+[train] step=002900 loss=0.760114 lr=8.16e-05
+[train] step=003000 loss=0.791192 lr=8.04e-05
+[val] step=003000 shape=0.862128
+[train] step=003100 loss=0.840654 lr=7.92e-05
+[train] step=003200 loss=0.752684 lr=7.80e-05
+[train] step=003300 loss=0.808577 lr=7.67e-05
+[train] step=003400 loss=0.809030 lr=7.54e-05
+[train] step=003500 loss=0.820824 lr=7.41e-05
+[val] step=003500 shape=0.854158
+[train] step=003600 loss=0.754219 lr=7.27e-05
+[train] step=003700 loss=0.802053 lr=7.14e-05
+[train] step=003800 loss=0.809837 lr=7.00e-05
+[train] step=003900 loss=0.800104 lr=6.86e-05
+[train] step=004000 loss=0.810291 lr=6.72e-05
+[val] step=004000 shape=0.849345
+[train] step=004100 loss=0.786186 lr=6.58e-05
+[train] step=004200 loss=0.788409 lr=6.43e-05
+[train] step=004300 loss=0.774475 lr=6.29e-05
+[train] step=004400 loss=0.779829 lr=6.14e-05
+[train] step=004500 loss=0.777706 lr=5.99e-05
+[val] step=004500 shape=0.845604
+[train] step=004600 loss=0.811783 lr=5.85e-05
+[train] step=004700 loss=0.697341 lr=5.70e-05
+[train] step=004800 loss=0.795626 lr=5.55e-05
+[train] step=004900 loss=0.800058 lr=5.40e-05
+[train] step=005000 loss=0.766770 lr=5.25e-05
+[val] step=005000 shape=0.842994
+[train] step=005100 loss=0.796161 lr=5.10e-05
+[train] step=005200 loss=0.790238 lr=4.95e-05
+[train] step=005300 loss=0.743721 lr=4.80e-05
+[train] step=005400 loss=0.823446 lr=4.65e-05
+[train] step=005500 loss=0.763690 lr=4.51e-05
+[val] step=005500 shape=0.839235
+[train] step=005600 loss=0.812967 lr=4.36e-05
+[train] step=005700 loss=0.767015 lr=4.21e-05
+[train] step=005800 loss=0.835840 lr=4.07e-05
+[train] step=005900 loss=0.818171 lr=3.92e-05
+[train] step=006000 loss=0.740306 lr=3.78e-05
+[val] step=006000 shape=0.837154
+[train] step=006100 loss=0.746354 lr=3.64e-05
+[train] step=006200 loss=0.796474 lr=3.50e-05
+[train] step=006300 loss=0.755229 lr=3.36e-05
+[train] step=006400 loss=0.777975 lr=3.23e-05
+[train] step=006500 loss=0.736784 lr=3.09e-05
+[val] step=006500 shape=0.834818
+[train] step=006600 loss=0.759910 lr=2.96e-05
+[train] step=006700 loss=0.792933 lr=2.83e-05
+[train] step=006800 loss=0.773349 lr=2.70e-05
+[train] step=006900 loss=0.778327 lr=2.58e-05
+[train] step=007000 loss=0.810156 lr=2.46e-05
+[val] step=007000 shape=0.832650
+[train] step=007100 loss=0.781457 lr=2.34e-05
+[train] step=007200 loss=0.791061 lr=2.22e-05
+[train] step=007300 loss=0.795334 lr=2.11e-05
+[train] step=007400 loss=0.822525 lr=2.00e-05
+[train] step=007500 loss=0.784184 lr=1.89e-05
+[val] step=007500 shape=0.831571
+[train] step=007600 loss=0.767253 lr=1.79e-05
+[train] step=007700 loss=0.718639 lr=1.69e-05
+[train] step=007800 loss=0.780921 lr=1.59e-05
+[train] step=007900 loss=0.812744 lr=1.50e-05
+[train] step=008000 loss=0.823719 lr=1.41e-05
+[val] step=008000 shape=0.829904
+[train] step=008100 loss=0.797902 lr=1.32e-05
+[train] step=008200 loss=0.792182 lr=1.24e-05
+[train] step=008300 loss=0.807765 lr=1.16e-05
+[train] step=008400 loss=0.791940 lr=1.09e-05
+[train] step=008500 loss=0.743415 lr=1.02e-05
+[val] step=008500 shape=0.828881
+[train] step=008600 loss=0.753404 lr=9.52e-06
+[train] step=008700 loss=0.743332 lr=8.91e-06
+[train] step=008800 loss=0.780957 lr=8.34e-06
+[train] step=008900 loss=0.738607 lr=7.81e-06
+[train] step=009000 loss=0.793354 lr=7.32e-06
+[val] step=009000 shape=0.829065
+[train] step=009100 loss=0.737742 lr=6.89e-06
+[train] step=009200 loss=0.791844 lr=6.49e-06
+[train] step=009300 loss=0.768181 lr=6.14e-06
+[train] step=009400 loss=0.774544 lr=5.84e-06
+[train] step=009500 loss=0.773345 lr=5.58e-06
+[val] step=009500 shape=0.827776
+[train] step=009600 loss=0.838825 lr=5.37e-06
+[train] step=009700 loss=0.754937 lr=5.21e-06
+[train] step=009800 loss=0.779906 lr=5.09e-06
+[train] step=009900 loss=0.724569 lr=5.02e-06
+[train] step=010000 loss=0.784705 lr=5.00e-06
+[val] step=010000 shape=0.827590
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/requirements.txt b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..2e10c9ab6bc99041706780299f9e3e5035f34676
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/requirements.txt
@@ -0,0 +1,302 @@
+shellingham==1.5.4
+importlib_metadata==8.7.0
+websocket-client==1.8.0
+rfc3986-validator==0.1.1
+aiohappyeyeballs==2.6.1
+tensorboardX==2.6.2.2
+matplotlib==3.8.4
+zstandard==0.23.0
+gradio_client==1.4.2
+nbclient==0.10.2
+pyaml==25.7.0
+argparse==1.4.0
+h11==0.16.0
+scooby==0.10.1
+pexpect==4.9.0
+jedi==0.19.2
+scikit-optimize==0.10.2
+omegaconf==2.3.0
+tomlkit==0.12.0
+yarl==1.20.1
+point-cloud-utils==0.34.0
+cffi==1.17.1
+asttokens==3.0.0
+webencodings==0.5.1
+imageio==2.34.1
+attrs==25.3.0
+hydra-core==1.3.2
+datasets==2.19.0
+pyquaternion==0.9.9
+trame-common==1.0.0
+wslink==2.3.4
+Pygments==2.19.2
+pydantic==2.11.7
+regex==2024.11.6
+timm==0.9.16
+ipykernel==6.29.5
+prompt_toolkit==3.0.51
+torchdiffeq==0.2.5
+pybind11==2.13.6
+debugpy==1.8.14
+pytz==2025.2
+transformers==4.40.1
+jupyterlab_pygments==0.3.0
+dash==3.1.1
+psutil==5.9.8
+tqdm==4.66.2
+multidict==6.6.3
+setproctitle==1.3.6
+nbformat==5.10.4
+prometheus_client==0.22.1
+jupyter_core==5.8.1
+gradio==5.5.0
+multiprocess==0.70.16
+fastapi==0.115.14
+simpervisor==1.0.0
+comm==0.2.2
+widgetsnbextension==4.0.14
+webcolors==24.11.1
+sentry-sdk==2.32.0
+tornado==6.5.1
+defusedxml==0.7.1
+Markdown==3.8.2
+numba==0.61.2
+pyarrow-hotfix==0.7
+trame-server==3.5.0
+nbconvert==7.16.6
+websockets==12.0
+semantic-version==2.10.0
+bpy==4.0.0
+smmap==5.0.2
+grpcio==1.73.1
+referencing==0.36.2
+annotated-types==0.7.0
+stack-data==0.6.3
+vtk==9.4.2
+safehttpx==0.1.6
+GitPython==3.1.44
+ipywidgets==8.1.7
+platformdirs==4.3.8
+tzdata==2025.2
+typer==0.16.0
+click==8.2.1
+networkx==3.3
+fsspec==2024.3.1
+plotly==6.2.0
+isoduration==20.11.0
+joblib==1.4.0
+aiosignal==1.4.0
+jsonschema-specifications==2025.4.1
+six==1.17.0
+pyzmq==27.0.0
+lazy_loader==0.4
+typeguard==2.13.3
+markdown-it-py==3.0.0
+httpx==0.28.1
+fqdn==1.5.1
+threadpoolctl==3.6.0
+orjson==3.10.18
+docker-pycreds==0.4.0
+jsonschema==4.24.0
+tensorboard-data-server==0.7.2
+nvidia-cufft-cu12==11.0.2.54
+aiofiles==23.2.1
+rpds-py==0.26.0
+Cython==3.1.2
+python-multipart==0.0.12
+nvidia-cuda-runtime-cu12==12.1.105
+huggingface-hub==0.26.2
+frozenlist==1.7.0
+onnxruntime==1.22.0
+cycler==0.12.1
+tifffile==2025.5.10
+safetensors==0.4.3
+nest-asyncio==1.6.0
+parso==0.8.4
+PyMCubes==0.1.4
+nvidia-nvtx-cu12==12.1.105
+pydub==0.25.1
+packaging==25.0
+ipython==8.37.0
+uvicorn==0.35.0
+pyarrow==20.0.0
+rich==14.0.0
+executing==2.2.0
+jupyter_server_terminals==0.5.3
+typing_extensions==4.12.2
+typing_extensions==4.15.0
+tensorboard==2.16.2
+argon2-cffi-bindings==21.2.0
+wcwidth==0.2.13
+typing-inspection==0.4.1
+pure_eval==0.2.3
+trimesh==4.3.2
+nvidia-cudnn-cu12==9.1.0.70
+humanfriendly==10.0
+Werkzeug==3.1.3
+decorator==5.2.1
+nvidia-cuda-nvrtc-cu12==12.1.105
+aiohttp==3.12.13
+uri-template==1.3.0
+anyio==4.9.0
+fastjsonschema==2.21.1
+nvidia-curand-cu12==10.3.2.106
+zipp==3.23.0
+rtree==1.4.1
+kiwisolver==1.4.8
+Flask==3.1.1
+nvidia-cuda-cupti-cu12==12.1.105
+pysdf==0.1.9
+sniffio==1.3.1
+itsdangerous==2.2.0
+rfc3339-validator==0.1.4
+trame==3.10.2
+open3d==0.18.0
+scikit-learn==1.7.0
+mistune==3.1.3
+diffusers==0.31.0
+absl-py==2.3.1
+cubvh==0.1.0
+trame-client==3.9.1
+einops==0.8.0
+plyfile==1.0.3
+bleach==6.2.0
+ptyprocess==0.7.0
+pycparser==2.22
+llvmlite==0.44.0
+addict==2.4.0
+diso==0.1.4
+Send2Trash==1.8.3
+more-itertools==10.7.0
+scikit-image==0.23.2
+propcache==0.3.2
+pandas==2.2.2
+flatbuffers==25.2.10
+gitdb==4.0.12
+jaxtyping==0.2.28
+soupsieve==2.7
+mdurl==0.1.2
+protobuf==5.29.5
+fonttools==4.58.5
+coloredlogs==15.0.1
+spaces==0.28.3
+jupyter_server==2.16.0
+trame-vtk==2.9.0
+terminado==0.18.1
+overrides==7.7.0
+pandocfilters==1.5.1
+nvidia-cusolver-cu12==11.4.5.107
+trame-vuetify==3.0.1
+async-timeout==5.0.1
+blinker==1.9.0
+exceptiongroup==1.3.0
+xxhash==3.5.0
+python-json-logger==3.3.0
+pyvista==0.45.2
+python-dateutil==2.9.0.post0
+ConfigArgParse==1.7.1
+PyMatting==1.1.14
+fpsample==0.3.3
+traitlets==5.14.3
+msgpack==1.1.1
+argon2-cffi==25.1.0
+tinycss2==1.4.0
+retrying==1.4.0
+jsonpointer==3.0.0
+matplotlib-inline==0.1.7
+jupyter_client==8.6.3
+contourpy==1.3.2
+lightning-utilities==0.11.2
+pydantic_core==2.33.2
+httpcore==1.0.9
+narwhals==1.46.0
+nvidia-cublas-cu12==12.1.3.1
+jupyterlab_widgets==3.0.15
+antlr4-python3-runtime==4.9.3
+pyparsing==3.1.2
+rembg==2.0.59
+ffmpy==0.6.0
+tokenizers==0.19.1
+beautifulsoup4==4.13.4
+pooch==1.8.2
+ninja==1.11.1.4
+jupyter_server_proxy==4.4.0
+ruff==0.12.2
+types-python-dateutil==2.9.0.20250708
+dill==0.3.8
+pytorch-lightning==2.2.4
+jupyter-events==0.12.0
+starlette==0.46.2
+arrow==1.3.0
+setuptools==78.1.1
+wheel==0.45.1
+pip==25.1
+mkl-service==2.5.2
+mpi4py==4.0.3
+h5py==3.14.0
+mkl_fft==2.1.1
+mkl_random==1.3.0
+mcubes==0.1.6
+wandb==0.22.3
+certifi==2026.1.4
+mpmath==1.3.0
+triton==3.1.0
+sympy==1.13.1
+nvidia-nccl-cu12==2.21.5
+nvidia-nvjitlink-cu12==12.9.86
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nvtx-cu11==11.8.86
+nvidia-nccl-cu11==2.21.5
+nvidia-cusparse-cu11==11.7.5.86
+nvidia-curand-cu11==10.3.0.86
+nvidia-cufft-cu11==10.9.0.58
+nvidia-cuda-runtime-cu11==11.8.89
+nvidia-cuda-nvrtc-cu11==11.8.89
+nvidia-cuda-cupti-cu11==11.8.87
+nvidia-cublas-cu11==11.11.3.6
+nvidia-cusolver-cu11==11.4.1.48
+nvidia-cudnn-cu11==9.1.0.70
+torch==2.5.1+cu118
+torchvision==0.20.1+cu118
+torchaudio==2.5.1+cu118
+PyYAML==6.0.3
+pillow==12.1.0
+torchmetrics==1.8.2
+accelerate==1.12.0
+urllib3==2.6.3
+idna==3.11
+charset-normalizer==3.4.4
+requests==2.32.5
+filelock==3.20.3
+torch_cluster==1.6.3+pt25cu118
+numpy==1.26.4
+opencv-python-headless==4.9.0.80
+PyOpenGL-accelerate==3.1.10
+PyOpenGL==3.1.0
+pyglet==2.1.14
+freetype-py==2.5.1
+pyrender==0.1.45
+scipy==1.15.3
+memory-profiler==0.61.0
+pyamg==5.3.0
+meshio==5.3.5
+torch-fem==0.6.2
+torch-geometric==2.7.0
+MarkupSafe==3.0.3
+Jinja2==3.1.6
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/wandb-metadata.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..65a30d8d686acbeff52ba2879b1439c15351c9cd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/wandb-metadata.json
@@ -0,0 +1,69 @@
+{
+  "os":  "Linux-6.8.0-124-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.10.18",
+  "startedAt":  "2026-08-16T19:15:48.715546Z",
+  "args":  [
+    "--mode",
+    "train",
+    "--image-view",
+    "side",
+    "--paired-latents",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz",
+    "--out-dir",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side",
+    "--steps",
+    "10000",
+    "--batch-size",
+    "128",
+    "--lr",
+    "1e-4",
+    "--weight-decay",
+    "0.01",
+    "--seed",
+    "20260526",
+    "--pos-embed",
+    "2d",
+    "--cond-dropout",
+    "0.1",
+    "--resume",
+    "--wandb",
+    "--wandb-run-name",
+    "drivaer3v-inputcond-cfg-side-0816c",
+    "--wandb-tags",
+    "drivaer,qwen3v,input-conditional,channel-concat,side"
+  ],
+  "program":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePath":  "code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePathLocal":  "train_cond_flow.py",
+  "git":  {
+    "commit":  "81d95989e1279e432c4145933bf9ce02c3f70f77"
+  },
+  "email":  "1193414549@qq.com",
+  "root":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side/wandb",
+  "host":  "inputcond-cfg-side-a100-40g-0816c-0-0",
+  "executable":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python",
+  "cpu_count":  32,
+  "cpu_count_logical":  64,
+  "gpu":  "NVIDIA A100-SXM4-40GB",
+  "gpu_count":  1,
+  "disk":  {
+    "/":  {
+      "total":  "7679361679360",
+      "used":  "3024004001792"
+    }
+  },
+  "memory":  {
+    "total":  "540742455296"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A100-SXM4-40GB",
+      "memoryTotal":  "42949672960",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-163dd541-2599-630b-f41a-fb118ab4205d"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "ft0m6ru8kwsntudviapr5e07h1ozshp2"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/wandb-summary.json b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..19eb4e97b32776a10f964fc9eb6ea5a30e9831e5
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/files/wandb-summary.json
@@ -0,0 +1 @@
+{"steps":10000,"conditioning":"sketch_latent_channel_concat","train/shape":0.7847051024436951,"val_samples":128,"_step":10000,"_wandb":{"runtime":2792},"train/lr":5e-06,"best_val_shape":0.8275899887084961,"_timestamp":1.786910537286624e+09,"image_view":"side","val/shape":0.8275899887084961,"image_scale":1.4969936040640752,"shape_scale":1.7132293052897125,"train_samples":3051,"_runtime":2792.260386112,"time/elapsed_sec":2786.8980469703674}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..965f111e9b1bbf9bb621601c72a002dde641980e
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:48.977817566Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:49.293408985Z","level":"INFO","msg":"stream: created new stream","id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.29350293Z","level":"INFO","msg":"handler: started","stream_id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.296351578Z","level":"INFO","msg":"stream: started","id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.296386794Z","level":"INFO","msg":"writer: started","stream_id":"42v9byh1"}
+{"time":"2026-08-16T19:15:49.296391062Z","level":"INFO","msg":"sender: started","stream_id":"42v9byh1"}
+{"time":"2026-08-16T20:02:22.815923116Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:02:23.068725923Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:02:23.075143059Z","level":"INFO","msg":"stream: closing","id":"42v9byh1"}
+{"time":"2026-08-16T20:02:23.07515987Z","level":"INFO","msg":"handler: closed","stream_id":"42v9byh1"}
+{"time":"2026-08-16T20:02:23.076217748Z","level":"INFO","msg":"sender: closed","stream_id":"42v9byh1"}
+{"time":"2026-08-16T20:02:23.076241342Z","level":"INFO","msg":"stream: closed","id":"42v9byh1"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug.log b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..db5896eedb5ee02c98e4f5e623debfa54e0bfdfb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug.log
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/logs/debug-internal.log
+2026-08-16 19:15:48,732 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:48,733 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/cfg/side', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz', 'image_view': 'side', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.1, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-cfg-side-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,side', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:48,733 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:48,959 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:48,970 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:48,972 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:48,994 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:49,617 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:50,192 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:50,207 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:02:21,876 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/42v9byh1
+2026-08-16 20:02:21,877 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:02:21,878 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:02:21,878 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:02:23,073 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/run-42v9byh1.wandb b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/run-42v9byh1.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..070f9ad3a79f9cc3324c78a43269aa1593c43e2f
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/cfg/side/wandb/wandb/run-20260816_191548-42v9byh1/run-42v9byh1.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94952d191b57bd436e317644742d14944c6a5d9881df2c71e4402436277e0715
+size 232419
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/code_provenance.json b/anchorgen_baselines_20260923/input_cond/runs/plain/front/code_provenance.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d65c77fa6bb7025a1484f4ad82c6b647f7a2315
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/code_provenance.json
@@ -0,0 +1,7 @@
+{
+  "file": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "commit": "81d95989e1279e432c4145933bf9ce02c3f70f77",
+  "branch": "flow-field",
+  "train_cond_flow_sha256": "095053598f77923a8c8fbdd63bb77280bd641d5f8ca004bc100e63b82bae7616",
+  "seiler_cond_backbone_sha256": "908f1f3b3756ac054e640b4e6e9fe7a40e2130adcf7ba0ce09432f74783dd97c"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_best.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_best.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1ab1c62b453b1fb2f36e2af22ceb4e034c239305
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_best.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6d6d22e5da0a75084acb9291d5c6b67174c820fc69574be85a601843fb03302
+size 329404702
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_last.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_last.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f483d05033144550078cdfce341d0a172975384f
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_last.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e8948dfb1b787c254a0b5232e0854dc9a28a40fcd57a270cd0d9cfbb52225d9
+size 988212490
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_step_005000.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_step_005000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..125a1848cd92c9d031c3d88053e4c981768ded46
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_step_005000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3f3691e21ef5e69986b98d160526bb8e7742d305e0bd0c9dcefee6b3cc62be0
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_step_010000.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_step_010000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6ed1bc61ac6a2f2b1a9eaab2a358cfedc8dc27dd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/model_step_010000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:827cc3ecb58cf7e7e6f80557dc46737f3dfea51d9259787a75e1147ed7c8ad2e
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/run_config.json b/anchorgen_baselines_20260923/input_cond/runs/plain/front/run_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..5aaa97bfe62be4db8ecee62d0b94411011f7dd35
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/run_config.json
@@ -0,0 +1,46 @@
+{
+  "mode": "train",
+  "out_dir": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front",
+  "paired_latents": "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz",
+  "image_view": "front",
+  "seed": 20260526,
+  "val_fraction": 0.05,
+  "max_val": 128,
+  "val_overlap_train": true,
+  "shape_scale": 0.0,
+  "image_scale": 0.0,
+  "hidden_size": 512,
+  "depth": 12,
+  "num_heads": 8,
+  "head_dim": 64,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "pos_embed": "2d",
+  "attention_backend": "sdpa",
+  "gradient_checkpointing": false,
+  "cond_dropout": 0.0,
+  "batch_size": 128,
+  "num_workers": 2,
+  "steps": 10000,
+  "lr": 0.0001,
+  "weight_decay": 0.01,
+  "log_every": 100,
+  "eval_every": 500,
+  "eval_noise_seed": 20260712,
+  "save_every": 5000,
+  "resume": true,
+  "sample_count": 8,
+  "sample_steps": 50,
+  "sample_seed": 20260712,
+  "sample_checkpoint": "model_best.pt",
+  "guidance_scale": 1.0,
+  "decode_mesh": false,
+  "shape_resolution": 512,
+  "shape_max_batch": 32768,
+  "wandb": true,
+  "wandb_project": "drivaer-qwen3v-input-cond",
+  "wandb_entity": "kanydao-epfl",
+  "wandb_run_name": "drivaer3v-inputcond-plain-front-0816c",
+  "wandb_tags": "drivaer,qwen3v,input-conditional,channel-concat,front",
+  "wandb_mode": "online"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/summary.json b/anchorgen_baselines_20260923/input_cond/runs/plain/front/summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..ebb894f1dc1c87054238d64b2d264b2df56cdde4
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/summary.json
@@ -0,0 +1,10 @@
+{
+  "image_view": "front",
+  "best_val_shape": 0.825598955154419,
+  "steps": 10000,
+  "train_samples": 3051,
+  "val_samples": 128,
+  "shape_scale": 1.7132293052897125,
+  "image_scale": 1.6039202568304356,
+  "conditioning": "sketch_latent_channel_concat"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/train_history.json b/anchorgen_baselines_20260923/input_cond/runs/plain/front/train_history.json
new file mode 100644
index 0000000000000000000000000000000000000000..f2808459a12e81774159b9e44511b6ac81b669a7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/train_history.json
@@ -0,0 +1,692 @@
+[
+  {
+    "step": 1,
+    "train_shape": 2.0095300674438477,
+    "lr": 9.999999765596898e-05,
+    "elapsed_sec": 1.0129308700561523
+  },
+  {
+    "step": 1,
+    "val_shape": 2.006056785583496
+  },
+  {
+    "step": 100,
+    "train_shape": 1.268984317779541,
+    "lr": 9.997656161737219e-05,
+    "elapsed_sec": 37.41489315032959
+  },
+  {
+    "step": 200,
+    "train_shape": 1.0299079418182373,
+    "lr": 9.990626960034288e-05,
+    "elapsed_sec": 66.01066493988037
+  },
+  {
+    "step": 300,
+    "train_shape": 0.9263432025909424,
+    "lr": 9.978919331864633e-05,
+    "elapsed_sec": 94.67685151100159
+  },
+  {
+    "step": 400,
+    "train_shape": 0.8832083940505981,
+    "lr": 9.962544831243784e-05,
+    "elapsed_sec": 123.1138174533844
+  },
+  {
+    "step": 500,
+    "train_shape": 0.8710054159164429,
+    "lr": 9.94151961782692e-05,
+    "elapsed_sec": 151.58315086364746
+  },
+  {
+    "step": 500,
+    "val_shape": 0.9266958236694336
+  },
+  {
+    "step": 600,
+    "train_shape": 0.869126558303833,
+    "lr": 9.915864440961283e-05,
+    "elapsed_sec": 180.99633479118347
+  },
+  {
+    "step": 700,
+    "train_shape": 0.8065810799598694,
+    "lr": 9.885604619209063e-05,
+    "elapsed_sec": 209.4696888923645
+  },
+  {
+    "step": 800,
+    "train_shape": 0.8864577412605286,
+    "lr": 9.85077001536101e-05,
+    "elapsed_sec": 238.14472723007202
+  },
+  {
+    "step": 900,
+    "train_shape": 0.8431312441825867,
+    "lr": 9.811395006965495e-05,
+    "elapsed_sec": 266.6265549659729
+  },
+  {
+    "step": 1000,
+    "train_shape": 0.8453853726387024,
+    "lr": 9.767518452401991e-05,
+    "elapsed_sec": 295.1067485809326
+  },
+  {
+    "step": 1000,
+    "val_shape": 0.9003417491912842
+  },
+  {
+    "step": 1100,
+    "train_shape": 0.7959873080253601,
+    "lr": 9.71918365253258e-05,
+    "elapsed_sec": 324.4971194267273
+  },
+  {
+    "step": 1200,
+    "train_shape": 0.8778124451637268,
+    "lr": 9.666438307969207e-05,
+    "elapsed_sec": 352.98753905296326
+  },
+  {
+    "step": 1300,
+    "train_shape": 0.8215359449386597,
+    "lr": 9.60933447199893e-05,
+    "elapsed_sec": 381.49539065361023
+  },
+  {
+    "step": 1400,
+    "train_shape": 0.8760049343109131,
+    "lr": 9.547928499213621e-05,
+    "elapsed_sec": 409.98854970932007
+  },
+  {
+    "step": 1500,
+    "train_shape": 0.8190972805023193,
+    "lr": 9.482280989894765e-05,
+    "elapsed_sec": 438.4830906391144
+  },
+  {
+    "step": 1500,
+    "val_shape": 0.8850529193878174
+  },
+  {
+    "step": 1600,
+    "train_shape": 0.8722271919250488,
+    "lr": 9.412456730208378e-05,
+    "elapsed_sec": 468.0896770954132
+  },
+  {
+    "step": 1700,
+    "train_shape": 0.7841859459877014,
+    "lr": 9.338524628268747e-05,
+    "elapsed_sec": 496.5826110839844
+  },
+  {
+    "step": 1800,
+    "train_shape": 0.8432357311248779,
+    "lr": 9.260557646134585e-05,
+    "elapsed_sec": 525.0737829208374
+  },
+  {
+    "step": 1900,
+    "train_shape": 0.8450958728790283,
+    "lr": 9.178632727804172e-05,
+    "elapsed_sec": 553.7921257019043
+  },
+  {
+    "step": 2000,
+    "train_shape": 0.8576456308364868,
+    "lr": 9.092830723281015e-05,
+    "elapsed_sec": 582.2766716480255
+  },
+  {
+    "step": 2000,
+    "val_shape": 0.8728685975074768
+  },
+  {
+    "step": 2100,
+    "train_shape": 0.8240158557891846,
+    "lr": 9.003236308784539e-05,
+    "elapsed_sec": 611.6931269168854
+  },
+  {
+    "step": 2200,
+    "train_shape": 0.8200933337211609,
+    "lr": 8.90993790318501e-05,
+    "elapsed_sec": 640.1839456558228
+  },
+  {
+    "step": 2300,
+    "train_shape": 0.8790189623832703,
+    "lr": 8.813027580744691e-05,
+    "elapsed_sec": 668.7048745155334
+  },
+  {
+    "step": 2400,
+    "train_shape": 0.7859760522842407,
+    "lr": 8.712600980251722e-05,
+    "elapsed_sec": 697.2973303794861
+  },
+  {
+    "step": 2500,
+    "train_shape": 0.8636320233345032,
+    "lr": 8.60875721063612e-05,
+    "elapsed_sec": 725.8184547424316
+  },
+  {
+    "step": 2500,
+    "val_shape": 0.8659746050834656
+  },
+  {
+    "step": 2600,
+    "train_shape": 0.8139171004295349,
+    "lr": 8.501598753161292e-05,
+    "elapsed_sec": 755.2402539253235
+  },
+  {
+    "step": 2700,
+    "train_shape": 0.8120536208152771,
+    "lr": 8.391231360287367e-05,
+    "elapsed_sec": 783.916437625885
+  },
+  {
+    "step": 2800,
+    "train_shape": 0.7672505378723145,
+    "lr": 8.277763951306297e-05,
+    "elapsed_sec": 812.4099626541138
+  },
+  {
+    "step": 2900,
+    "train_shape": 0.826414942741394,
+    "lr": 8.161308504851662e-05,
+    "elapsed_sec": 840.9039149284363
+  },
+  {
+    "step": 3000,
+    "train_shape": 0.8040660619735718,
+    "lr": 8.04197994838928e-05,
+    "elapsed_sec": 869.4113848209381
+  },
+  {
+    "step": 3000,
+    "val_shape": 0.860557496547699
+  },
+  {
+    "step": 3100,
+    "train_shape": 0.7908036708831787,
+    "lr": 7.919896044797652e-05,
+    "elapsed_sec": 898.8148198127747
+  },
+  {
+    "step": 3200,
+    "train_shape": 0.8107312917709351,
+    "lr": 7.79517727615026e-05,
+    "elapsed_sec": 927.3218214511871
+  },
+  {
+    "step": 3300,
+    "train_shape": 0.8105617761611938,
+    "lr": 7.667946724814286e-05,
+    "elapsed_sec": 955.821857213974
+  },
+  {
+    "step": 3400,
+    "train_shape": 0.8060175180435181,
+    "lr": 7.538329951983173e-05,
+    "elapsed_sec": 984.329461812973
+  },
+  {
+    "step": 3500,
+    "train_shape": 0.8581928610801697,
+    "lr": 7.406454873762866e-05,
+    "elapsed_sec": 1012.8501288890839
+  },
+  {
+    "step": 3500,
+    "val_shape": 0.8549162149429321
+  },
+  {
+    "step": 3600,
+    "train_shape": 0.7955557107925415,
+    "lr": 7.272451634934115e-05,
+    "elapsed_sec": 1042.239506483078
+  },
+  {
+    "step": 3700,
+    "train_shape": 0.7935442924499512,
+    "lr": 7.136452480515233e-05,
+    "elapsed_sec": 1070.735456943512
+  },
+  {
+    "step": 3800,
+    "train_shape": 0.8031713962554932,
+    "lr": 6.99859162525224e-05,
+    "elapsed_sec": 1099.3385877609253
+  },
+  {
+    "step": 3900,
+    "train_shape": 0.8015480041503906,
+    "lr": 6.859005121165148e-05,
+    "elapsed_sec": 1127.845385313034
+  },
+  {
+    "step": 4000,
+    "train_shape": 0.8419222831726074,
+    "lr": 6.71783072328101e-05,
+    "elapsed_sec": 1156.3482077121735
+  },
+  {
+    "step": 4000,
+    "val_shape": 0.8485618829727173
+  },
+  {
+    "step": 4100,
+    "train_shape": 0.7777034044265747,
+    "lr": 6.575207753686356e-05,
+    "elapsed_sec": 1185.8670597076416
+  },
+  {
+    "step": 4200,
+    "train_shape": 0.7778583765029907,
+    "lr": 6.431276964033111e-05,
+    "elapsed_sec": 1214.3739998340607
+  },
+  {
+    "step": 4300,
+    "train_shape": 0.7850451469421387,
+    "lr": 6.286180396633663e-05,
+    "elapsed_sec": 1242.878157377243
+  },
+  {
+    "step": 4400,
+    "train_shape": 0.8449586629867554,
+    "lr": 6.140061244282311e-05,
+    "elapsed_sec": 1271.3767983913422
+  },
+  {
+    "step": 4500,
+    "train_shape": 0.7664374113082886,
+    "lr": 5.9930637089412255e-05,
+    "elapsed_sec": 1299.8652172088623
+  },
+  {
+    "step": 4500,
+    "val_shape": 0.8460968136787415
+  },
+  {
+    "step": 4600,
+    "train_shape": 0.746739387512207,
+    "lr": 5.845332859430572e-05,
+    "elapsed_sec": 1329.2865314483643
+  },
+  {
+    "step": 4700,
+    "train_shape": 0.7821491360664368,
+    "lr": 5.697014488263067e-05,
+    "elapsed_sec": 1357.8854684829712
+  },
+  {
+    "step": 4800,
+    "train_shape": 0.7888714671134949,
+    "lr": 5.548254967764354e-05,
+    "elapsed_sec": 1386.4099190235138
+  },
+  {
+    "step": 4900,
+    "train_shape": 0.8487134575843811,
+    "lr": 5.3992011056212214e-05,
+    "elapsed_sec": 1415.0614676475525
+  },
+  {
+    "step": 5000,
+    "train_shape": 0.8082961440086365,
+    "lr": 5.250000000000107e-05,
+    "elapsed_sec": 1443.5634636878967
+  },
+  {
+    "step": 5000,
+    "val_shape": 0.8414024710655212
+  },
+  {
+    "step": 5100,
+    "train_shape": 0.8482471704483032,
+    "lr": 5.100798894378995e-05,
+    "elapsed_sec": 1475.0750432014465
+  },
+  {
+    "step": 5200,
+    "train_shape": 0.8305211067199707,
+    "lr": 4.951745032235864e-05,
+    "elapsed_sec": 1503.5678346157074
+  },
+  {
+    "step": 5300,
+    "train_shape": 0.7925020456314087,
+    "lr": 4.8029855117371535e-05,
+    "elapsed_sec": 1532.0675690174103
+  },
+  {
+    "step": 5400,
+    "train_shape": 0.7910262942314148,
+    "lr": 4.654667140569645e-05,
+    "elapsed_sec": 1560.5577907562256
+  },
+  {
+    "step": 5500,
+    "train_shape": 0.7911370396614075,
+    "lr": 4.506936291058995e-05,
+    "elapsed_sec": 1589.0530514717102
+  },
+  {
+    "step": 5500,
+    "val_shape": 0.8389391899108887
+  },
+  {
+    "step": 5600,
+    "train_shape": 0.7431558966636658,
+    "lr": 4.359938755717899e-05,
+    "elapsed_sec": 1618.5013201236725
+  },
+  {
+    "step": 5700,
+    "train_shape": 0.7530568838119507,
+    "lr": 4.213819603366514e-05,
+    "elapsed_sec": 1646.9949688911438
+  },
+  {
+    "step": 5800,
+    "train_shape": 0.8088139295578003,
+    "lr": 4.0687230359670296e-05,
+    "elapsed_sec": 1675.4983611106873
+  },
+  {
+    "step": 5900,
+    "train_shape": 0.7627294063568115,
+    "lr": 3.924792246313749e-05,
+    "elapsed_sec": 1703.990826368332
+  },
+  {
+    "step": 6000,
+    "train_shape": 0.7609977722167969,
+    "lr": 3.7821692767190845e-05,
+    "elapsed_sec": 1732.4819633960724
+  },
+  {
+    "step": 6000,
+    "val_shape": 0.8357729911804199
+  },
+  {
+    "step": 6100,
+    "train_shape": 0.7633602619171143,
+    "lr": 3.640994878834947e-05,
+    "elapsed_sec": 1761.9376239776611
+  },
+  {
+    "step": 6200,
+    "train_shape": 0.7686607837677002,
+    "lr": 3.501408374747862e-05,
+    "elapsed_sec": 1790.4302072525024
+  },
+  {
+    "step": 6300,
+    "train_shape": 0.724835991859436,
+    "lr": 3.363547519484887e-05,
+    "elapsed_sec": 1818.9151651859283
+  },
+  {
+    "step": 6400,
+    "train_shape": 0.7138386964797974,
+    "lr": 3.227548365066013e-05,
+    "elapsed_sec": 1847.4060201644897
+  },
+  {
+    "step": 6500,
+    "train_shape": 0.7321893572807312,
+    "lr": 3.0935451262372706e-05,
+    "elapsed_sec": 1875.9163136482239
+  },
+  {
+    "step": 6500,
+    "val_shape": 0.8322945833206177
+  },
+  {
+    "step": 6600,
+    "train_shape": 0.7628726959228516,
+    "lr": 2.9616700480169673e-05,
+    "elapsed_sec": 1905.3679604530334
+  },
+  {
+    "step": 6700,
+    "train_shape": 0.8032939434051514,
+    "lr": 2.8320532751858457e-05,
+    "elapsed_sec": 1933.9090111255646
+  },
+  {
+    "step": 6800,
+    "train_shape": 0.7815189957618713,
+    "lr": 2.7048227238498715e-05,
+    "elapsed_sec": 1962.396309375763
+  },
+  {
+    "step": 6900,
+    "train_shape": 0.7566124796867371,
+    "lr": 2.5801039552024805e-05,
+    "elapsed_sec": 1990.93736577034
+  },
+  {
+    "step": 7000,
+    "train_shape": 0.8027796149253845,
+    "lr": 2.458020051610849e-05,
+    "elapsed_sec": 2019.5351889133453
+  },
+  {
+    "step": 7000,
+    "val_shape": 0.8310120105743408
+  },
+  {
+    "step": 7100,
+    "train_shape": 0.7617167234420776,
+    "lr": 2.3386914951484544e-05,
+    "elapsed_sec": 2048.991213083267
+  },
+  {
+    "step": 7200,
+    "train_shape": 0.7901880741119385,
+    "lr": 2.222236048693809e-05,
+    "elapsed_sec": 2077.478776693344
+  },
+  {
+    "step": 7300,
+    "train_shape": 0.8377978801727295,
+    "lr": 2.1087686397127328e-05,
+    "elapsed_sec": 2105.965715408325
+  },
+  {
+    "step": 7400,
+    "train_shape": 0.8018462657928467,
+    "lr": 1.998401246838802e-05,
+    "elapsed_sec": 2134.462625026703
+  },
+  {
+    "step": 7500,
+    "train_shape": 0.7704964876174927,
+    "lr": 1.891242789363976e-05,
+    "elapsed_sec": 2162.948211669922
+  },
+  {
+    "step": 7500,
+    "val_shape": 0.8296697735786438
+  },
+  {
+    "step": 7600,
+    "train_shape": 0.7745089530944824,
+    "lr": 1.787399019748373e-05,
+    "elapsed_sec": 2192.3724250793457
+  },
+  {
+    "step": 7700,
+    "train_shape": 0.8334223031997681,
+    "lr": 1.6869724192553968e-05,
+    "elapsed_sec": 2220.852392911911
+  },
+  {
+    "step": 7800,
+    "train_shape": 0.7865667343139648,
+    "lr": 1.5900620968150808e-05,
+    "elapsed_sec": 2249.6258058547974
+  },
+  {
+    "step": 7900,
+    "train_shape": 0.7727748155593872,
+    "lr": 1.496763691215551e-05,
+    "elapsed_sec": 2278.113433599472
+  },
+  {
+    "step": 8000,
+    "train_shape": 0.7897841930389404,
+    "lr": 1.4071692767190806e-05,
+    "elapsed_sec": 2306.6025247573853
+  },
+  {
+    "step": 8000,
+    "val_shape": 0.8280940055847168
+  },
+  {
+    "step": 8100,
+    "train_shape": 0.8125532865524292,
+    "lr": 1.3213672721959122e-05,
+    "elapsed_sec": 2336.050815820694
+  },
+  {
+    "step": 8200,
+    "train_shape": 0.7164137363433838,
+    "lr": 1.2394423538655062e-05,
+    "elapsed_sec": 2364.542544364929
+  },
+  {
+    "step": 8300,
+    "train_shape": 0.788827121257782,
+    "lr": 1.1614753717313381e-05,
+    "elapsed_sec": 2393.028420448303
+  },
+  {
+    "step": 8400,
+    "train_shape": 0.7654622793197632,
+    "lr": 1.0875432697917097e-05,
+    "elapsed_sec": 2421.5211737155914
+  },
+  {
+    "step": 8500,
+    "train_shape": 0.7531470060348511,
+    "lr": 1.0177190101053074e-05,
+    "elapsed_sec": 2450.013347864151
+  },
+  {
+    "step": 8500,
+    "val_shape": 0.8269951343536377
+  },
+  {
+    "step": 8600,
+    "train_shape": 0.7753704786300659,
+    "lr": 9.520715007864551e-06,
+    "elapsed_sec": 2479.501224040985
+  },
+  {
+    "step": 8700,
+    "train_shape": 0.7819291353225708,
+    "lr": 8.906655280011289e-06,
+    "elapsed_sec": 2507.9908044338226
+  },
+  {
+    "step": 8800,
+    "train_shape": 0.8123645782470703,
+    "lr": 8.335616920308386e-06,
+    "elapsed_sec": 2536.475681781769
+  },
+  {
+    "step": 8900,
+    "train_shape": 0.7559694051742554,
+    "lr": 7.808163474674577e-06,
+    "elapsed_sec": 2565.1452107429504
+  },
+  {
+    "step": 9000,
+    "train_shape": 0.7610799074172974,
+    "lr": 7.32481547598044e-06,
+    "elapsed_sec": 2593.632852077484
+  },
+  {
+    "step": 9000,
+    "val_shape": 0.8260221481323242
+  },
+  {
+    "step": 9100,
+    "train_shape": 0.8018679618835449,
+    "lr": 6.886049930345396e-06,
+    "elapsed_sec": 2623.0626072883606
+  },
+  {
+    "step": 9200,
+    "train_shape": 0.8077034950256348,
+    "lr": 6.492299846390179e-06,
+    "elapsed_sec": 2651.543227672577
+  },
+  {
+    "step": 9300,
+    "train_shape": 0.7794169187545776,
+    "lr": 6.143953807909618e-06,
+    "elapsed_sec": 2680.1222846508026
+  },
+  {
+    "step": 9400,
+    "train_shape": 0.7416903972625732,
+    "lr": 5.841355590387376e-06,
+    "elapsed_sec": 2708.6076984405518
+  },
+  {
+    "step": 9500,
+    "train_shape": 0.7934188842773438,
+    "lr": 5.5848038217310204e-06,
+    "elapsed_sec": 2737.092348575592
+  },
+  {
+    "step": 9500,
+    "val_shape": 0.8256533145904541
+  },
+  {
+    "step": 9600,
+    "train_shape": 0.7625489830970764,
+    "lr": 5.374551687562344e-06,
+    "elapsed_sec": 2766.477079629898
+  },
+  {
+    "step": 9700,
+    "train_shape": 0.7572126388549805,
+    "lr": 5.2108066813537186e-06,
+    "elapsed_sec": 2794.95827126503
+  },
+  {
+    "step": 9800,
+    "train_shape": 0.7812173962593079,
+    "lr": 5.093730399657109e-06,
+    "elapsed_sec": 2823.44531083107
+  },
+  {
+    "step": 9900,
+    "train_shape": 0.7539259195327759,
+    "lr": 5.023438382627753e-06,
+    "elapsed_sec": 2851.9267842769623
+  },
+  {
+    "step": 10000,
+    "train_shape": 0.8094518780708313,
+    "lr": 5e-06,
+    "elapsed_sec": 2880.5140516757965
+  },
+  {
+    "step": 10000,
+    "val_shape": 0.825598955154419
+  }
+]
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..ad00856974a61e47f9a7e93da674691ce0f084dd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:42.721940446Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:43.043406151Z","level":"INFO","msg":"stream: created new stream","id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.043475229Z","level":"INFO","msg":"handler: started","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.045208076Z","level":"INFO","msg":"stream: started","id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.045281642Z","level":"INFO","msg":"writer: started","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.04529117Z","level":"INFO","msg":"sender: started","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:49.850800315Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:50.087652489Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:50.094525199Z","level":"INFO","msg":"stream: closing","id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:50.094537812Z","level":"INFO","msg":"handler: closed","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:50.095593506Z","level":"INFO","msg":"sender: closed","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:50.095603835Z","level":"INFO","msg":"stream: closed","id":"xp14m7ij"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/debug.log b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..4dc63d5b7d07ae9f6c26192d27b9149522cfcc46
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Configure stats pid to 95
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug.log
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug-internal.log
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz', 'image_view': 'front', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.0, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-plain-front-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,front', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:42,713 INFO    MainThread:95 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:42,718 INFO    MainThread:95 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:42,721 INFO    MainThread:95 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:42,734 INFO    MainThread:95 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:43,401 INFO    MainThread:95 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:43,851 INFO    MainThread:95 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:49,189 INFO    MainThread:95 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/xp14m7ij
+2026-08-16 20:03:49,190 INFO    MainThread:95 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:49,190 INFO    MainThread:95 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:49,190 INFO    MainThread:95 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:50,090 INFO    MainThread:95 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/config.yaml b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..3943e7b345a98c00e0dd9c5fa0b7a29d3d632cf7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/config.yaml
@@ -0,0 +1,172 @@
+_wandb:
+    value:
+        cli_version: 0.22.3
+        e:
+            yt6lqfoly09z2oyaes0vzdfa45aw3rt4:
+                args:
+                    - --mode
+                    - train
+                    - --image-view
+                    - front
+                    - --paired-latents
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz
+                    - --out-dir
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front
+                    - --steps
+                    - "10000"
+                    - --batch-size
+                    - "128"
+                    - --lr
+                    - "1e-4"
+                    - --weight-decay
+                    - "0.01"
+                    - --seed
+                    - "20260526"
+                    - --pos-embed
+                    - 2d
+                    - --cond-dropout
+                    - "0.0"
+                    - --resume
+                    - --wandb
+                    - --wandb-run-name
+                    - drivaer3v-inputcond-plain-front-0816c
+                    - --wandb-tags
+                    - drivaer,qwen3v,input-conditional,channel-concat,front
+                codePath: code/ICLR/customized-design/input_cond/train_cond_flow.py
+                codePathLocal: train_cond_flow.py
+                cpu_count: 32
+                cpu_count_logical: 64
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "7679361679360"
+                        used: "2800196116480"
+                email: 1193414549@qq.com
+                executable: /mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python
+                git:
+                    commit: 81d95989e1279e432c4145933bf9ce02c3f70f77
+                gpu: NVIDIA A100-SXM4-40GB
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "42949672960"
+                      name: NVIDIA A100-SXM4-40GB
+                      uuid: GPU-9b7f1446-d17a-347f-1b5a-20d18d7c7002
+                host: inputcond-plain-front-a100-40g-0816c-0-0
+                memory:
+                    total: "540741947392"
+                os: Linux-6.8.0-124-generic-x86_64-with-glibc2.35
+                program: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py
+                python: CPython 3.10.18
+                root: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front/wandb
+                startedAt: "2026-08-16T19:15:42.471231Z"
+                writerId: yt6lqfoly09z2oyaes0vzdfa45aw3rt4
+        m: []
+        python_version: 3.10.18
+        t:
+            "1":
+                - 1
+                - 105
+            "2":
+                - 1
+                - 105
+            "3":
+                - 2
+                - 13
+                - 15
+                - 16
+                - 61
+                - 62
+            "4": 3.10.18
+            "5": 0.22.3
+            "12": 0.22.3
+            "13": linux-x86_64
+attention_backend:
+    value: sdpa
+batch_size:
+    value: 128
+cond_dropout:
+    value: 0
+decode_mesh:
+    value: false
+depth:
+    value: 12
+dropout:
+    value: 0
+eval_every:
+    value: 500
+eval_noise_seed:
+    value: 20260712
+gradient_checkpointing:
+    value: false
+guidance_scale:
+    value: 1
+head_dim:
+    value: 64
+hidden_size:
+    value: 512
+image_scale:
+    value: 0
+image_view:
+    value: front
+log_every:
+    value: 100
+lr:
+    value: 0.0001
+max_val:
+    value: 128
+mlp_ratio:
+    value: 4
+mode:
+    value: train
+num_heads:
+    value: 8
+num_workers:
+    value: 2
+out_dir:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front
+paired_latents:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz
+pos_embed:
+    value: 2d
+resume:
+    value: true
+sample_checkpoint:
+    value: model_best.pt
+sample_count:
+    value: 8
+sample_seed:
+    value: 20260712
+sample_steps:
+    value: 50
+save_every:
+    value: 5000
+seed:
+    value: 20260526
+shape_max_batch:
+    value: 32768
+shape_resolution:
+    value: 512
+shape_scale:
+    value: 0
+steps:
+    value: 10000
+val_fraction:
+    value: 0.05
+val_overlap_train:
+    value: true
+wandb:
+    value: true
+wandb_entity:
+    value: kanydao-epfl
+wandb_mode:
+    value: online
+wandb_project:
+    value: drivaer-qwen3v-input-cond
+wandb_run_name:
+    value: drivaer3v-inputcond-plain-front-0816c
+wandb_tags:
+    value: drivaer,qwen3v,input-conditional,channel-concat,front
+weight_decay:
+    value: 0.01
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/output.log b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..c07d0694a574bd6bc7e7e0d28d047c69fed06c98
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/output.log
@@ -0,0 +1,123 @@
+[wandb] project=drivaer-qwen3v-input-cond run=drivaer3v-inputcond-plain-front-0816c mode=online
+[train] step=000001 loss=2.009530 lr=1.00e-04
+[val] step=000001 shape=2.006057
+[train] step=000100 loss=1.268984 lr=1.00e-04
+[train] step=000200 loss=1.029908 lr=9.99e-05
+[train] step=000300 loss=0.926343 lr=9.98e-05
+[train] step=000400 loss=0.883208 lr=9.96e-05
+[train] step=000500 loss=0.871005 lr=9.94e-05
+[val] step=000500 shape=0.926696
+[train] step=000600 loss=0.869127 lr=9.92e-05
+[train] step=000700 loss=0.806581 lr=9.89e-05
+[train] step=000800 loss=0.886458 lr=9.85e-05
+[train] step=000900 loss=0.843131 lr=9.81e-05
+[train] step=001000 loss=0.845385 lr=9.77e-05
+[val] step=001000 shape=0.900342
+[train] step=001100 loss=0.795987 lr=9.72e-05
+[train] step=001200 loss=0.877812 lr=9.67e-05
+[train] step=001300 loss=0.821536 lr=9.61e-05
+[train] step=001400 loss=0.876005 lr=9.55e-05
+[train] step=001500 loss=0.819097 lr=9.48e-05
+[val] step=001500 shape=0.885053
+[train] step=001600 loss=0.872227 lr=9.41e-05
+[train] step=001700 loss=0.784186 lr=9.34e-05
+[train] step=001800 loss=0.843236 lr=9.26e-05
+[train] step=001900 loss=0.845096 lr=9.18e-05
+[train] step=002000 loss=0.857646 lr=9.09e-05
+[val] step=002000 shape=0.872869
+[train] step=002100 loss=0.824016 lr=9.00e-05
+[train] step=002200 loss=0.820093 lr=8.91e-05
+[train] step=002300 loss=0.879019 lr=8.81e-05
+[train] step=002400 loss=0.785976 lr=8.71e-05
+[train] step=002500 loss=0.863632 lr=8.61e-05
+[val] step=002500 shape=0.865975
+[train] step=002600 loss=0.813917 lr=8.50e-05
+[train] step=002700 loss=0.812054 lr=8.39e-05
+[train] step=002800 loss=0.767251 lr=8.28e-05
+[train] step=002900 loss=0.826415 lr=8.16e-05
+[train] step=003000 loss=0.804066 lr=8.04e-05
+[val] step=003000 shape=0.860557
+[train] step=003100 loss=0.790804 lr=7.92e-05
+[train] step=003200 loss=0.810731 lr=7.80e-05
+[train] step=003300 loss=0.810562 lr=7.67e-05
+[train] step=003400 loss=0.806018 lr=7.54e-05
+[train] step=003500 loss=0.858193 lr=7.41e-05
+[val] step=003500 shape=0.854916
+[train] step=003600 loss=0.795556 lr=7.27e-05
+[train] step=003700 loss=0.793544 lr=7.14e-05
+[train] step=003800 loss=0.803171 lr=7.00e-05
+[train] step=003900 loss=0.801548 lr=6.86e-05
+[train] step=004000 loss=0.841922 lr=6.72e-05
+[val] step=004000 shape=0.848562
+[train] step=004100 loss=0.777703 lr=6.58e-05
+[train] step=004200 loss=0.777858 lr=6.43e-05
+[train] step=004300 loss=0.785045 lr=6.29e-05
+[train] step=004400 loss=0.844959 lr=6.14e-05
+[train] step=004500 loss=0.766437 lr=5.99e-05
+[val] step=004500 shape=0.846097
+[train] step=004600 loss=0.746739 lr=5.85e-05
+[train] step=004700 loss=0.782149 lr=5.70e-05
+[train] step=004800 loss=0.788871 lr=5.55e-05
+[train] step=004900 loss=0.848713 lr=5.40e-05
+[train] step=005000 loss=0.808296 lr=5.25e-05
+[val] step=005000 shape=0.841402
+[train] step=005100 loss=0.848247 lr=5.10e-05
+[train] step=005200 loss=0.830521 lr=4.95e-05
+[train] step=005300 loss=0.792502 lr=4.80e-05
+[train] step=005400 loss=0.791026 lr=4.65e-05
+[train] step=005500 loss=0.791137 lr=4.51e-05
+[val] step=005500 shape=0.838939
+[train] step=005600 loss=0.743156 lr=4.36e-05
+[train] step=005700 loss=0.753057 lr=4.21e-05
+[train] step=005800 loss=0.808814 lr=4.07e-05
+[train] step=005900 loss=0.762729 lr=3.92e-05
+[train] step=006000 loss=0.760998 lr=3.78e-05
+[val] step=006000 shape=0.835773
+[train] step=006100 loss=0.763360 lr=3.64e-05
+[train] step=006200 loss=0.768661 lr=3.50e-05
+[train] step=006300 loss=0.724836 lr=3.36e-05
+[train] step=006400 loss=0.713839 lr=3.23e-05
+[train] step=006500 loss=0.732189 lr=3.09e-05
+[val] step=006500 shape=0.832295
+[train] step=006600 loss=0.762873 lr=2.96e-05
+[train] step=006700 loss=0.803294 lr=2.83e-05
+[train] step=006800 loss=0.781519 lr=2.70e-05
+[train] step=006900 loss=0.756612 lr=2.58e-05
+[train] step=007000 loss=0.802780 lr=2.46e-05
+[val] step=007000 shape=0.831012
+[train] step=007100 loss=0.761717 lr=2.34e-05
+[train] step=007200 loss=0.790188 lr=2.22e-05
+[train] step=007300 loss=0.837798 lr=2.11e-05
+[train] step=007400 loss=0.801846 lr=2.00e-05
+[train] step=007500 loss=0.770496 lr=1.89e-05
+[val] step=007500 shape=0.829670
+[train] step=007600 loss=0.774509 lr=1.79e-05
+[train] step=007700 loss=0.833422 lr=1.69e-05
+[train] step=007800 loss=0.786567 lr=1.59e-05
+[train] step=007900 loss=0.772775 lr=1.50e-05
+[train] step=008000 loss=0.789784 lr=1.41e-05
+[val] step=008000 shape=0.828094
+[train] step=008100 loss=0.812553 lr=1.32e-05
+[train] step=008200 loss=0.716414 lr=1.24e-05
+[train] step=008300 loss=0.788827 lr=1.16e-05
+[train] step=008400 loss=0.765462 lr=1.09e-05
+[train] step=008500 loss=0.753147 lr=1.02e-05
+[val] step=008500 shape=0.826995
+[train] step=008600 loss=0.775370 lr=9.52e-06
+[train] step=008700 loss=0.781929 lr=8.91e-06
+[train] step=008800 loss=0.812365 lr=8.34e-06
+[train] step=008900 loss=0.755969 lr=7.81e-06
+[train] step=009000 loss=0.761080 lr=7.32e-06
+[val] step=009000 shape=0.826022
+[train] step=009100 loss=0.801868 lr=6.89e-06
+[train] step=009200 loss=0.807703 lr=6.49e-06
+[train] step=009300 loss=0.779417 lr=6.14e-06
+[train] step=009400 loss=0.741690 lr=5.84e-06
+[train] step=009500 loss=0.793419 lr=5.58e-06
+[val] step=009500 shape=0.825653
+[train] step=009600 loss=0.762549 lr=5.37e-06
+[train] step=009700 loss=0.757213 lr=5.21e-06
+[train] step=009800 loss=0.781217 lr=5.09e-06
+[train] step=009900 loss=0.753926 lr=5.02e-06
+[train] step=010000 loss=0.809452 lr=5.00e-06
+[val] step=010000 shape=0.825599
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/requirements.txt b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..2e10c9ab6bc99041706780299f9e3e5035f34676
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/requirements.txt
@@ -0,0 +1,302 @@
+shellingham==1.5.4
+importlib_metadata==8.7.0
+websocket-client==1.8.0
+rfc3986-validator==0.1.1
+aiohappyeyeballs==2.6.1
+tensorboardX==2.6.2.2
+matplotlib==3.8.4
+zstandard==0.23.0
+gradio_client==1.4.2
+nbclient==0.10.2
+pyaml==25.7.0
+argparse==1.4.0
+h11==0.16.0
+scooby==0.10.1
+pexpect==4.9.0
+jedi==0.19.2
+scikit-optimize==0.10.2
+omegaconf==2.3.0
+tomlkit==0.12.0
+yarl==1.20.1
+point-cloud-utils==0.34.0
+cffi==1.17.1
+asttokens==3.0.0
+webencodings==0.5.1
+imageio==2.34.1
+attrs==25.3.0
+hydra-core==1.3.2
+datasets==2.19.0
+pyquaternion==0.9.9
+trame-common==1.0.0
+wslink==2.3.4
+Pygments==2.19.2
+pydantic==2.11.7
+regex==2024.11.6
+timm==0.9.16
+ipykernel==6.29.5
+prompt_toolkit==3.0.51
+torchdiffeq==0.2.5
+pybind11==2.13.6
+debugpy==1.8.14
+pytz==2025.2
+transformers==4.40.1
+jupyterlab_pygments==0.3.0
+dash==3.1.1
+psutil==5.9.8
+tqdm==4.66.2
+multidict==6.6.3
+setproctitle==1.3.6
+nbformat==5.10.4
+prometheus_client==0.22.1
+jupyter_core==5.8.1
+gradio==5.5.0
+multiprocess==0.70.16
+fastapi==0.115.14
+simpervisor==1.0.0
+comm==0.2.2
+widgetsnbextension==4.0.14
+webcolors==24.11.1
+sentry-sdk==2.32.0
+tornado==6.5.1
+defusedxml==0.7.1
+Markdown==3.8.2
+numba==0.61.2
+pyarrow-hotfix==0.7
+trame-server==3.5.0
+nbconvert==7.16.6
+websockets==12.0
+semantic-version==2.10.0
+bpy==4.0.0
+smmap==5.0.2
+grpcio==1.73.1
+referencing==0.36.2
+annotated-types==0.7.0
+stack-data==0.6.3
+vtk==9.4.2
+safehttpx==0.1.6
+GitPython==3.1.44
+ipywidgets==8.1.7
+platformdirs==4.3.8
+tzdata==2025.2
+typer==0.16.0
+click==8.2.1
+networkx==3.3
+fsspec==2024.3.1
+plotly==6.2.0
+isoduration==20.11.0
+joblib==1.4.0
+aiosignal==1.4.0
+jsonschema-specifications==2025.4.1
+six==1.17.0
+pyzmq==27.0.0
+lazy_loader==0.4
+typeguard==2.13.3
+markdown-it-py==3.0.0
+httpx==0.28.1
+fqdn==1.5.1
+threadpoolctl==3.6.0
+orjson==3.10.18
+docker-pycreds==0.4.0
+jsonschema==4.24.0
+tensorboard-data-server==0.7.2
+nvidia-cufft-cu12==11.0.2.54
+aiofiles==23.2.1
+rpds-py==0.26.0
+Cython==3.1.2
+python-multipart==0.0.12
+nvidia-cuda-runtime-cu12==12.1.105
+huggingface-hub==0.26.2
+frozenlist==1.7.0
+onnxruntime==1.22.0
+cycler==0.12.1
+tifffile==2025.5.10
+safetensors==0.4.3
+nest-asyncio==1.6.0
+parso==0.8.4
+PyMCubes==0.1.4
+nvidia-nvtx-cu12==12.1.105
+pydub==0.25.1
+packaging==25.0
+ipython==8.37.0
+uvicorn==0.35.0
+pyarrow==20.0.0
+rich==14.0.0
+executing==2.2.0
+jupyter_server_terminals==0.5.3
+typing_extensions==4.12.2
+typing_extensions==4.15.0
+tensorboard==2.16.2
+argon2-cffi-bindings==21.2.0
+wcwidth==0.2.13
+typing-inspection==0.4.1
+pure_eval==0.2.3
+trimesh==4.3.2
+nvidia-cudnn-cu12==9.1.0.70
+humanfriendly==10.0
+Werkzeug==3.1.3
+decorator==5.2.1
+nvidia-cuda-nvrtc-cu12==12.1.105
+aiohttp==3.12.13
+uri-template==1.3.0
+anyio==4.9.0
+fastjsonschema==2.21.1
+nvidia-curand-cu12==10.3.2.106
+zipp==3.23.0
+rtree==1.4.1
+kiwisolver==1.4.8
+Flask==3.1.1
+nvidia-cuda-cupti-cu12==12.1.105
+pysdf==0.1.9
+sniffio==1.3.1
+itsdangerous==2.2.0
+rfc3339-validator==0.1.4
+trame==3.10.2
+open3d==0.18.0
+scikit-learn==1.7.0
+mistune==3.1.3
+diffusers==0.31.0
+absl-py==2.3.1
+cubvh==0.1.0
+trame-client==3.9.1
+einops==0.8.0
+plyfile==1.0.3
+bleach==6.2.0
+ptyprocess==0.7.0
+pycparser==2.22
+llvmlite==0.44.0
+addict==2.4.0
+diso==0.1.4
+Send2Trash==1.8.3
+more-itertools==10.7.0
+scikit-image==0.23.2
+propcache==0.3.2
+pandas==2.2.2
+flatbuffers==25.2.10
+gitdb==4.0.12
+jaxtyping==0.2.28
+soupsieve==2.7
+mdurl==0.1.2
+protobuf==5.29.5
+fonttools==4.58.5
+coloredlogs==15.0.1
+spaces==0.28.3
+jupyter_server==2.16.0
+trame-vtk==2.9.0
+terminado==0.18.1
+overrides==7.7.0
+pandocfilters==1.5.1
+nvidia-cusolver-cu12==11.4.5.107
+trame-vuetify==3.0.1
+async-timeout==5.0.1
+blinker==1.9.0
+exceptiongroup==1.3.0
+xxhash==3.5.0
+python-json-logger==3.3.0
+pyvista==0.45.2
+python-dateutil==2.9.0.post0
+ConfigArgParse==1.7.1
+PyMatting==1.1.14
+fpsample==0.3.3
+traitlets==5.14.3
+msgpack==1.1.1
+argon2-cffi==25.1.0
+tinycss2==1.4.0
+retrying==1.4.0
+jsonpointer==3.0.0
+matplotlib-inline==0.1.7
+jupyter_client==8.6.3
+contourpy==1.3.2
+lightning-utilities==0.11.2
+pydantic_core==2.33.2
+httpcore==1.0.9
+narwhals==1.46.0
+nvidia-cublas-cu12==12.1.3.1
+jupyterlab_widgets==3.0.15
+antlr4-python3-runtime==4.9.3
+pyparsing==3.1.2
+rembg==2.0.59
+ffmpy==0.6.0
+tokenizers==0.19.1
+beautifulsoup4==4.13.4
+pooch==1.8.2
+ninja==1.11.1.4
+jupyter_server_proxy==4.4.0
+ruff==0.12.2
+types-python-dateutil==2.9.0.20250708
+dill==0.3.8
+pytorch-lightning==2.2.4
+jupyter-events==0.12.0
+starlette==0.46.2
+arrow==1.3.0
+setuptools==78.1.1
+wheel==0.45.1
+pip==25.1
+mkl-service==2.5.2
+mpi4py==4.0.3
+h5py==3.14.0
+mkl_fft==2.1.1
+mkl_random==1.3.0
+mcubes==0.1.6
+wandb==0.22.3
+certifi==2026.1.4
+mpmath==1.3.0
+triton==3.1.0
+sympy==1.13.1
+nvidia-nccl-cu12==2.21.5
+nvidia-nvjitlink-cu12==12.9.86
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nvtx-cu11==11.8.86
+nvidia-nccl-cu11==2.21.5
+nvidia-cusparse-cu11==11.7.5.86
+nvidia-curand-cu11==10.3.0.86
+nvidia-cufft-cu11==10.9.0.58
+nvidia-cuda-runtime-cu11==11.8.89
+nvidia-cuda-nvrtc-cu11==11.8.89
+nvidia-cuda-cupti-cu11==11.8.87
+nvidia-cublas-cu11==11.11.3.6
+nvidia-cusolver-cu11==11.4.1.48
+nvidia-cudnn-cu11==9.1.0.70
+torch==2.5.1+cu118
+torchvision==0.20.1+cu118
+torchaudio==2.5.1+cu118
+PyYAML==6.0.3
+pillow==12.1.0
+torchmetrics==1.8.2
+accelerate==1.12.0
+urllib3==2.6.3
+idna==3.11
+charset-normalizer==3.4.4
+requests==2.32.5
+filelock==3.20.3
+torch_cluster==1.6.3+pt25cu118
+numpy==1.26.4
+opencv-python-headless==4.9.0.80
+PyOpenGL-accelerate==3.1.10
+PyOpenGL==3.1.0
+pyglet==2.1.14
+freetype-py==2.5.1
+pyrender==0.1.45
+scipy==1.15.3
+memory-profiler==0.61.0
+pyamg==5.3.0
+meshio==5.3.5
+torch-fem==0.6.2
+torch-geometric==2.7.0
+MarkupSafe==3.0.3
+Jinja2==3.1.6
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/wandb-metadata.json b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..24261fb9bdae3286e6b0299f945df235ef1c32b4
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/wandb-metadata.json
@@ -0,0 +1,69 @@
+{
+  "os":  "Linux-6.8.0-124-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.10.18",
+  "startedAt":  "2026-08-16T19:15:42.471231Z",
+  "args":  [
+    "--mode",
+    "train",
+    "--image-view",
+    "front",
+    "--paired-latents",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz",
+    "--out-dir",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front",
+    "--steps",
+    "10000",
+    "--batch-size",
+    "128",
+    "--lr",
+    "1e-4",
+    "--weight-decay",
+    "0.01",
+    "--seed",
+    "20260526",
+    "--pos-embed",
+    "2d",
+    "--cond-dropout",
+    "0.0",
+    "--resume",
+    "--wandb",
+    "--wandb-run-name",
+    "drivaer3v-inputcond-plain-front-0816c",
+    "--wandb-tags",
+    "drivaer,qwen3v,input-conditional,channel-concat,front"
+  ],
+  "program":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePath":  "code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePathLocal":  "train_cond_flow.py",
+  "git":  {
+    "commit":  "81d95989e1279e432c4145933bf9ce02c3f70f77"
+  },
+  "email":  "1193414549@qq.com",
+  "root":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front/wandb",
+  "host":  "inputcond-plain-front-a100-40g-0816c-0-0",
+  "executable":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python",
+  "cpu_count":  32,
+  "cpu_count_logical":  64,
+  "gpu":  "NVIDIA A100-SXM4-40GB",
+  "gpu_count":  1,
+  "disk":  {
+    "/":  {
+      "total":  "7679361679360",
+      "used":  "2800196116480"
+    }
+  },
+  "memory":  {
+    "total":  "540741947392"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A100-SXM4-40GB",
+      "memoryTotal":  "42949672960",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-9b7f1446-d17a-347f-1b5a-20d18d7c7002"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "yt6lqfoly09z2oyaes0vzdfa45aw3rt4"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/wandb-summary.json b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..a0c2965b1109201dd03348a6fdfc38ae67d9b462
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_wandb":{"runtime":2885},"val/shape":0.825598955154419,"shape_scale":1.7132293052897125,"train_samples":3051,"val_samples":128,"_runtime":2885.789498281,"image_view":"front","image_scale":1.6039202568304356,"_timestamp":1.786910624548519e+09,"steps":10000,"time/elapsed_sec":2880.5140516757965,"best_val_shape":0.825598955154419,"conditioning":"sketch_latent_channel_concat","train/lr":5e-06,"_step":10000,"train/shape":0.8094518780708313}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..ad00856974a61e47f9a7e93da674691ce0f084dd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:42.721940446Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:43.043406151Z","level":"INFO","msg":"stream: created new stream","id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.043475229Z","level":"INFO","msg":"handler: started","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.045208076Z","level":"INFO","msg":"stream: started","id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.045281642Z","level":"INFO","msg":"writer: started","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T19:15:43.04529117Z","level":"INFO","msg":"sender: started","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:49.850800315Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:50.087652489Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:50.094525199Z","level":"INFO","msg":"stream: closing","id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:50.094537812Z","level":"INFO","msg":"handler: closed","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:50.095593506Z","level":"INFO","msg":"sender: closed","stream_id":"xp14m7ij"}
+{"time":"2026-08-16T20:03:50.095603835Z","level":"INFO","msg":"stream: closed","id":"xp14m7ij"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug.log b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..4dc63d5b7d07ae9f6c26192d27b9149522cfcc46
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Configure stats pid to 95
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug.log
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/logs/debug-internal.log
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/front', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/front/paired_latents_train.npz', 'image_view': 'front', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.0, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-plain-front-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,front', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:42,498 INFO    MainThread:95 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:42,713 INFO    MainThread:95 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:42,718 INFO    MainThread:95 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:42,721 INFO    MainThread:95 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:42,734 INFO    MainThread:95 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:43,401 INFO    MainThread:95 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:43,844 INFO    MainThread:95 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:43,851 INFO    MainThread:95 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:49,189 INFO    MainThread:95 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/xp14m7ij
+2026-08-16 20:03:49,190 INFO    MainThread:95 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:49,190 INFO    MainThread:95 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:49,190 INFO    MainThread:95 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:50,090 INFO    MainThread:95 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/run-xp14m7ij.wandb b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/run-xp14m7ij.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..da24c2aeadf7cb37fb97421bf9fc79494839e841
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/front/wandb/wandb/run-20260816_191542-xp14m7ij/run-xp14m7ij.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3d821e7f03fb5995b6505d937232139b8f5f0d32308d392b2b9a7fdfd4a39f8
+size 238194
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/code_provenance.json b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/code_provenance.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d65c77fa6bb7025a1484f4ad82c6b647f7a2315
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/code_provenance.json
@@ -0,0 +1,7 @@
+{
+  "file": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "commit": "81d95989e1279e432c4145933bf9ce02c3f70f77",
+  "branch": "flow-field",
+  "train_cond_flow_sha256": "095053598f77923a8c8fbdd63bb77280bd641d5f8ca004bc100e63b82bae7616",
+  "seiler_cond_backbone_sha256": "908f1f3b3756ac054e640b4e6e9fe7a40e2130adcf7ba0ce09432f74783dd97c"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_best.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_best.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6c27b9120ddae1d2643cf48823ec9913482f3198
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_best.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bed471e5b6579aca1d41f1edd95af862fc6a53cbcfe11582553bb01d53babc4a
+size 329404702
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_last.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_last.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c682b8f4a5cb9ddba4b2790dd114a173c97ecb6a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_last.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df4e9289bd813f9c44da790f2e7620172f2bd8554bef02398383bdfca4f640ed
+size 988212490
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_step_005000.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_step_005000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5f9daae46a705b5d15392d9ffe4dd20a304e9583
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_step_005000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49b9d29efc08d6ae8125ec062a6c699b3463c6cf9bbeeaf74047827f69beddb9
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_step_010000.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_step_010000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f10f9cc4c75f187f514d7df082116be88239f4a1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/model_step_010000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8af3fcf6d2480a3442dc2f25059c1c8dc9e5efeff1dcedc08c9c069ebde52c1a
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/run_config.json b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/run_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..d6f4514c7e9e1ac215e880d34b149ab7051366bb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/run_config.json
@@ -0,0 +1,46 @@
+{
+  "mode": "train",
+  "out_dir": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear",
+  "paired_latents": "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz",
+  "image_view": "rear",
+  "seed": 20260526,
+  "val_fraction": 0.05,
+  "max_val": 128,
+  "val_overlap_train": true,
+  "shape_scale": 0.0,
+  "image_scale": 0.0,
+  "hidden_size": 512,
+  "depth": 12,
+  "num_heads": 8,
+  "head_dim": 64,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "pos_embed": "2d",
+  "attention_backend": "sdpa",
+  "gradient_checkpointing": false,
+  "cond_dropout": 0.0,
+  "batch_size": 128,
+  "num_workers": 2,
+  "steps": 10000,
+  "lr": 0.0001,
+  "weight_decay": 0.01,
+  "log_every": 100,
+  "eval_every": 500,
+  "eval_noise_seed": 20260712,
+  "save_every": 5000,
+  "resume": true,
+  "sample_count": 8,
+  "sample_steps": 50,
+  "sample_seed": 20260712,
+  "sample_checkpoint": "model_best.pt",
+  "guidance_scale": 1.0,
+  "decode_mesh": false,
+  "shape_resolution": 512,
+  "shape_max_batch": 32768,
+  "wandb": true,
+  "wandb_project": "drivaer-qwen3v-input-cond",
+  "wandb_entity": "kanydao-epfl",
+  "wandb_run_name": "drivaer3v-inputcond-plain-rear-0816c",
+  "wandb_tags": "drivaer,qwen3v,input-conditional,channel-concat,rear",
+  "wandb_mode": "online"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/summary.json b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..46507f7c3f8d4c94072d97c5295de1f2cee24448
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/summary.json
@@ -0,0 +1,10 @@
+{
+  "image_view": "rear",
+  "best_val_shape": 0.8280910849571228,
+  "steps": 10000,
+  "train_samples": 3051,
+  "val_samples": 128,
+  "shape_scale": 1.7132293052897125,
+  "image_scale": 1.6045079043256218,
+  "conditioning": "sketch_latent_channel_concat"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/train_history.json b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/train_history.json
new file mode 100644
index 0000000000000000000000000000000000000000..3ba882a18f589e72a78192c9e3b654fe7d562ad5
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/train_history.json
@@ -0,0 +1,692 @@
+[
+  {
+    "step": 1,
+    "train_shape": 2.0095300674438477,
+    "lr": 9.999999765596898e-05,
+    "elapsed_sec": 0.9428911209106445
+  },
+  {
+    "step": 1,
+    "val_shape": 2.006044864654541
+  },
+  {
+    "step": 100,
+    "train_shape": 1.269827127456665,
+    "lr": 9.997656161737219e-05,
+    "elapsed_sec": 37.3280463218689
+  },
+  {
+    "step": 200,
+    "train_shape": 1.0325653553009033,
+    "lr": 9.990626960034288e-05,
+    "elapsed_sec": 65.63210988044739
+  },
+  {
+    "step": 300,
+    "train_shape": 0.9296151399612427,
+    "lr": 9.978919331864633e-05,
+    "elapsed_sec": 94.09020042419434
+  },
+  {
+    "step": 400,
+    "train_shape": 0.8819944858551025,
+    "lr": 9.962544831243784e-05,
+    "elapsed_sec": 122.25593328475952
+  },
+  {
+    "step": 500,
+    "train_shape": 0.8715932369232178,
+    "lr": 9.94151961782692e-05,
+    "elapsed_sec": 150.44360637664795
+  },
+  {
+    "step": 500,
+    "val_shape": 0.930525004863739
+  },
+  {
+    "step": 600,
+    "train_shape": 0.8687801361083984,
+    "lr": 9.915864440961283e-05,
+    "elapsed_sec": 179.70851397514343
+  },
+  {
+    "step": 700,
+    "train_shape": 0.8079893589019775,
+    "lr": 9.885604619209063e-05,
+    "elapsed_sec": 207.916668176651
+  },
+  {
+    "step": 800,
+    "train_shape": 0.8868737816810608,
+    "lr": 9.85077001536101e-05,
+    "elapsed_sec": 236.35090899467468
+  },
+  {
+    "step": 900,
+    "train_shape": 0.8433308005332947,
+    "lr": 9.811395006965495e-05,
+    "elapsed_sec": 264.5979514122009
+  },
+  {
+    "step": 1000,
+    "train_shape": 0.8450058102607727,
+    "lr": 9.767518452401991e-05,
+    "elapsed_sec": 293.17318654060364
+  },
+  {
+    "step": 1000,
+    "val_shape": 0.8978108167648315
+  },
+  {
+    "step": 1100,
+    "train_shape": 0.7957796454429626,
+    "lr": 9.71918365253258e-05,
+    "elapsed_sec": 322.32773303985596
+  },
+  {
+    "step": 1200,
+    "train_shape": 0.8763790130615234,
+    "lr": 9.666438307969207e-05,
+    "elapsed_sec": 350.5464277267456
+  },
+  {
+    "step": 1300,
+    "train_shape": 0.8170459866523743,
+    "lr": 9.60933447199893e-05,
+    "elapsed_sec": 378.77299070358276
+  },
+  {
+    "step": 1400,
+    "train_shape": 0.8748599290847778,
+    "lr": 9.547928499213621e-05,
+    "elapsed_sec": 406.99155020713806
+  },
+  {
+    "step": 1500,
+    "train_shape": 0.8163318634033203,
+    "lr": 9.482280989894765e-05,
+    "elapsed_sec": 435.2138216495514
+  },
+  {
+    "step": 1500,
+    "val_shape": 0.8852865695953369
+  },
+  {
+    "step": 1600,
+    "train_shape": 0.8705066442489624,
+    "lr": 9.412456730208378e-05,
+    "elapsed_sec": 464.4463083744049
+  },
+  {
+    "step": 1700,
+    "train_shape": 0.7831807732582092,
+    "lr": 9.338524628268747e-05,
+    "elapsed_sec": 492.66426157951355
+  },
+  {
+    "step": 1800,
+    "train_shape": 0.8418937921524048,
+    "lr": 9.260557646134585e-05,
+    "elapsed_sec": 520.8827862739563
+  },
+  {
+    "step": 1900,
+    "train_shape": 0.8447175025939941,
+    "lr": 9.178632727804172e-05,
+    "elapsed_sec": 549.1207213401794
+  },
+  {
+    "step": 2000,
+    "train_shape": 0.8559203147888184,
+    "lr": 9.092830723281015e-05,
+    "elapsed_sec": 577.3382239341736
+  },
+  {
+    "step": 2000,
+    "val_shape": 0.8728982210159302
+  },
+  {
+    "step": 2100,
+    "train_shape": 0.8235125541687012,
+    "lr": 9.003236308784539e-05,
+    "elapsed_sec": 606.7079167366028
+  },
+  {
+    "step": 2200,
+    "train_shape": 0.8179695010185242,
+    "lr": 8.90993790318501e-05,
+    "elapsed_sec": 634.9284703731537
+  },
+  {
+    "step": 2300,
+    "train_shape": 0.8786155581474304,
+    "lr": 8.813027580744691e-05,
+    "elapsed_sec": 663.2199635505676
+  },
+  {
+    "step": 2400,
+    "train_shape": 0.7865079045295715,
+    "lr": 8.712600980251722e-05,
+    "elapsed_sec": 691.5572016239166
+  },
+  {
+    "step": 2500,
+    "train_shape": 0.862334132194519,
+    "lr": 8.60875721063612e-05,
+    "elapsed_sec": 719.7881026268005
+  },
+  {
+    "step": 2500,
+    "val_shape": 0.8647055625915527
+  },
+  {
+    "step": 2600,
+    "train_shape": 0.8143407702445984,
+    "lr": 8.501598753161292e-05,
+    "elapsed_sec": 749.037504196167
+  },
+  {
+    "step": 2700,
+    "train_shape": 0.8142162561416626,
+    "lr": 8.391231360287367e-05,
+    "elapsed_sec": 777.2557532787323
+  },
+  {
+    "step": 2800,
+    "train_shape": 0.7679807543754578,
+    "lr": 8.277763951306297e-05,
+    "elapsed_sec": 805.4911637306213
+  },
+  {
+    "step": 2900,
+    "train_shape": 0.8266807794570923,
+    "lr": 8.161308504851662e-05,
+    "elapsed_sec": 833.7185759544373
+  },
+  {
+    "step": 3000,
+    "train_shape": 0.805258572101593,
+    "lr": 8.04197994838928e-05,
+    "elapsed_sec": 861.9650771617889
+  },
+  {
+    "step": 3000,
+    "val_shape": 0.859959602355957
+  },
+  {
+    "step": 3100,
+    "train_shape": 0.7929704189300537,
+    "lr": 7.919896044797652e-05,
+    "elapsed_sec": 891.2469742298126
+  },
+  {
+    "step": 3200,
+    "train_shape": 0.8101007342338562,
+    "lr": 7.79517727615026e-05,
+    "elapsed_sec": 919.4709761142731
+  },
+  {
+    "step": 3300,
+    "train_shape": 0.8105075359344482,
+    "lr": 7.667946724814286e-05,
+    "elapsed_sec": 947.6924600601196
+  },
+  {
+    "step": 3400,
+    "train_shape": 0.80492103099823,
+    "lr": 7.538329951983173e-05,
+    "elapsed_sec": 976.0182056427002
+  },
+  {
+    "step": 3500,
+    "train_shape": 0.8560464382171631,
+    "lr": 7.406454873762866e-05,
+    "elapsed_sec": 1004.2428452968597
+  },
+  {
+    "step": 3500,
+    "val_shape": 0.8548734188079834
+  },
+  {
+    "step": 3600,
+    "train_shape": 0.7947094440460205,
+    "lr": 7.272451634934115e-05,
+    "elapsed_sec": 1033.416578054428
+  },
+  {
+    "step": 3700,
+    "train_shape": 0.7939572334289551,
+    "lr": 7.136452480515233e-05,
+    "elapsed_sec": 1061.6481804847717
+  },
+  {
+    "step": 3800,
+    "train_shape": 0.801822304725647,
+    "lr": 6.99859162525224e-05,
+    "elapsed_sec": 1089.8824651241302
+  },
+  {
+    "step": 3900,
+    "train_shape": 0.8018317818641663,
+    "lr": 6.859005121165148e-05,
+    "elapsed_sec": 1118.1285510063171
+  },
+  {
+    "step": 4000,
+    "train_shape": 0.8431285619735718,
+    "lr": 6.71783072328101e-05,
+    "elapsed_sec": 1146.3597824573517
+  },
+  {
+    "step": 4000,
+    "val_shape": 0.8495839834213257
+  },
+  {
+    "step": 4100,
+    "train_shape": 0.7751585841178894,
+    "lr": 6.575207753686356e-05,
+    "elapsed_sec": 1175.5649752616882
+  },
+  {
+    "step": 4200,
+    "train_shape": 0.7778671383857727,
+    "lr": 6.431276964033111e-05,
+    "elapsed_sec": 1203.7970151901245
+  },
+  {
+    "step": 4300,
+    "train_shape": 0.785365104675293,
+    "lr": 6.286180396633663e-05,
+    "elapsed_sec": 1232.0446407794952
+  },
+  {
+    "step": 4400,
+    "train_shape": 0.845032811164856,
+    "lr": 6.140061244282311e-05,
+    "elapsed_sec": 1260.28538107872
+  },
+  {
+    "step": 4500,
+    "train_shape": 0.7689575552940369,
+    "lr": 5.9930637089412255e-05,
+    "elapsed_sec": 1288.5169897079468
+  },
+  {
+    "step": 4500,
+    "val_shape": 0.8456140160560608
+  },
+  {
+    "step": 4600,
+    "train_shape": 0.7452852725982666,
+    "lr": 5.845332859430572e-05,
+    "elapsed_sec": 1317.7465991973877
+  },
+  {
+    "step": 4700,
+    "train_shape": 0.7812906503677368,
+    "lr": 5.697014488263067e-05,
+    "elapsed_sec": 1346.089480638504
+  },
+  {
+    "step": 4800,
+    "train_shape": 0.7887287139892578,
+    "lr": 5.548254967764354e-05,
+    "elapsed_sec": 1374.5268805027008
+  },
+  {
+    "step": 4900,
+    "train_shape": 0.8496115207672119,
+    "lr": 5.3992011056212214e-05,
+    "elapsed_sec": 1402.7835450172424
+  },
+  {
+    "step": 5000,
+    "train_shape": 0.8087153434753418,
+    "lr": 5.250000000000107e-05,
+    "elapsed_sec": 1431.0209577083588
+  },
+  {
+    "step": 5000,
+    "val_shape": 0.8412057161331177
+  },
+  {
+    "step": 5100,
+    "train_shape": 0.8487837910652161,
+    "lr": 5.100798894378995e-05,
+    "elapsed_sec": 1462.11345744133
+  },
+  {
+    "step": 5200,
+    "train_shape": 0.8302334547042847,
+    "lr": 4.951745032235864e-05,
+    "elapsed_sec": 1490.3205099105835
+  },
+  {
+    "step": 5300,
+    "train_shape": 0.7922009229660034,
+    "lr": 4.8029855117371535e-05,
+    "elapsed_sec": 1518.5206182003021
+  },
+  {
+    "step": 5400,
+    "train_shape": 0.7913670539855957,
+    "lr": 4.654667140569645e-05,
+    "elapsed_sec": 1546.7638883590698
+  },
+  {
+    "step": 5500,
+    "train_shape": 0.7895956635475159,
+    "lr": 4.506936291058995e-05,
+    "elapsed_sec": 1574.981760263443
+  },
+  {
+    "step": 5500,
+    "val_shape": 0.8404234647750854
+  },
+  {
+    "step": 5600,
+    "train_shape": 0.7408543825149536,
+    "lr": 4.359938755717899e-05,
+    "elapsed_sec": 1604.1752305030823
+  },
+  {
+    "step": 5700,
+    "train_shape": 0.7522714734077454,
+    "lr": 4.213819603366514e-05,
+    "elapsed_sec": 1632.3973755836487
+  },
+  {
+    "step": 5800,
+    "train_shape": 0.8080376982688904,
+    "lr": 4.0687230359670296e-05,
+    "elapsed_sec": 1660.6248710155487
+  },
+  {
+    "step": 5900,
+    "train_shape": 0.7631515264511108,
+    "lr": 3.924792246313749e-05,
+    "elapsed_sec": 1688.95654296875
+  },
+  {
+    "step": 6000,
+    "train_shape": 0.761693000793457,
+    "lr": 3.7821692767190845e-05,
+    "elapsed_sec": 1717.1791653633118
+  },
+  {
+    "step": 6000,
+    "val_shape": 0.8373721837997437
+  },
+  {
+    "step": 6100,
+    "train_shape": 0.7622103691101074,
+    "lr": 3.640994878834947e-05,
+    "elapsed_sec": 1746.3542127609253
+  },
+  {
+    "step": 6200,
+    "train_shape": 0.768337607383728,
+    "lr": 3.501408374747862e-05,
+    "elapsed_sec": 1774.5794394016266
+  },
+  {
+    "step": 6300,
+    "train_shape": 0.7231592535972595,
+    "lr": 3.363547519484887e-05,
+    "elapsed_sec": 1802.7991178035736
+  },
+  {
+    "step": 6400,
+    "train_shape": 0.7120444178581238,
+    "lr": 3.227548365066013e-05,
+    "elapsed_sec": 1831.0241284370422
+  },
+  {
+    "step": 6500,
+    "train_shape": 0.7312938570976257,
+    "lr": 3.0935451262372706e-05,
+    "elapsed_sec": 1859.2460734844208
+  },
+  {
+    "step": 6500,
+    "val_shape": 0.8339642882347107
+  },
+  {
+    "step": 6600,
+    "train_shape": 0.763182520866394,
+    "lr": 2.9616700480169673e-05,
+    "elapsed_sec": 1888.3847861289978
+  },
+  {
+    "step": 6700,
+    "train_shape": 0.8041389584541321,
+    "lr": 2.8320532751858457e-05,
+    "elapsed_sec": 1916.6041235923767
+  },
+  {
+    "step": 6800,
+    "train_shape": 0.7822360396385193,
+    "lr": 2.7048227238498715e-05,
+    "elapsed_sec": 1944.8252398967743
+  },
+  {
+    "step": 6900,
+    "train_shape": 0.7559687495231628,
+    "lr": 2.5801039552024805e-05,
+    "elapsed_sec": 1973.0428183078766
+  },
+  {
+    "step": 7000,
+    "train_shape": 0.803432047367096,
+    "lr": 2.458020051610849e-05,
+    "elapsed_sec": 2001.3757619857788
+  },
+  {
+    "step": 7000,
+    "val_shape": 0.8327170610427856
+  },
+  {
+    "step": 7100,
+    "train_shape": 0.7607059478759766,
+    "lr": 2.3386914951484544e-05,
+    "elapsed_sec": 2030.5706520080566
+  },
+  {
+    "step": 7200,
+    "train_shape": 0.7893930673599243,
+    "lr": 2.222236048693809e-05,
+    "elapsed_sec": 2058.9274191856384
+  },
+  {
+    "step": 7300,
+    "train_shape": 0.8376253843307495,
+    "lr": 2.1087686397127328e-05,
+    "elapsed_sec": 2087.1605219841003
+  },
+  {
+    "step": 7400,
+    "train_shape": 0.8037107586860657,
+    "lr": 1.998401246838802e-05,
+    "elapsed_sec": 2115.555244445801
+  },
+  {
+    "step": 7500,
+    "train_shape": 0.7703807950019836,
+    "lr": 1.891242789363976e-05,
+    "elapsed_sec": 2143.7768216133118
+  },
+  {
+    "step": 7500,
+    "val_shape": 0.8324167728424072
+  },
+  {
+    "step": 7600,
+    "train_shape": 0.7770482301712036,
+    "lr": 1.787399019748373e-05,
+    "elapsed_sec": 2172.9288878440857
+  },
+  {
+    "step": 7700,
+    "train_shape": 0.8342878222465515,
+    "lr": 1.6869724192553968e-05,
+    "elapsed_sec": 2201.1405262947083
+  },
+  {
+    "step": 7800,
+    "train_shape": 0.7862416505813599,
+    "lr": 1.5900620968150808e-05,
+    "elapsed_sec": 2229.359471797943
+  },
+  {
+    "step": 7900,
+    "train_shape": 0.7752629518508911,
+    "lr": 1.496763691215551e-05,
+    "elapsed_sec": 2257.5899555683136
+  },
+  {
+    "step": 8000,
+    "train_shape": 0.7890541553497314,
+    "lr": 1.4071692767190806e-05,
+    "elapsed_sec": 2285.8047819137573
+  },
+  {
+    "step": 8000,
+    "val_shape": 0.8307053446769714
+  },
+  {
+    "step": 8100,
+    "train_shape": 0.8134850263595581,
+    "lr": 1.3213672721959122e-05,
+    "elapsed_sec": 2314.985892534256
+  },
+  {
+    "step": 8200,
+    "train_shape": 0.7168530225753784,
+    "lr": 1.2394423538655062e-05,
+    "elapsed_sec": 2343.2096168994904
+  },
+  {
+    "step": 8300,
+    "train_shape": 0.7898730039596558,
+    "lr": 1.1614753717313381e-05,
+    "elapsed_sec": 2371.4606697559357
+  },
+  {
+    "step": 8400,
+    "train_shape": 0.7653496265411377,
+    "lr": 1.0875432697917097e-05,
+    "elapsed_sec": 2399.671991109848
+  },
+  {
+    "step": 8500,
+    "train_shape": 0.7535018920898438,
+    "lr": 1.0177190101053074e-05,
+    "elapsed_sec": 2428.137418270111
+  },
+  {
+    "step": 8500,
+    "val_shape": 0.8296800851821899
+  },
+  {
+    "step": 8600,
+    "train_shape": 0.7750765681266785,
+    "lr": 9.520715007864551e-06,
+    "elapsed_sec": 2457.283500432968
+  },
+  {
+    "step": 8700,
+    "train_shape": 0.7837339639663696,
+    "lr": 8.906655280011289e-06,
+    "elapsed_sec": 2485.5041477680206
+  },
+  {
+    "step": 8800,
+    "train_shape": 0.8121090531349182,
+    "lr": 8.335616920308386e-06,
+    "elapsed_sec": 2513.7187955379486
+  },
+  {
+    "step": 8900,
+    "train_shape": 0.7563372850418091,
+    "lr": 7.808163474674577e-06,
+    "elapsed_sec": 2541.9335572719574
+  },
+  {
+    "step": 9000,
+    "train_shape": 0.7611607313156128,
+    "lr": 7.32481547598044e-06,
+    "elapsed_sec": 2570.1557500362396
+  },
+  {
+    "step": 9000,
+    "val_shape": 0.8286348581314087
+  },
+  {
+    "step": 9100,
+    "train_shape": 0.8008730411529541,
+    "lr": 6.886049930345396e-06,
+    "elapsed_sec": 2599.314120054245
+  },
+  {
+    "step": 9200,
+    "train_shape": 0.8060568571090698,
+    "lr": 6.492299846390179e-06,
+    "elapsed_sec": 2627.563369512558
+  },
+  {
+    "step": 9300,
+    "train_shape": 0.779072642326355,
+    "lr": 6.143953807909618e-06,
+    "elapsed_sec": 2655.8930566310883
+  },
+  {
+    "step": 9400,
+    "train_shape": 0.7388461828231812,
+    "lr": 5.841355590387376e-06,
+    "elapsed_sec": 2684.1053919792175
+  },
+  {
+    "step": 9500,
+    "train_shape": 0.794291079044342,
+    "lr": 5.5848038217310204e-06,
+    "elapsed_sec": 2712.3224008083344
+  },
+  {
+    "step": 9500,
+    "val_shape": 0.8284382820129395
+  },
+  {
+    "step": 9600,
+    "train_shape": 0.7644393444061279,
+    "lr": 5.374551687562344e-06,
+    "elapsed_sec": 2741.4729866981506
+  },
+  {
+    "step": 9700,
+    "train_shape": 0.7576324343681335,
+    "lr": 5.2108066813537186e-06,
+    "elapsed_sec": 2769.6907596588135
+  },
+  {
+    "step": 9800,
+    "train_shape": 0.7811599373817444,
+    "lr": 5.093730399657109e-06,
+    "elapsed_sec": 2797.906686067581
+  },
+  {
+    "step": 9900,
+    "train_shape": 0.7527590394020081,
+    "lr": 5.023438382627753e-06,
+    "elapsed_sec": 2826.213297843933
+  },
+  {
+    "step": 10000,
+    "train_shape": 0.8111909031867981,
+    "lr": 5e-06,
+    "elapsed_sec": 2854.4251403808594
+  },
+  {
+    "step": 10000,
+    "val_shape": 0.8280910849571228
+  }
+]
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..6269dabac112cdeb4b6514759f810f92fed22599
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:42.721948411Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:43.040119359Z","level":"INFO","msg":"stream: created new stream","id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.040191492Z","level":"INFO","msg":"handler: started","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.041928367Z","level":"INFO","msg":"stream: started","id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.04195729Z","level":"INFO","msg":"writer: started","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.041963542Z","level":"INFO","msg":"sender: started","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.32809079Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:24.566097831Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:24.571834409Z","level":"INFO","msg":"stream: closing","id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.571847733Z","level":"INFO","msg":"handler: closed","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.572982283Z","level":"INFO","msg":"sender: closed","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.572991139Z","level":"INFO","msg":"stream: closed","id":"3zi7cjs8"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/debug.log b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..b3ada1423e9cfe7d1e6f766be5131f2ff659066d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug.log
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug-internal.log
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz', 'image_view': 'rear', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.0, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-plain-rear-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,rear', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:42,712 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:42,719 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:42,721 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:42,735 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:43,376 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:43,836 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:23,366 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/3zi7cjs8
+2026-08-16 20:03:23,367 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:23,367 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:23,367 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:24,569 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/config.yaml b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..2447b7cdbbab0e432c3c03dbdd325f8b0d4e0993
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/config.yaml
@@ -0,0 +1,172 @@
+_wandb:
+    value:
+        cli_version: 0.22.3
+        e:
+            jl4d8brfqpi3oqxiym52khxxr8xmz166:
+                args:
+                    - --mode
+                    - train
+                    - --image-view
+                    - rear
+                    - --paired-latents
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz
+                    - --out-dir
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear
+                    - --steps
+                    - "10000"
+                    - --batch-size
+                    - "128"
+                    - --lr
+                    - "1e-4"
+                    - --weight-decay
+                    - "0.01"
+                    - --seed
+                    - "20260526"
+                    - --pos-embed
+                    - 2d
+                    - --cond-dropout
+                    - "0.0"
+                    - --resume
+                    - --wandb
+                    - --wandb-run-name
+                    - drivaer3v-inputcond-plain-rear-0816c
+                    - --wandb-tags
+                    - drivaer,qwen3v,input-conditional,channel-concat,rear
+                codePath: code/ICLR/customized-design/input_cond/train_cond_flow.py
+                codePathLocal: train_cond_flow.py
+                cpu_count: 32
+                cpu_count_logical: 64
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "7679361679360"
+                        used: "2800196116480"
+                email: 1193414549@qq.com
+                executable: /mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python
+                git:
+                    commit: 81d95989e1279e432c4145933bf9ce02c3f70f77
+                gpu: NVIDIA A100-SXM4-40GB
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "42949672960"
+                      name: NVIDIA A100-SXM4-40GB
+                      uuid: GPU-6b0e8c05-a91f-b234-b772-05ea54f42725
+                host: inputcond-plain-rear-a100-40g-0816c-0-0
+                memory:
+                    total: "540741947392"
+                os: Linux-6.8.0-124-generic-x86_64-with-glibc2.35
+                program: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py
+                python: CPython 3.10.18
+                root: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear/wandb
+                startedAt: "2026-08-16T19:15:42.469535Z"
+                writerId: jl4d8brfqpi3oqxiym52khxxr8xmz166
+        m: []
+        python_version: 3.10.18
+        t:
+            "1":
+                - 1
+                - 105
+            "2":
+                - 1
+                - 105
+            "3":
+                - 2
+                - 13
+                - 15
+                - 16
+                - 61
+                - 62
+            "4": 3.10.18
+            "5": 0.22.3
+            "12": 0.22.3
+            "13": linux-x86_64
+attention_backend:
+    value: sdpa
+batch_size:
+    value: 128
+cond_dropout:
+    value: 0
+decode_mesh:
+    value: false
+depth:
+    value: 12
+dropout:
+    value: 0
+eval_every:
+    value: 500
+eval_noise_seed:
+    value: 20260712
+gradient_checkpointing:
+    value: false
+guidance_scale:
+    value: 1
+head_dim:
+    value: 64
+hidden_size:
+    value: 512
+image_scale:
+    value: 0
+image_view:
+    value: rear
+log_every:
+    value: 100
+lr:
+    value: 0.0001
+max_val:
+    value: 128
+mlp_ratio:
+    value: 4
+mode:
+    value: train
+num_heads:
+    value: 8
+num_workers:
+    value: 2
+out_dir:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear
+paired_latents:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz
+pos_embed:
+    value: 2d
+resume:
+    value: true
+sample_checkpoint:
+    value: model_best.pt
+sample_count:
+    value: 8
+sample_seed:
+    value: 20260712
+sample_steps:
+    value: 50
+save_every:
+    value: 5000
+seed:
+    value: 20260526
+shape_max_batch:
+    value: 32768
+shape_resolution:
+    value: 512
+shape_scale:
+    value: 0
+steps:
+    value: 10000
+val_fraction:
+    value: 0.05
+val_overlap_train:
+    value: true
+wandb:
+    value: true
+wandb_entity:
+    value: kanydao-epfl
+wandb_mode:
+    value: online
+wandb_project:
+    value: drivaer-qwen3v-input-cond
+wandb_run_name:
+    value: drivaer3v-inputcond-plain-rear-0816c
+wandb_tags:
+    value: drivaer,qwen3v,input-conditional,channel-concat,rear
+weight_decay:
+    value: 0.01
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/output.log b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..a62788b790048433ccb2e7b549c1bb1505358573
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/output.log
@@ -0,0 +1,123 @@
+[wandb] project=drivaer-qwen3v-input-cond run=drivaer3v-inputcond-plain-rear-0816c mode=online
+[train] step=000001 loss=2.009530 lr=1.00e-04
+[val] step=000001 shape=2.006045
+[train] step=000100 loss=1.269827 lr=1.00e-04
+[train] step=000200 loss=1.032565 lr=9.99e-05
+[train] step=000300 loss=0.929615 lr=9.98e-05
+[train] step=000400 loss=0.881994 lr=9.96e-05
+[train] step=000500 loss=0.871593 lr=9.94e-05
+[val] step=000500 shape=0.930525
+[train] step=000600 loss=0.868780 lr=9.92e-05
+[train] step=000700 loss=0.807989 lr=9.89e-05
+[train] step=000800 loss=0.886874 lr=9.85e-05
+[train] step=000900 loss=0.843331 lr=9.81e-05
+[train] step=001000 loss=0.845006 lr=9.77e-05
+[val] step=001000 shape=0.897811
+[train] step=001100 loss=0.795780 lr=9.72e-05
+[train] step=001200 loss=0.876379 lr=9.67e-05
+[train] step=001300 loss=0.817046 lr=9.61e-05
+[train] step=001400 loss=0.874860 lr=9.55e-05
+[train] step=001500 loss=0.816332 lr=9.48e-05
+[val] step=001500 shape=0.885287
+[train] step=001600 loss=0.870507 lr=9.41e-05
+[train] step=001700 loss=0.783181 lr=9.34e-05
+[train] step=001800 loss=0.841894 lr=9.26e-05
+[train] step=001900 loss=0.844718 lr=9.18e-05
+[train] step=002000 loss=0.855920 lr=9.09e-05
+[val] step=002000 shape=0.872898
+[train] step=002100 loss=0.823513 lr=9.00e-05
+[train] step=002200 loss=0.817970 lr=8.91e-05
+[train] step=002300 loss=0.878616 lr=8.81e-05
+[train] step=002400 loss=0.786508 lr=8.71e-05
+[train] step=002500 loss=0.862334 lr=8.61e-05
+[val] step=002500 shape=0.864706
+[train] step=002600 loss=0.814341 lr=8.50e-05
+[train] step=002700 loss=0.814216 lr=8.39e-05
+[train] step=002800 loss=0.767981 lr=8.28e-05
+[train] step=002900 loss=0.826681 lr=8.16e-05
+[train] step=003000 loss=0.805259 lr=8.04e-05
+[val] step=003000 shape=0.859960
+[train] step=003100 loss=0.792970 lr=7.92e-05
+[train] step=003200 loss=0.810101 lr=7.80e-05
+[train] step=003300 loss=0.810508 lr=7.67e-05
+[train] step=003400 loss=0.804921 lr=7.54e-05
+[train] step=003500 loss=0.856046 lr=7.41e-05
+[val] step=003500 shape=0.854873
+[train] step=003600 loss=0.794709 lr=7.27e-05
+[train] step=003700 loss=0.793957 lr=7.14e-05
+[train] step=003800 loss=0.801822 lr=7.00e-05
+[train] step=003900 loss=0.801832 lr=6.86e-05
+[train] step=004000 loss=0.843129 lr=6.72e-05
+[val] step=004000 shape=0.849584
+[train] step=004100 loss=0.775159 lr=6.58e-05
+[train] step=004200 loss=0.777867 lr=6.43e-05
+[train] step=004300 loss=0.785365 lr=6.29e-05
+[train] step=004400 loss=0.845033 lr=6.14e-05
+[train] step=004500 loss=0.768958 lr=5.99e-05
+[val] step=004500 shape=0.845614
+[train] step=004600 loss=0.745285 lr=5.85e-05
+[train] step=004700 loss=0.781291 lr=5.70e-05
+[train] step=004800 loss=0.788729 lr=5.55e-05
+[train] step=004900 loss=0.849612 lr=5.40e-05
+[train] step=005000 loss=0.808715 lr=5.25e-05
+[val] step=005000 shape=0.841206
+[train] step=005100 loss=0.848784 lr=5.10e-05
+[train] step=005200 loss=0.830233 lr=4.95e-05
+[train] step=005300 loss=0.792201 lr=4.80e-05
+[train] step=005400 loss=0.791367 lr=4.65e-05
+[train] step=005500 loss=0.789596 lr=4.51e-05
+[val] step=005500 shape=0.840423
+[train] step=005600 loss=0.740854 lr=4.36e-05
+[train] step=005700 loss=0.752271 lr=4.21e-05
+[train] step=005800 loss=0.808038 lr=4.07e-05
+[train] step=005900 loss=0.763152 lr=3.92e-05
+[train] step=006000 loss=0.761693 lr=3.78e-05
+[val] step=006000 shape=0.837372
+[train] step=006100 loss=0.762210 lr=3.64e-05
+[train] step=006200 loss=0.768338 lr=3.50e-05
+[train] step=006300 loss=0.723159 lr=3.36e-05
+[train] step=006400 loss=0.712044 lr=3.23e-05
+[train] step=006500 loss=0.731294 lr=3.09e-05
+[val] step=006500 shape=0.833964
+[train] step=006600 loss=0.763183 lr=2.96e-05
+[train] step=006700 loss=0.804139 lr=2.83e-05
+[train] step=006800 loss=0.782236 lr=2.70e-05
+[train] step=006900 loss=0.755969 lr=2.58e-05
+[train] step=007000 loss=0.803432 lr=2.46e-05
+[val] step=007000 shape=0.832717
+[train] step=007100 loss=0.760706 lr=2.34e-05
+[train] step=007200 loss=0.789393 lr=2.22e-05
+[train] step=007300 loss=0.837625 lr=2.11e-05
+[train] step=007400 loss=0.803711 lr=2.00e-05
+[train] step=007500 loss=0.770381 lr=1.89e-05
+[val] step=007500 shape=0.832417
+[train] step=007600 loss=0.777048 lr=1.79e-05
+[train] step=007700 loss=0.834288 lr=1.69e-05
+[train] step=007800 loss=0.786242 lr=1.59e-05
+[train] step=007900 loss=0.775263 lr=1.50e-05
+[train] step=008000 loss=0.789054 lr=1.41e-05
+[val] step=008000 shape=0.830705
+[train] step=008100 loss=0.813485 lr=1.32e-05
+[train] step=008200 loss=0.716853 lr=1.24e-05
+[train] step=008300 loss=0.789873 lr=1.16e-05
+[train] step=008400 loss=0.765350 lr=1.09e-05
+[train] step=008500 loss=0.753502 lr=1.02e-05
+[val] step=008500 shape=0.829680
+[train] step=008600 loss=0.775077 lr=9.52e-06
+[train] step=008700 loss=0.783734 lr=8.91e-06
+[train] step=008800 loss=0.812109 lr=8.34e-06
+[train] step=008900 loss=0.756337 lr=7.81e-06
+[train] step=009000 loss=0.761161 lr=7.32e-06
+[val] step=009000 shape=0.828635
+[train] step=009100 loss=0.800873 lr=6.89e-06
+[train] step=009200 loss=0.806057 lr=6.49e-06
+[train] step=009300 loss=0.779073 lr=6.14e-06
+[train] step=009400 loss=0.738846 lr=5.84e-06
+[train] step=009500 loss=0.794291 lr=5.58e-06
+[val] step=009500 shape=0.828438
+[train] step=009600 loss=0.764439 lr=5.37e-06
+[train] step=009700 loss=0.757632 lr=5.21e-06
+[train] step=009800 loss=0.781160 lr=5.09e-06
+[train] step=009900 loss=0.752759 lr=5.02e-06
+[train] step=010000 loss=0.811191 lr=5.00e-06
+[val] step=010000 shape=0.828091
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/requirements.txt b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..2e10c9ab6bc99041706780299f9e3e5035f34676
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/requirements.txt
@@ -0,0 +1,302 @@
+shellingham==1.5.4
+importlib_metadata==8.7.0
+websocket-client==1.8.0
+rfc3986-validator==0.1.1
+aiohappyeyeballs==2.6.1
+tensorboardX==2.6.2.2
+matplotlib==3.8.4
+zstandard==0.23.0
+gradio_client==1.4.2
+nbclient==0.10.2
+pyaml==25.7.0
+argparse==1.4.0
+h11==0.16.0
+scooby==0.10.1
+pexpect==4.9.0
+jedi==0.19.2
+scikit-optimize==0.10.2
+omegaconf==2.3.0
+tomlkit==0.12.0
+yarl==1.20.1
+point-cloud-utils==0.34.0
+cffi==1.17.1
+asttokens==3.0.0
+webencodings==0.5.1
+imageio==2.34.1
+attrs==25.3.0
+hydra-core==1.3.2
+datasets==2.19.0
+pyquaternion==0.9.9
+trame-common==1.0.0
+wslink==2.3.4
+Pygments==2.19.2
+pydantic==2.11.7
+regex==2024.11.6
+timm==0.9.16
+ipykernel==6.29.5
+prompt_toolkit==3.0.51
+torchdiffeq==0.2.5
+pybind11==2.13.6
+debugpy==1.8.14
+pytz==2025.2
+transformers==4.40.1
+jupyterlab_pygments==0.3.0
+dash==3.1.1
+psutil==5.9.8
+tqdm==4.66.2
+multidict==6.6.3
+setproctitle==1.3.6
+nbformat==5.10.4
+prometheus_client==0.22.1
+jupyter_core==5.8.1
+gradio==5.5.0
+multiprocess==0.70.16
+fastapi==0.115.14
+simpervisor==1.0.0
+comm==0.2.2
+widgetsnbextension==4.0.14
+webcolors==24.11.1
+sentry-sdk==2.32.0
+tornado==6.5.1
+defusedxml==0.7.1
+Markdown==3.8.2
+numba==0.61.2
+pyarrow-hotfix==0.7
+trame-server==3.5.0
+nbconvert==7.16.6
+websockets==12.0
+semantic-version==2.10.0
+bpy==4.0.0
+smmap==5.0.2
+grpcio==1.73.1
+referencing==0.36.2
+annotated-types==0.7.0
+stack-data==0.6.3
+vtk==9.4.2
+safehttpx==0.1.6
+GitPython==3.1.44
+ipywidgets==8.1.7
+platformdirs==4.3.8
+tzdata==2025.2
+typer==0.16.0
+click==8.2.1
+networkx==3.3
+fsspec==2024.3.1
+plotly==6.2.0
+isoduration==20.11.0
+joblib==1.4.0
+aiosignal==1.4.0
+jsonschema-specifications==2025.4.1
+six==1.17.0
+pyzmq==27.0.0
+lazy_loader==0.4
+typeguard==2.13.3
+markdown-it-py==3.0.0
+httpx==0.28.1
+fqdn==1.5.1
+threadpoolctl==3.6.0
+orjson==3.10.18
+docker-pycreds==0.4.0
+jsonschema==4.24.0
+tensorboard-data-server==0.7.2
+nvidia-cufft-cu12==11.0.2.54
+aiofiles==23.2.1
+rpds-py==0.26.0
+Cython==3.1.2
+python-multipart==0.0.12
+nvidia-cuda-runtime-cu12==12.1.105
+huggingface-hub==0.26.2
+frozenlist==1.7.0
+onnxruntime==1.22.0
+cycler==0.12.1
+tifffile==2025.5.10
+safetensors==0.4.3
+nest-asyncio==1.6.0
+parso==0.8.4
+PyMCubes==0.1.4
+nvidia-nvtx-cu12==12.1.105
+pydub==0.25.1
+packaging==25.0
+ipython==8.37.0
+uvicorn==0.35.0
+pyarrow==20.0.0
+rich==14.0.0
+executing==2.2.0
+jupyter_server_terminals==0.5.3
+typing_extensions==4.12.2
+typing_extensions==4.15.0
+tensorboard==2.16.2
+argon2-cffi-bindings==21.2.0
+wcwidth==0.2.13
+typing-inspection==0.4.1
+pure_eval==0.2.3
+trimesh==4.3.2
+nvidia-cudnn-cu12==9.1.0.70
+humanfriendly==10.0
+Werkzeug==3.1.3
+decorator==5.2.1
+nvidia-cuda-nvrtc-cu12==12.1.105
+aiohttp==3.12.13
+uri-template==1.3.0
+anyio==4.9.0
+fastjsonschema==2.21.1
+nvidia-curand-cu12==10.3.2.106
+zipp==3.23.0
+rtree==1.4.1
+kiwisolver==1.4.8
+Flask==3.1.1
+nvidia-cuda-cupti-cu12==12.1.105
+pysdf==0.1.9
+sniffio==1.3.1
+itsdangerous==2.2.0
+rfc3339-validator==0.1.4
+trame==3.10.2
+open3d==0.18.0
+scikit-learn==1.7.0
+mistune==3.1.3
+diffusers==0.31.0
+absl-py==2.3.1
+cubvh==0.1.0
+trame-client==3.9.1
+einops==0.8.0
+plyfile==1.0.3
+bleach==6.2.0
+ptyprocess==0.7.0
+pycparser==2.22
+llvmlite==0.44.0
+addict==2.4.0
+diso==0.1.4
+Send2Trash==1.8.3
+more-itertools==10.7.0
+scikit-image==0.23.2
+propcache==0.3.2
+pandas==2.2.2
+flatbuffers==25.2.10
+gitdb==4.0.12
+jaxtyping==0.2.28
+soupsieve==2.7
+mdurl==0.1.2
+protobuf==5.29.5
+fonttools==4.58.5
+coloredlogs==15.0.1
+spaces==0.28.3
+jupyter_server==2.16.0
+trame-vtk==2.9.0
+terminado==0.18.1
+overrides==7.7.0
+pandocfilters==1.5.1
+nvidia-cusolver-cu12==11.4.5.107
+trame-vuetify==3.0.1
+async-timeout==5.0.1
+blinker==1.9.0
+exceptiongroup==1.3.0
+xxhash==3.5.0
+python-json-logger==3.3.0
+pyvista==0.45.2
+python-dateutil==2.9.0.post0
+ConfigArgParse==1.7.1
+PyMatting==1.1.14
+fpsample==0.3.3
+traitlets==5.14.3
+msgpack==1.1.1
+argon2-cffi==25.1.0
+tinycss2==1.4.0
+retrying==1.4.0
+jsonpointer==3.0.0
+matplotlib-inline==0.1.7
+jupyter_client==8.6.3
+contourpy==1.3.2
+lightning-utilities==0.11.2
+pydantic_core==2.33.2
+httpcore==1.0.9
+narwhals==1.46.0
+nvidia-cublas-cu12==12.1.3.1
+jupyterlab_widgets==3.0.15
+antlr4-python3-runtime==4.9.3
+pyparsing==3.1.2
+rembg==2.0.59
+ffmpy==0.6.0
+tokenizers==0.19.1
+beautifulsoup4==4.13.4
+pooch==1.8.2
+ninja==1.11.1.4
+jupyter_server_proxy==4.4.0
+ruff==0.12.2
+types-python-dateutil==2.9.0.20250708
+dill==0.3.8
+pytorch-lightning==2.2.4
+jupyter-events==0.12.0
+starlette==0.46.2
+arrow==1.3.0
+setuptools==78.1.1
+wheel==0.45.1
+pip==25.1
+mkl-service==2.5.2
+mpi4py==4.0.3
+h5py==3.14.0
+mkl_fft==2.1.1
+mkl_random==1.3.0
+mcubes==0.1.6
+wandb==0.22.3
+certifi==2026.1.4
+mpmath==1.3.0
+triton==3.1.0
+sympy==1.13.1
+nvidia-nccl-cu12==2.21.5
+nvidia-nvjitlink-cu12==12.9.86
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nvtx-cu11==11.8.86
+nvidia-nccl-cu11==2.21.5
+nvidia-cusparse-cu11==11.7.5.86
+nvidia-curand-cu11==10.3.0.86
+nvidia-cufft-cu11==10.9.0.58
+nvidia-cuda-runtime-cu11==11.8.89
+nvidia-cuda-nvrtc-cu11==11.8.89
+nvidia-cuda-cupti-cu11==11.8.87
+nvidia-cublas-cu11==11.11.3.6
+nvidia-cusolver-cu11==11.4.1.48
+nvidia-cudnn-cu11==9.1.0.70
+torch==2.5.1+cu118
+torchvision==0.20.1+cu118
+torchaudio==2.5.1+cu118
+PyYAML==6.0.3
+pillow==12.1.0
+torchmetrics==1.8.2
+accelerate==1.12.0
+urllib3==2.6.3
+idna==3.11
+charset-normalizer==3.4.4
+requests==2.32.5
+filelock==3.20.3
+torch_cluster==1.6.3+pt25cu118
+numpy==1.26.4
+opencv-python-headless==4.9.0.80
+PyOpenGL-accelerate==3.1.10
+PyOpenGL==3.1.0
+pyglet==2.1.14
+freetype-py==2.5.1
+pyrender==0.1.45
+scipy==1.15.3
+memory-profiler==0.61.0
+pyamg==5.3.0
+meshio==5.3.5
+torch-fem==0.6.2
+torch-geometric==2.7.0
+MarkupSafe==3.0.3
+Jinja2==3.1.6
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/wandb-metadata.json b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..f874df5d261d55e6cae09586b1eb5cb90883738f
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/wandb-metadata.json
@@ -0,0 +1,69 @@
+{
+  "os":  "Linux-6.8.0-124-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.10.18",
+  "startedAt":  "2026-08-16T19:15:42.469535Z",
+  "args":  [
+    "--mode",
+    "train",
+    "--image-view",
+    "rear",
+    "--paired-latents",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz",
+    "--out-dir",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear",
+    "--steps",
+    "10000",
+    "--batch-size",
+    "128",
+    "--lr",
+    "1e-4",
+    "--weight-decay",
+    "0.01",
+    "--seed",
+    "20260526",
+    "--pos-embed",
+    "2d",
+    "--cond-dropout",
+    "0.0",
+    "--resume",
+    "--wandb",
+    "--wandb-run-name",
+    "drivaer3v-inputcond-plain-rear-0816c",
+    "--wandb-tags",
+    "drivaer,qwen3v,input-conditional,channel-concat,rear"
+  ],
+  "program":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePath":  "code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePathLocal":  "train_cond_flow.py",
+  "git":  {
+    "commit":  "81d95989e1279e432c4145933bf9ce02c3f70f77"
+  },
+  "email":  "1193414549@qq.com",
+  "root":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear/wandb",
+  "host":  "inputcond-plain-rear-a100-40g-0816c-0-0",
+  "executable":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python",
+  "cpu_count":  32,
+  "cpu_count_logical":  64,
+  "gpu":  "NVIDIA A100-SXM4-40GB",
+  "gpu_count":  1,
+  "disk":  {
+    "/":  {
+      "total":  "7679361679360",
+      "used":  "2800196116480"
+    }
+  },
+  "memory":  {
+    "total":  "540741947392"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A100-SXM4-40GB",
+      "memoryTotal":  "42949672960",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-6b0e8c05-a91f-b234-b772-05ea54f42725"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "jl4d8brfqpi3oqxiym52khxxr8xmz166"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/wandb-summary.json b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..5f5cea3f3a85a24697738b64a4597d4f82849e41
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/files/wandb-summary.json
@@ -0,0 +1 @@
+{"val/shape":0.8280910849571228,"_timestamp":1.7869105984416685e+09,"image_view":"rear","_runtime":2859.990426401,"train/shape":0.8111909031867981,"time/elapsed_sec":2854.4251403808594,"steps":10000,"shape_scale":1.7132293052897125,"train/lr":5e-06,"val_samples":128,"_wandb":{"runtime":2859},"image_scale":1.6045079043256218,"_step":10000,"train_samples":3051,"best_val_shape":0.8280910849571228,"conditioning":"sketch_latent_channel_concat"}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..6269dabac112cdeb4b6514759f810f92fed22599
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:42.721948411Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:43.040119359Z","level":"INFO","msg":"stream: created new stream","id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.040191492Z","level":"INFO","msg":"handler: started","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.041928367Z","level":"INFO","msg":"stream: started","id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.04195729Z","level":"INFO","msg":"writer: started","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T19:15:43.041963542Z","level":"INFO","msg":"sender: started","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.32809079Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:24.566097831Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:24.571834409Z","level":"INFO","msg":"stream: closing","id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.571847733Z","level":"INFO","msg":"handler: closed","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.572982283Z","level":"INFO","msg":"sender: closed","stream_id":"3zi7cjs8"}
+{"time":"2026-08-16T20:03:24.572991139Z","level":"INFO","msg":"stream: closed","id":"3zi7cjs8"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug.log b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..b3ada1423e9cfe7d1e6f766be5131f2ff659066d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:42,499 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug.log
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/logs/debug-internal.log
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/rear', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/rear/paired_latents_train.npz', 'image_view': 'rear', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.0, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-plain-rear-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,rear', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:42,500 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:42,712 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:42,719 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:42,721 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:42,735 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:43,376 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:43,836 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:23,366 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/3zi7cjs8
+2026-08-16 20:03:23,367 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:23,367 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:23,367 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:24,569 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/run-3zi7cjs8.wandb b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/run-3zi7cjs8.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..b828d698aadbb2aab44f9eae74ca6cdaacf158cb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/rear/wandb/wandb/run-20260816_191542-3zi7cjs8/run-3zi7cjs8.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dd14ebe335b400307b794db103b0d8ca16918663508b77de29f871ea3711b53
+size 236444
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/code_provenance.json b/anchorgen_baselines_20260923/input_cond/runs/plain/side/code_provenance.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d65c77fa6bb7025a1484f4ad82c6b647f7a2315
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/code_provenance.json
@@ -0,0 +1,7 @@
+{
+  "file": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "commit": "81d95989e1279e432c4145933bf9ce02c3f70f77",
+  "branch": "flow-field",
+  "train_cond_flow_sha256": "095053598f77923a8c8fbdd63bb77280bd641d5f8ca004bc100e63b82bae7616",
+  "seiler_cond_backbone_sha256": "908f1f3b3756ac054e640b4e6e9fe7a40e2130adcf7ba0ce09432f74783dd97c"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_best.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_best.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0720ad2e8a739ffcdba9013405ec9b9333dbd14e
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_best.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aadccaa88bf8cbbf256506d2913ae72ea50b630fe562db8cf0e4d4f62806dbab
+size 329404702
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_last.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_last.pt
new file mode 100644
index 0000000000000000000000000000000000000000..13cbc01adbb789d993f4ed8cdb1c19d073f691a6
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_last.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58b3b27f687f171bcc8c5b99e678579464435bfd20c20ce4df23eabbdc052f40
+size 988212490
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_step_005000.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_step_005000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..03413655b3a3b1718449f8849c82078785585808
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_step_005000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d55150d7fcfec39dca8af1b4ba242b3254d43c18ed7f5d070dae6c4a23552bf2
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_step_010000.pt b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_step_010000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fca264e1180c6850d1e194f8ac3d4767cfe77e2e
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/model_step_010000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:921a0699d0ea9123af1f13fe026d5da7a204a773e3cbb11cd86ae3964189fbbb
+size 988220540
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/run_config.json b/anchorgen_baselines_20260923/input_cond/runs/plain/side/run_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..74dffa4ae592fa113f04bcbbb9228fd08a03781d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/run_config.json
@@ -0,0 +1,46 @@
+{
+  "mode": "train",
+  "out_dir": "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side",
+  "paired_latents": "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz",
+  "image_view": "side",
+  "seed": 20260526,
+  "val_fraction": 0.05,
+  "max_val": 128,
+  "val_overlap_train": true,
+  "shape_scale": 0.0,
+  "image_scale": 0.0,
+  "hidden_size": 512,
+  "depth": 12,
+  "num_heads": 8,
+  "head_dim": 64,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "pos_embed": "2d",
+  "attention_backend": "sdpa",
+  "gradient_checkpointing": false,
+  "cond_dropout": 0.0,
+  "batch_size": 128,
+  "num_workers": 2,
+  "steps": 10000,
+  "lr": 0.0001,
+  "weight_decay": 0.01,
+  "log_every": 100,
+  "eval_every": 500,
+  "eval_noise_seed": 20260712,
+  "save_every": 5000,
+  "resume": true,
+  "sample_count": 8,
+  "sample_steps": 50,
+  "sample_seed": 20260712,
+  "sample_checkpoint": "model_best.pt",
+  "guidance_scale": 1.0,
+  "decode_mesh": false,
+  "shape_resolution": 512,
+  "shape_max_batch": 32768,
+  "wandb": true,
+  "wandb_project": "drivaer-qwen3v-input-cond",
+  "wandb_entity": "kanydao-epfl",
+  "wandb_run_name": "drivaer3v-inputcond-plain-side-0816c",
+  "wandb_tags": "drivaer,qwen3v,input-conditional,channel-concat,side",
+  "wandb_mode": "online"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/summary.json b/anchorgen_baselines_20260923/input_cond/runs/plain/side/summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..67e01799af7bb0510751c6be0e02f72919b4e3c5
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/summary.json
@@ -0,0 +1,10 @@
+{
+  "image_view": "side",
+  "best_val_shape": 0.8226728439331055,
+  "steps": 10000,
+  "train_samples": 3051,
+  "val_samples": 128,
+  "shape_scale": 1.7132293052897125,
+  "image_scale": 1.4969936040640752,
+  "conditioning": "sketch_latent_channel_concat"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/train_history.json b/anchorgen_baselines_20260923/input_cond/runs/plain/side/train_history.json
new file mode 100644
index 0000000000000000000000000000000000000000..2346d4f200287765770a4c4f8bd5b3fd324da031
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/train_history.json
@@ -0,0 +1,692 @@
+[
+  {
+    "step": 1,
+    "train_shape": 2.0095300674438477,
+    "lr": 9.999999765596898e-05,
+    "elapsed_sec": 0.936082124710083
+  },
+  {
+    "step": 1,
+    "val_shape": 2.006117820739746
+  },
+  {
+    "step": 100,
+    "train_shape": 1.2791863679885864,
+    "lr": 9.997656161737219e-05,
+    "elapsed_sec": 37.12446141242981
+  },
+  {
+    "step": 200,
+    "train_shape": 1.0408809185028076,
+    "lr": 9.990626960034288e-05,
+    "elapsed_sec": 65.54530620574951
+  },
+  {
+    "step": 300,
+    "train_shape": 0.9307267665863037,
+    "lr": 9.978919331864633e-05,
+    "elapsed_sec": 94.04075717926025
+  },
+  {
+    "step": 400,
+    "train_shape": 0.8859299421310425,
+    "lr": 9.962544831243784e-05,
+    "elapsed_sec": 122.20862770080566
+  },
+  {
+    "step": 500,
+    "train_shape": 0.8706417083740234,
+    "lr": 9.94151961782692e-05,
+    "elapsed_sec": 150.41223788261414
+  },
+  {
+    "step": 500,
+    "val_shape": 0.9312366247177124
+  },
+  {
+    "step": 600,
+    "train_shape": 0.8679609894752502,
+    "lr": 9.915864440961283e-05,
+    "elapsed_sec": 179.6932246685028
+  },
+  {
+    "step": 700,
+    "train_shape": 0.8059194087982178,
+    "lr": 9.885604619209063e-05,
+    "elapsed_sec": 207.90899848937988
+  },
+  {
+    "step": 800,
+    "train_shape": 0.8870972394943237,
+    "lr": 9.85077001536101e-05,
+    "elapsed_sec": 236.34254956245422
+  },
+  {
+    "step": 900,
+    "train_shape": 0.841090202331543,
+    "lr": 9.811395006965495e-05,
+    "elapsed_sec": 264.59082078933716
+  },
+  {
+    "step": 1000,
+    "train_shape": 0.8450446128845215,
+    "lr": 9.767518452401991e-05,
+    "elapsed_sec": 293.05759048461914
+  },
+  {
+    "step": 1000,
+    "val_shape": 0.8987793922424316
+  },
+  {
+    "step": 1100,
+    "train_shape": 0.7974262237548828,
+    "lr": 9.71918365253258e-05,
+    "elapsed_sec": 322.1970007419586
+  },
+  {
+    "step": 1200,
+    "train_shape": 0.8797049522399902,
+    "lr": 9.666438307969207e-05,
+    "elapsed_sec": 350.4005882740021
+  },
+  {
+    "step": 1300,
+    "train_shape": 0.8169320225715637,
+    "lr": 9.60933447199893e-05,
+    "elapsed_sec": 378.65963315963745
+  },
+  {
+    "step": 1400,
+    "train_shape": 0.8758187890052795,
+    "lr": 9.547928499213621e-05,
+    "elapsed_sec": 406.8931975364685
+  },
+  {
+    "step": 1500,
+    "train_shape": 0.8163073062896729,
+    "lr": 9.482280989894765e-05,
+    "elapsed_sec": 435.12877130508423
+  },
+  {
+    "step": 1500,
+    "val_shape": 0.8851141929626465
+  },
+  {
+    "step": 1600,
+    "train_shape": 0.8725255727767944,
+    "lr": 9.412456730208378e-05,
+    "elapsed_sec": 464.2909164428711
+  },
+  {
+    "step": 1700,
+    "train_shape": 0.7835960984230042,
+    "lr": 9.338524628268747e-05,
+    "elapsed_sec": 492.5186266899109
+  },
+  {
+    "step": 1800,
+    "train_shape": 0.8417016267776489,
+    "lr": 9.260557646134585e-05,
+    "elapsed_sec": 520.7459514141083
+  },
+  {
+    "step": 1900,
+    "train_shape": 0.8423107862472534,
+    "lr": 9.178632727804172e-05,
+    "elapsed_sec": 548.9857332706451
+  },
+  {
+    "step": 2000,
+    "train_shape": 0.8594784736633301,
+    "lr": 9.092830723281015e-05,
+    "elapsed_sec": 577.18394947052
+  },
+  {
+    "step": 2000,
+    "val_shape": 0.8726917505264282
+  },
+  {
+    "step": 2100,
+    "train_shape": 0.8251680135726929,
+    "lr": 9.003236308784539e-05,
+    "elapsed_sec": 606.5172808170319
+  },
+  {
+    "step": 2200,
+    "train_shape": 0.817196249961853,
+    "lr": 8.90993790318501e-05,
+    "elapsed_sec": 634.7245216369629
+  },
+  {
+    "step": 2300,
+    "train_shape": 0.8782278895378113,
+    "lr": 8.813027580744691e-05,
+    "elapsed_sec": 663.0194749832153
+  },
+  {
+    "step": 2400,
+    "train_shape": 0.784439206123352,
+    "lr": 8.712600980251722e-05,
+    "elapsed_sec": 691.3315060138702
+  },
+  {
+    "step": 2500,
+    "train_shape": 0.8639475107192993,
+    "lr": 8.60875721063612e-05,
+    "elapsed_sec": 719.6213495731354
+  },
+  {
+    "step": 2500,
+    "val_shape": 0.8647449016571045
+  },
+  {
+    "step": 2600,
+    "train_shape": 0.8142738342285156,
+    "lr": 8.501598753161292e-05,
+    "elapsed_sec": 748.7612233161926
+  },
+  {
+    "step": 2700,
+    "train_shape": 0.8120519518852234,
+    "lr": 8.391231360287367e-05,
+    "elapsed_sec": 776.9595711231232
+  },
+  {
+    "step": 2800,
+    "train_shape": 0.7666445970535278,
+    "lr": 8.277763951306297e-05,
+    "elapsed_sec": 805.1747786998749
+  },
+  {
+    "step": 2900,
+    "train_shape": 0.823058009147644,
+    "lr": 8.161308504851662e-05,
+    "elapsed_sec": 833.3803806304932
+  },
+  {
+    "step": 3000,
+    "train_shape": 0.8043828010559082,
+    "lr": 8.04197994838928e-05,
+    "elapsed_sec": 861.5887203216553
+  },
+  {
+    "step": 3000,
+    "val_shape": 0.8559789061546326
+  },
+  {
+    "step": 3100,
+    "train_shape": 0.7878702282905579,
+    "lr": 7.919896044797652e-05,
+    "elapsed_sec": 890.8248317241669
+  },
+  {
+    "step": 3200,
+    "train_shape": 0.808750331401825,
+    "lr": 7.79517727615026e-05,
+    "elapsed_sec": 919.0373272895813
+  },
+  {
+    "step": 3300,
+    "train_shape": 0.8068431615829468,
+    "lr": 7.667946724814286e-05,
+    "elapsed_sec": 947.2532444000244
+  },
+  {
+    "step": 3400,
+    "train_shape": 0.8046305179595947,
+    "lr": 7.538329951983173e-05,
+    "elapsed_sec": 975.69184923172
+  },
+  {
+    "step": 3500,
+    "train_shape": 0.8582855463027954,
+    "lr": 7.406454873762866e-05,
+    "elapsed_sec": 1003.8970856666565
+  },
+  {
+    "step": 3500,
+    "val_shape": 0.8492363691329956
+  },
+  {
+    "step": 3600,
+    "train_shape": 0.7914978265762329,
+    "lr": 7.272451634934115e-05,
+    "elapsed_sec": 1033.0489664077759
+  },
+  {
+    "step": 3700,
+    "train_shape": 0.7868252992630005,
+    "lr": 7.136452480515233e-05,
+    "elapsed_sec": 1061.2489976882935
+  },
+  {
+    "step": 3800,
+    "train_shape": 0.797909677028656,
+    "lr": 6.99859162525224e-05,
+    "elapsed_sec": 1089.458578824997
+  },
+  {
+    "step": 3900,
+    "train_shape": 0.7972847819328308,
+    "lr": 6.859005121165148e-05,
+    "elapsed_sec": 1117.6733033657074
+  },
+  {
+    "step": 4000,
+    "train_shape": 0.8413028120994568,
+    "lr": 6.71783072328101e-05,
+    "elapsed_sec": 1145.882479429245
+  },
+  {
+    "step": 4000,
+    "val_shape": 0.8442273139953613
+  },
+  {
+    "step": 4100,
+    "train_shape": 0.7729045152664185,
+    "lr": 6.575207753686356e-05,
+    "elapsed_sec": 1175.0235650539398
+  },
+  {
+    "step": 4200,
+    "train_shape": 0.7737760543823242,
+    "lr": 6.431276964033111e-05,
+    "elapsed_sec": 1203.2464413642883
+  },
+  {
+    "step": 4300,
+    "train_shape": 0.7827517986297607,
+    "lr": 6.286180396633663e-05,
+    "elapsed_sec": 1231.4651429653168
+  },
+  {
+    "step": 4400,
+    "train_shape": 0.8428584337234497,
+    "lr": 6.140061244282311e-05,
+    "elapsed_sec": 1259.6782433986664
+  },
+  {
+    "step": 4500,
+    "train_shape": 0.7634469270706177,
+    "lr": 5.9930637089412255e-05,
+    "elapsed_sec": 1287.8907580375671
+  },
+  {
+    "step": 4500,
+    "val_shape": 0.8420872688293457
+  },
+  {
+    "step": 4600,
+    "train_shape": 0.7443527579307556,
+    "lr": 5.845332859430572e-05,
+    "elapsed_sec": 1317.0973958969116
+  },
+  {
+    "step": 4700,
+    "train_shape": 0.7786215543746948,
+    "lr": 5.697014488263067e-05,
+    "elapsed_sec": 1345.4626286029816
+  },
+  {
+    "step": 4800,
+    "train_shape": 0.7865493297576904,
+    "lr": 5.548254967764354e-05,
+    "elapsed_sec": 1373.8456404209137
+  },
+  {
+    "step": 4900,
+    "train_shape": 0.8457694053649902,
+    "lr": 5.3992011056212214e-05,
+    "elapsed_sec": 1402.1039729118347
+  },
+  {
+    "step": 5000,
+    "train_shape": 0.8066558837890625,
+    "lr": 5.250000000000107e-05,
+    "elapsed_sec": 1430.4693086147308
+  },
+  {
+    "step": 5000,
+    "val_shape": 0.8387514352798462
+  },
+  {
+    "step": 5100,
+    "train_shape": 0.8486970663070679,
+    "lr": 5.100798894378995e-05,
+    "elapsed_sec": 1461.4542706012726
+  },
+  {
+    "step": 5200,
+    "train_shape": 0.826057493686676,
+    "lr": 4.951745032235864e-05,
+    "elapsed_sec": 1489.6694586277008
+  },
+  {
+    "step": 5300,
+    "train_shape": 0.7918369174003601,
+    "lr": 4.8029855117371535e-05,
+    "elapsed_sec": 1517.8915855884552
+  },
+  {
+    "step": 5400,
+    "train_shape": 0.7885835766792297,
+    "lr": 4.654667140569645e-05,
+    "elapsed_sec": 1546.1066854000092
+  },
+  {
+    "step": 5500,
+    "train_shape": 0.7876046299934387,
+    "lr": 4.506936291058995e-05,
+    "elapsed_sec": 1574.3269317150116
+  },
+  {
+    "step": 5500,
+    "val_shape": 0.8344401121139526
+  },
+  {
+    "step": 5600,
+    "train_shape": 0.7392200231552124,
+    "lr": 4.359938755717899e-05,
+    "elapsed_sec": 1603.4569492340088
+  },
+  {
+    "step": 5700,
+    "train_shape": 0.7479463219642639,
+    "lr": 4.213819603366514e-05,
+    "elapsed_sec": 1631.6661064624786
+  },
+  {
+    "step": 5800,
+    "train_shape": 0.8059396743774414,
+    "lr": 4.0687230359670296e-05,
+    "elapsed_sec": 1659.8773169517517
+  },
+  {
+    "step": 5900,
+    "train_shape": 0.7620919942855835,
+    "lr": 3.924792246313749e-05,
+    "elapsed_sec": 1688.080369234085
+  },
+  {
+    "step": 6000,
+    "train_shape": 0.7597255706787109,
+    "lr": 3.7821692767190845e-05,
+    "elapsed_sec": 1716.3113341331482
+  },
+  {
+    "step": 6000,
+    "val_shape": 0.8316431045532227
+  },
+  {
+    "step": 6100,
+    "train_shape": 0.7596187591552734,
+    "lr": 3.640994878834947e-05,
+    "elapsed_sec": 1745.7161622047424
+  },
+  {
+    "step": 6200,
+    "train_shape": 0.7682507038116455,
+    "lr": 3.501408374747862e-05,
+    "elapsed_sec": 1773.9208846092224
+  },
+  {
+    "step": 6300,
+    "train_shape": 0.7231128811836243,
+    "lr": 3.363547519484887e-05,
+    "elapsed_sec": 1802.1198375225067
+  },
+  {
+    "step": 6400,
+    "train_shape": 0.7122642993927002,
+    "lr": 3.227548365066013e-05,
+    "elapsed_sec": 1830.3174185752869
+  },
+  {
+    "step": 6500,
+    "train_shape": 0.7290594577789307,
+    "lr": 3.0935451262372706e-05,
+    "elapsed_sec": 1858.5141904354095
+  },
+  {
+    "step": 6500,
+    "val_shape": 0.8293938636779785
+  },
+  {
+    "step": 6600,
+    "train_shape": 0.7610007524490356,
+    "lr": 2.9616700480169673e-05,
+    "elapsed_sec": 1887.6447110176086
+  },
+  {
+    "step": 6700,
+    "train_shape": 0.8001368045806885,
+    "lr": 2.8320532751858457e-05,
+    "elapsed_sec": 1915.8440294265747
+  },
+  {
+    "step": 6800,
+    "train_shape": 0.7795954942703247,
+    "lr": 2.7048227238498715e-05,
+    "elapsed_sec": 1944.0467538833618
+  },
+  {
+    "step": 6900,
+    "train_shape": 0.7534531354904175,
+    "lr": 2.5801039552024805e-05,
+    "elapsed_sec": 1972.2502312660217
+  },
+  {
+    "step": 7000,
+    "train_shape": 0.8005579113960266,
+    "lr": 2.458020051610849e-05,
+    "elapsed_sec": 2000.5970103740692
+  },
+  {
+    "step": 7000,
+    "val_shape": 0.8283576965332031
+  },
+  {
+    "step": 7100,
+    "train_shape": 0.7591400146484375,
+    "lr": 2.3386914951484544e-05,
+    "elapsed_sec": 2029.684012889862
+  },
+  {
+    "step": 7200,
+    "train_shape": 0.7876743674278259,
+    "lr": 2.222236048693809e-05,
+    "elapsed_sec": 2057.8901505470276
+  },
+  {
+    "step": 7300,
+    "train_shape": 0.8336132168769836,
+    "lr": 2.1087686397127328e-05,
+    "elapsed_sec": 2086.090241909027
+  },
+  {
+    "step": 7400,
+    "train_shape": 0.7999813556671143,
+    "lr": 1.998401246838802e-05,
+    "elapsed_sec": 2114.40247130394
+  },
+  {
+    "step": 7500,
+    "train_shape": 0.7676112055778503,
+    "lr": 1.891242789363976e-05,
+    "elapsed_sec": 2142.599452495575
+  },
+  {
+    "step": 7500,
+    "val_shape": 0.8270952701568604
+  },
+  {
+    "step": 7600,
+    "train_shape": 0.7755910158157349,
+    "lr": 1.787399019748373e-05,
+    "elapsed_sec": 2171.7514951229095
+  },
+  {
+    "step": 7700,
+    "train_shape": 0.829744815826416,
+    "lr": 1.6869724192553968e-05,
+    "elapsed_sec": 2199.941335916519
+  },
+  {
+    "step": 7800,
+    "train_shape": 0.7845275402069092,
+    "lr": 1.5900620968150808e-05,
+    "elapsed_sec": 2228.135841369629
+  },
+  {
+    "step": 7900,
+    "train_shape": 0.7701389193534851,
+    "lr": 1.496763691215551e-05,
+    "elapsed_sec": 2256.3702108860016
+  },
+  {
+    "step": 8000,
+    "train_shape": 0.7890292406082153,
+    "lr": 1.4071692767190806e-05,
+    "elapsed_sec": 2284.5678894519806
+  },
+  {
+    "step": 8000,
+    "val_shape": 0.8253313302993774
+  },
+  {
+    "step": 8100,
+    "train_shape": 0.8098048567771912,
+    "lr": 1.3213672721959122e-05,
+    "elapsed_sec": 2313.6740515232086
+  },
+  {
+    "step": 8200,
+    "train_shape": 0.7152670621871948,
+    "lr": 1.2394423538655062e-05,
+    "elapsed_sec": 2341.874612569809
+  },
+  {
+    "step": 8300,
+    "train_shape": 0.784916341304779,
+    "lr": 1.1614753717313381e-05,
+    "elapsed_sec": 2370.0865817070007
+  },
+  {
+    "step": 8400,
+    "train_shape": 0.7638041973114014,
+    "lr": 1.0875432697917097e-05,
+    "elapsed_sec": 2398.284777879715
+  },
+  {
+    "step": 8500,
+    "train_shape": 0.7512559294700623,
+    "lr": 1.0177190101053074e-05,
+    "elapsed_sec": 2426.4895071983337
+  },
+  {
+    "step": 8500,
+    "val_shape": 0.8244471549987793
+  },
+  {
+    "step": 8600,
+    "train_shape": 0.7741738557815552,
+    "lr": 9.520715007864551e-06,
+    "elapsed_sec": 2455.603847026825
+  },
+  {
+    "step": 8700,
+    "train_shape": 0.7788733243942261,
+    "lr": 8.906655280011289e-06,
+    "elapsed_sec": 2483.8072555065155
+  },
+  {
+    "step": 8800,
+    "train_shape": 0.8083540797233582,
+    "lr": 8.335616920308386e-06,
+    "elapsed_sec": 2512.2269921302795
+  },
+  {
+    "step": 8900,
+    "train_shape": 0.7531920671463013,
+    "lr": 7.808163474674577e-06,
+    "elapsed_sec": 2540.422973394394
+  },
+  {
+    "step": 9000,
+    "train_shape": 0.7555859088897705,
+    "lr": 7.32481547598044e-06,
+    "elapsed_sec": 2568.6235115528107
+  },
+  {
+    "step": 9000,
+    "val_shape": 0.823600172996521
+  },
+  {
+    "step": 9100,
+    "train_shape": 0.8012001514434814,
+    "lr": 6.886049930345396e-06,
+    "elapsed_sec": 2597.822489976883
+  },
+  {
+    "step": 9200,
+    "train_shape": 0.8056222796440125,
+    "lr": 6.492299846390179e-06,
+    "elapsed_sec": 2626.02307677269
+  },
+  {
+    "step": 9300,
+    "train_shape": 0.7770637273788452,
+    "lr": 6.143953807909618e-06,
+    "elapsed_sec": 2654.3207063674927
+  },
+  {
+    "step": 9400,
+    "train_shape": 0.7398402690887451,
+    "lr": 5.841355590387376e-06,
+    "elapsed_sec": 2682.5214760303497
+  },
+  {
+    "step": 9500,
+    "train_shape": 0.7921051979064941,
+    "lr": 5.5848038217310204e-06,
+    "elapsed_sec": 2710.7262189388275
+  },
+  {
+    "step": 9500,
+    "val_shape": 0.8229491710662842
+  },
+  {
+    "step": 9600,
+    "train_shape": 0.7612913250923157,
+    "lr": 5.374551687562344e-06,
+    "elapsed_sec": 2739.8656091690063
+  },
+  {
+    "step": 9700,
+    "train_shape": 0.7580114603042603,
+    "lr": 5.2108066813537186e-06,
+    "elapsed_sec": 2768.0596215724945
+  },
+  {
+    "step": 9800,
+    "train_shape": 0.7784627079963684,
+    "lr": 5.093730399657109e-06,
+    "elapsed_sec": 2796.2510480880737
+  },
+  {
+    "step": 9900,
+    "train_shape": 0.7518530488014221,
+    "lr": 5.023438382627753e-06,
+    "elapsed_sec": 2824.4482271671295
+  },
+  {
+    "step": 10000,
+    "train_shape": 0.8090463876724243,
+    "lr": 5e-06,
+    "elapsed_sec": 2852.640844106674
+  },
+  {
+    "step": 10000,
+    "val_shape": 0.8226728439331055
+  }
+]
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..e01a62366e6d949d6be7204798e90da22f494efb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:42.722379238Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:43.036296394Z","level":"INFO","msg":"stream: created new stream","id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.036382323Z","level":"INFO","msg":"handler: started","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.038275557Z","level":"INFO","msg":"stream: started","id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.038311473Z","level":"INFO","msg":"writer: started","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.038311263Z","level":"INFO","msg":"sender: started","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.280634788Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:22.52020952Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:22.525542581Z","level":"INFO","msg":"stream: closing","id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.525560664Z","level":"INFO","msg":"handler: closed","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.526744074Z","level":"INFO","msg":"sender: closed","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.526758551Z","level":"INFO","msg":"stream: closed","id":"3mjkolf3"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/debug.log b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..d92fa14ef50d0e9b596849c58579b0499c711efb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:42,504 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug.log
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug-internal.log
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz', 'image_view': 'side', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.0, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-plain-side-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,side', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:42,714 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:42,719 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:42,720 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:42,735 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:43,362 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:21,358 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/3mjkolf3
+2026-08-16 20:03:21,366 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:21,366 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:21,366 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:22,523 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/config.yaml b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..7716fcb9ea170b0093c167836513a0e29cf2fec8
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/config.yaml
@@ -0,0 +1,172 @@
+_wandb:
+    value:
+        cli_version: 0.22.3
+        e:
+            dtwg8ceag5t2ngn2g8nlsdlom0o38mgy:
+                args:
+                    - --mode
+                    - train
+                    - --image-view
+                    - side
+                    - --paired-latents
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz
+                    - --out-dir
+                    - /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side
+                    - --steps
+                    - "10000"
+                    - --batch-size
+                    - "128"
+                    - --lr
+                    - "1e-4"
+                    - --weight-decay
+                    - "0.01"
+                    - --seed
+                    - "20260526"
+                    - --pos-embed
+                    - 2d
+                    - --cond-dropout
+                    - "0.0"
+                    - --resume
+                    - --wandb
+                    - --wandb-run-name
+                    - drivaer3v-inputcond-plain-side-0816c
+                    - --wandb-tags
+                    - drivaer,qwen3v,input-conditional,channel-concat,side
+                codePath: code/ICLR/customized-design/input_cond/train_cond_flow.py
+                codePathLocal: train_cond_flow.py
+                cpu_count: 32
+                cpu_count_logical: 64
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "7679361679360"
+                        used: "2800196116480"
+                email: 1193414549@qq.com
+                executable: /mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python
+                git:
+                    commit: 81d95989e1279e432c4145933bf9ce02c3f70f77
+                gpu: NVIDIA A100-SXM4-40GB
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "42949672960"
+                      name: NVIDIA A100-SXM4-40GB
+                      uuid: GPU-086d5a87-2d0c-ecc2-e858-d39983c380ed
+                host: inputcond-plain-side-a100-40g-0816c-0-0
+                memory:
+                    total: "540741947392"
+                os: Linux-6.8.0-124-generic-x86_64-with-glibc2.35
+                program: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py
+                python: CPython 3.10.18
+                root: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side/wandb
+                startedAt: "2026-08-16T19:15:42.479849Z"
+                writerId: dtwg8ceag5t2ngn2g8nlsdlom0o38mgy
+        m: []
+        python_version: 3.10.18
+        t:
+            "1":
+                - 1
+                - 105
+            "2":
+                - 1
+                - 105
+            "3":
+                - 2
+                - 13
+                - 15
+                - 16
+                - 61
+                - 62
+            "4": 3.10.18
+            "5": 0.22.3
+            "12": 0.22.3
+            "13": linux-x86_64
+attention_backend:
+    value: sdpa
+batch_size:
+    value: 128
+cond_dropout:
+    value: 0
+decode_mesh:
+    value: false
+depth:
+    value: 12
+dropout:
+    value: 0
+eval_every:
+    value: 500
+eval_noise_seed:
+    value: 20260712
+gradient_checkpointing:
+    value: false
+guidance_scale:
+    value: 1
+head_dim:
+    value: 64
+hidden_size:
+    value: 512
+image_scale:
+    value: 0
+image_view:
+    value: side
+log_every:
+    value: 100
+lr:
+    value: 0.0001
+max_val:
+    value: 128
+mlp_ratio:
+    value: 4
+mode:
+    value: train
+num_heads:
+    value: 8
+num_workers:
+    value: 2
+out_dir:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side
+paired_latents:
+    value: /mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz
+pos_embed:
+    value: 2d
+resume:
+    value: true
+sample_checkpoint:
+    value: model_best.pt
+sample_count:
+    value: 8
+sample_seed:
+    value: 20260712
+sample_steps:
+    value: 50
+save_every:
+    value: 5000
+seed:
+    value: 20260526
+shape_max_batch:
+    value: 32768
+shape_resolution:
+    value: 512
+shape_scale:
+    value: 0
+steps:
+    value: 10000
+val_fraction:
+    value: 0.05
+val_overlap_train:
+    value: true
+wandb:
+    value: true
+wandb_entity:
+    value: kanydao-epfl
+wandb_mode:
+    value: online
+wandb_project:
+    value: drivaer-qwen3v-input-cond
+wandb_run_name:
+    value: drivaer3v-inputcond-plain-side-0816c
+wandb_tags:
+    value: drivaer,qwen3v,input-conditional,channel-concat,side
+weight_decay:
+    value: 0.01
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/output.log b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..ee17b7d8016b66c890cbd3c25f45c3fde4f5d979
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/output.log
@@ -0,0 +1,123 @@
+[wandb] project=drivaer-qwen3v-input-cond run=drivaer3v-inputcond-plain-side-0816c mode=online
+[train] step=000001 loss=2.009530 lr=1.00e-04
+[val] step=000001 shape=2.006118
+[train] step=000100 loss=1.279186 lr=1.00e-04
+[train] step=000200 loss=1.040881 lr=9.99e-05
+[train] step=000300 loss=0.930727 lr=9.98e-05
+[train] step=000400 loss=0.885930 lr=9.96e-05
+[train] step=000500 loss=0.870642 lr=9.94e-05
+[val] step=000500 shape=0.931237
+[train] step=000600 loss=0.867961 lr=9.92e-05
+[train] step=000700 loss=0.805919 lr=9.89e-05
+[train] step=000800 loss=0.887097 lr=9.85e-05
+[train] step=000900 loss=0.841090 lr=9.81e-05
+[train] step=001000 loss=0.845045 lr=9.77e-05
+[val] step=001000 shape=0.898779
+[train] step=001100 loss=0.797426 lr=9.72e-05
+[train] step=001200 loss=0.879705 lr=9.67e-05
+[train] step=001300 loss=0.816932 lr=9.61e-05
+[train] step=001400 loss=0.875819 lr=9.55e-05
+[train] step=001500 loss=0.816307 lr=9.48e-05
+[val] step=001500 shape=0.885114
+[train] step=001600 loss=0.872526 lr=9.41e-05
+[train] step=001700 loss=0.783596 lr=9.34e-05
+[train] step=001800 loss=0.841702 lr=9.26e-05
+[train] step=001900 loss=0.842311 lr=9.18e-05
+[train] step=002000 loss=0.859478 lr=9.09e-05
+[val] step=002000 shape=0.872692
+[train] step=002100 loss=0.825168 lr=9.00e-05
+[train] step=002200 loss=0.817196 lr=8.91e-05
+[train] step=002300 loss=0.878228 lr=8.81e-05
+[train] step=002400 loss=0.784439 lr=8.71e-05
+[train] step=002500 loss=0.863948 lr=8.61e-05
+[val] step=002500 shape=0.864745
+[train] step=002600 loss=0.814274 lr=8.50e-05
+[train] step=002700 loss=0.812052 lr=8.39e-05
+[train] step=002800 loss=0.766645 lr=8.28e-05
+[train] step=002900 loss=0.823058 lr=8.16e-05
+[train] step=003000 loss=0.804383 lr=8.04e-05
+[val] step=003000 shape=0.855979
+[train] step=003100 loss=0.787870 lr=7.92e-05
+[train] step=003200 loss=0.808750 lr=7.80e-05
+[train] step=003300 loss=0.806843 lr=7.67e-05
+[train] step=003400 loss=0.804631 lr=7.54e-05
+[train] step=003500 loss=0.858286 lr=7.41e-05
+[val] step=003500 shape=0.849236
+[train] step=003600 loss=0.791498 lr=7.27e-05
+[train] step=003700 loss=0.786825 lr=7.14e-05
+[train] step=003800 loss=0.797910 lr=7.00e-05
+[train] step=003900 loss=0.797285 lr=6.86e-05
+[train] step=004000 loss=0.841303 lr=6.72e-05
+[val] step=004000 shape=0.844227
+[train] step=004100 loss=0.772905 lr=6.58e-05
+[train] step=004200 loss=0.773776 lr=6.43e-05
+[train] step=004300 loss=0.782752 lr=6.29e-05
+[train] step=004400 loss=0.842858 lr=6.14e-05
+[train] step=004500 loss=0.763447 lr=5.99e-05
+[val] step=004500 shape=0.842087
+[train] step=004600 loss=0.744353 lr=5.85e-05
+[train] step=004700 loss=0.778622 lr=5.70e-05
+[train] step=004800 loss=0.786549 lr=5.55e-05
+[train] step=004900 loss=0.845769 lr=5.40e-05
+[train] step=005000 loss=0.806656 lr=5.25e-05
+[val] step=005000 shape=0.838751
+[train] step=005100 loss=0.848697 lr=5.10e-05
+[train] step=005200 loss=0.826057 lr=4.95e-05
+[train] step=005300 loss=0.791837 lr=4.80e-05
+[train] step=005400 loss=0.788584 lr=4.65e-05
+[train] step=005500 loss=0.787605 lr=4.51e-05
+[val] step=005500 shape=0.834440
+[train] step=005600 loss=0.739220 lr=4.36e-05
+[train] step=005700 loss=0.747946 lr=4.21e-05
+[train] step=005800 loss=0.805940 lr=4.07e-05
+[train] step=005900 loss=0.762092 lr=3.92e-05
+[train] step=006000 loss=0.759726 lr=3.78e-05
+[val] step=006000 shape=0.831643
+[train] step=006100 loss=0.759619 lr=3.64e-05
+[train] step=006200 loss=0.768251 lr=3.50e-05
+[train] step=006300 loss=0.723113 lr=3.36e-05
+[train] step=006400 loss=0.712264 lr=3.23e-05
+[train] step=006500 loss=0.729059 lr=3.09e-05
+[val] step=006500 shape=0.829394
+[train] step=006600 loss=0.761001 lr=2.96e-05
+[train] step=006700 loss=0.800137 lr=2.83e-05
+[train] step=006800 loss=0.779595 lr=2.70e-05
+[train] step=006900 loss=0.753453 lr=2.58e-05
+[train] step=007000 loss=0.800558 lr=2.46e-05
+[val] step=007000 shape=0.828358
+[train] step=007100 loss=0.759140 lr=2.34e-05
+[train] step=007200 loss=0.787674 lr=2.22e-05
+[train] step=007300 loss=0.833613 lr=2.11e-05
+[train] step=007400 loss=0.799981 lr=2.00e-05
+[train] step=007500 loss=0.767611 lr=1.89e-05
+[val] step=007500 shape=0.827095
+[train] step=007600 loss=0.775591 lr=1.79e-05
+[train] step=007700 loss=0.829745 lr=1.69e-05
+[train] step=007800 loss=0.784528 lr=1.59e-05
+[train] step=007900 loss=0.770139 lr=1.50e-05
+[train] step=008000 loss=0.789029 lr=1.41e-05
+[val] step=008000 shape=0.825331
+[train] step=008100 loss=0.809805 lr=1.32e-05
+[train] step=008200 loss=0.715267 lr=1.24e-05
+[train] step=008300 loss=0.784916 lr=1.16e-05
+[train] step=008400 loss=0.763804 lr=1.09e-05
+[train] step=008500 loss=0.751256 lr=1.02e-05
+[val] step=008500 shape=0.824447
+[train] step=008600 loss=0.774174 lr=9.52e-06
+[train] step=008700 loss=0.778873 lr=8.91e-06
+[train] step=008800 loss=0.808354 lr=8.34e-06
+[train] step=008900 loss=0.753192 lr=7.81e-06
+[train] step=009000 loss=0.755586 lr=7.32e-06
+[val] step=009000 shape=0.823600
+[train] step=009100 loss=0.801200 lr=6.89e-06
+[train] step=009200 loss=0.805622 lr=6.49e-06
+[train] step=009300 loss=0.777064 lr=6.14e-06
+[train] step=009400 loss=0.739840 lr=5.84e-06
+[train] step=009500 loss=0.792105 lr=5.58e-06
+[val] step=009500 shape=0.822949
+[train] step=009600 loss=0.761291 lr=5.37e-06
+[train] step=009700 loss=0.758011 lr=5.21e-06
+[train] step=009800 loss=0.778463 lr=5.09e-06
+[train] step=009900 loss=0.751853 lr=5.02e-06
+[train] step=010000 loss=0.809046 lr=5.00e-06
+[val] step=010000 shape=0.822673
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/requirements.txt b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..2e10c9ab6bc99041706780299f9e3e5035f34676
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/requirements.txt
@@ -0,0 +1,302 @@
+shellingham==1.5.4
+importlib_metadata==8.7.0
+websocket-client==1.8.0
+rfc3986-validator==0.1.1
+aiohappyeyeballs==2.6.1
+tensorboardX==2.6.2.2
+matplotlib==3.8.4
+zstandard==0.23.0
+gradio_client==1.4.2
+nbclient==0.10.2
+pyaml==25.7.0
+argparse==1.4.0
+h11==0.16.0
+scooby==0.10.1
+pexpect==4.9.0
+jedi==0.19.2
+scikit-optimize==0.10.2
+omegaconf==2.3.0
+tomlkit==0.12.0
+yarl==1.20.1
+point-cloud-utils==0.34.0
+cffi==1.17.1
+asttokens==3.0.0
+webencodings==0.5.1
+imageio==2.34.1
+attrs==25.3.0
+hydra-core==1.3.2
+datasets==2.19.0
+pyquaternion==0.9.9
+trame-common==1.0.0
+wslink==2.3.4
+Pygments==2.19.2
+pydantic==2.11.7
+regex==2024.11.6
+timm==0.9.16
+ipykernel==6.29.5
+prompt_toolkit==3.0.51
+torchdiffeq==0.2.5
+pybind11==2.13.6
+debugpy==1.8.14
+pytz==2025.2
+transformers==4.40.1
+jupyterlab_pygments==0.3.0
+dash==3.1.1
+psutil==5.9.8
+tqdm==4.66.2
+multidict==6.6.3
+setproctitle==1.3.6
+nbformat==5.10.4
+prometheus_client==0.22.1
+jupyter_core==5.8.1
+gradio==5.5.0
+multiprocess==0.70.16
+fastapi==0.115.14
+simpervisor==1.0.0
+comm==0.2.2
+widgetsnbextension==4.0.14
+webcolors==24.11.1
+sentry-sdk==2.32.0
+tornado==6.5.1
+defusedxml==0.7.1
+Markdown==3.8.2
+numba==0.61.2
+pyarrow-hotfix==0.7
+trame-server==3.5.0
+nbconvert==7.16.6
+websockets==12.0
+semantic-version==2.10.0
+bpy==4.0.0
+smmap==5.0.2
+grpcio==1.73.1
+referencing==0.36.2
+annotated-types==0.7.0
+stack-data==0.6.3
+vtk==9.4.2
+safehttpx==0.1.6
+GitPython==3.1.44
+ipywidgets==8.1.7
+platformdirs==4.3.8
+tzdata==2025.2
+typer==0.16.0
+click==8.2.1
+networkx==3.3
+fsspec==2024.3.1
+plotly==6.2.0
+isoduration==20.11.0
+joblib==1.4.0
+aiosignal==1.4.0
+jsonschema-specifications==2025.4.1
+six==1.17.0
+pyzmq==27.0.0
+lazy_loader==0.4
+typeguard==2.13.3
+markdown-it-py==3.0.0
+httpx==0.28.1
+fqdn==1.5.1
+threadpoolctl==3.6.0
+orjson==3.10.18
+docker-pycreds==0.4.0
+jsonschema==4.24.0
+tensorboard-data-server==0.7.2
+nvidia-cufft-cu12==11.0.2.54
+aiofiles==23.2.1
+rpds-py==0.26.0
+Cython==3.1.2
+python-multipart==0.0.12
+nvidia-cuda-runtime-cu12==12.1.105
+huggingface-hub==0.26.2
+frozenlist==1.7.0
+onnxruntime==1.22.0
+cycler==0.12.1
+tifffile==2025.5.10
+safetensors==0.4.3
+nest-asyncio==1.6.0
+parso==0.8.4
+PyMCubes==0.1.4
+nvidia-nvtx-cu12==12.1.105
+pydub==0.25.1
+packaging==25.0
+ipython==8.37.0
+uvicorn==0.35.0
+pyarrow==20.0.0
+rich==14.0.0
+executing==2.2.0
+jupyter_server_terminals==0.5.3
+typing_extensions==4.12.2
+typing_extensions==4.15.0
+tensorboard==2.16.2
+argon2-cffi-bindings==21.2.0
+wcwidth==0.2.13
+typing-inspection==0.4.1
+pure_eval==0.2.3
+trimesh==4.3.2
+nvidia-cudnn-cu12==9.1.0.70
+humanfriendly==10.0
+Werkzeug==3.1.3
+decorator==5.2.1
+nvidia-cuda-nvrtc-cu12==12.1.105
+aiohttp==3.12.13
+uri-template==1.3.0
+anyio==4.9.0
+fastjsonschema==2.21.1
+nvidia-curand-cu12==10.3.2.106
+zipp==3.23.0
+rtree==1.4.1
+kiwisolver==1.4.8
+Flask==3.1.1
+nvidia-cuda-cupti-cu12==12.1.105
+pysdf==0.1.9
+sniffio==1.3.1
+itsdangerous==2.2.0
+rfc3339-validator==0.1.4
+trame==3.10.2
+open3d==0.18.0
+scikit-learn==1.7.0
+mistune==3.1.3
+diffusers==0.31.0
+absl-py==2.3.1
+cubvh==0.1.0
+trame-client==3.9.1
+einops==0.8.0
+plyfile==1.0.3
+bleach==6.2.0
+ptyprocess==0.7.0
+pycparser==2.22
+llvmlite==0.44.0
+addict==2.4.0
+diso==0.1.4
+Send2Trash==1.8.3
+more-itertools==10.7.0
+scikit-image==0.23.2
+propcache==0.3.2
+pandas==2.2.2
+flatbuffers==25.2.10
+gitdb==4.0.12
+jaxtyping==0.2.28
+soupsieve==2.7
+mdurl==0.1.2
+protobuf==5.29.5
+fonttools==4.58.5
+coloredlogs==15.0.1
+spaces==0.28.3
+jupyter_server==2.16.0
+trame-vtk==2.9.0
+terminado==0.18.1
+overrides==7.7.0
+pandocfilters==1.5.1
+nvidia-cusolver-cu12==11.4.5.107
+trame-vuetify==3.0.1
+async-timeout==5.0.1
+blinker==1.9.0
+exceptiongroup==1.3.0
+xxhash==3.5.0
+python-json-logger==3.3.0
+pyvista==0.45.2
+python-dateutil==2.9.0.post0
+ConfigArgParse==1.7.1
+PyMatting==1.1.14
+fpsample==0.3.3
+traitlets==5.14.3
+msgpack==1.1.1
+argon2-cffi==25.1.0
+tinycss2==1.4.0
+retrying==1.4.0
+jsonpointer==3.0.0
+matplotlib-inline==0.1.7
+jupyter_client==8.6.3
+contourpy==1.3.2
+lightning-utilities==0.11.2
+pydantic_core==2.33.2
+httpcore==1.0.9
+narwhals==1.46.0
+nvidia-cublas-cu12==12.1.3.1
+jupyterlab_widgets==3.0.15
+antlr4-python3-runtime==4.9.3
+pyparsing==3.1.2
+rembg==2.0.59
+ffmpy==0.6.0
+tokenizers==0.19.1
+beautifulsoup4==4.13.4
+pooch==1.8.2
+ninja==1.11.1.4
+jupyter_server_proxy==4.4.0
+ruff==0.12.2
+types-python-dateutil==2.9.0.20250708
+dill==0.3.8
+pytorch-lightning==2.2.4
+jupyter-events==0.12.0
+starlette==0.46.2
+arrow==1.3.0
+setuptools==78.1.1
+wheel==0.45.1
+pip==25.1
+mkl-service==2.5.2
+mpi4py==4.0.3
+h5py==3.14.0
+mkl_fft==2.1.1
+mkl_random==1.3.0
+mcubes==0.1.6
+wandb==0.22.3
+certifi==2026.1.4
+mpmath==1.3.0
+triton==3.1.0
+sympy==1.13.1
+nvidia-nccl-cu12==2.21.5
+nvidia-nvjitlink-cu12==12.9.86
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nvtx-cu11==11.8.86
+nvidia-nccl-cu11==2.21.5
+nvidia-cusparse-cu11==11.7.5.86
+nvidia-curand-cu11==10.3.0.86
+nvidia-cufft-cu11==10.9.0.58
+nvidia-cuda-runtime-cu11==11.8.89
+nvidia-cuda-nvrtc-cu11==11.8.89
+nvidia-cuda-cupti-cu11==11.8.87
+nvidia-cublas-cu11==11.11.3.6
+nvidia-cusolver-cu11==11.4.1.48
+nvidia-cudnn-cu11==9.1.0.70
+torch==2.5.1+cu118
+torchvision==0.20.1+cu118
+torchaudio==2.5.1+cu118
+PyYAML==6.0.3
+pillow==12.1.0
+torchmetrics==1.8.2
+accelerate==1.12.0
+urllib3==2.6.3
+idna==3.11
+charset-normalizer==3.4.4
+requests==2.32.5
+filelock==3.20.3
+torch_cluster==1.6.3+pt25cu118
+numpy==1.26.4
+opencv-python-headless==4.9.0.80
+PyOpenGL-accelerate==3.1.10
+PyOpenGL==3.1.0
+pyglet==2.1.14
+freetype-py==2.5.1
+pyrender==0.1.45
+scipy==1.15.3
+memory-profiler==0.61.0
+pyamg==5.3.0
+meshio==5.3.5
+torch-fem==0.6.2
+torch-geometric==2.7.0
+MarkupSafe==3.0.3
+Jinja2==3.1.6
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/wandb-metadata.json b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..13492c3e524e025e1abcbe2ee94c1fa417826726
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/wandb-metadata.json
@@ -0,0 +1,69 @@
+{
+  "os":  "Linux-6.8.0-124-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.10.18",
+  "startedAt":  "2026-08-16T19:15:42.479849Z",
+  "args":  [
+    "--mode",
+    "train",
+    "--image-view",
+    "side",
+    "--paired-latents",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz",
+    "--out-dir",
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side",
+    "--steps",
+    "10000",
+    "--batch-size",
+    "128",
+    "--lr",
+    "1e-4",
+    "--weight-decay",
+    "0.01",
+    "--seed",
+    "20260526",
+    "--pos-embed",
+    "2d",
+    "--cond-dropout",
+    "0.0",
+    "--resume",
+    "--wandb",
+    "--wandb-run-name",
+    "drivaer3v-inputcond-plain-side-0816c",
+    "--wandb-tags",
+    "drivaer,qwen3v,input-conditional,channel-concat,side"
+  ],
+  "program":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePath":  "code/ICLR/customized-design/input_cond/train_cond_flow.py",
+  "codePathLocal":  "train_cond_flow.py",
+  "git":  {
+    "commit":  "81d95989e1279e432c4145933bf9ce02c3f70f77"
+  },
+  "email":  "1193414549@qq.com",
+  "root":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side/wandb",
+  "host":  "inputcond-plain-side-a100-40g-0816c-0-0",
+  "executable":  "/mnt/cvlab/scratch/cvlab/home/hantzhan/anaconda3/envs/dora/bin/python",
+  "cpu_count":  32,
+  "cpu_count_logical":  64,
+  "gpu":  "NVIDIA A100-SXM4-40GB",
+  "gpu_count":  1,
+  "disk":  {
+    "/":  {
+      "total":  "7679361679360",
+      "used":  "2800196116480"
+    }
+  },
+  "memory":  {
+    "total":  "540741947392"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A100-SXM4-40GB",
+      "memoryTotal":  "42949672960",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-086d5a87-2d0c-ecc2-e858-d39983c380ed"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "dtwg8ceag5t2ngn2g8nlsdlom0o38mgy"
+}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/wandb-summary.json b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..2521599e90a93cbb7e0d64416b35df43cd1efcb7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/files/wandb-summary.json
@@ -0,0 +1 @@
+{"steps":10000,"val_samples":128,"val/shape":0.8226728439331055,"_timestamp":1.786910596681807e+09,"best_val_shape":0.8226728439331055,"time/elapsed_sec":2852.640844106674,"train/lr":5e-06,"image_view":"side","conditioning":"sketch_latent_channel_concat","train/shape":0.8090463876724243,"image_scale":1.4969936040640752,"shape_scale":1.7132293052897125,"_step":10000,"_wandb":{"runtime":2858},"_runtime":2858.003965344,"train_samples":3051}
\ No newline at end of file
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug-internal.log b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..e01a62366e6d949d6be7204798e90da22f494efb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2026-08-16T19:15:42.722379238Z","level":"INFO","msg":"stream: starting","core version":"0.22.3"}
+{"time":"2026-08-16T19:15:43.036296394Z","level":"INFO","msg":"stream: created new stream","id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.036382323Z","level":"INFO","msg":"handler: started","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.038275557Z","level":"INFO","msg":"stream: started","id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.038311473Z","level":"INFO","msg":"writer: started","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T19:15:43.038311263Z","level":"INFO","msg":"sender: started","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.280634788Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-08-16T20:03:22.52020952Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-08-16T20:03:22.525542581Z","level":"INFO","msg":"stream: closing","id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.525560664Z","level":"INFO","msg":"handler: closed","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.526744074Z","level":"INFO","msg":"sender: closed","stream_id":"3mjkolf3"}
+{"time":"2026-08-16T20:03:22.526758551Z","level":"INFO","msg":"stream: closed","id":"3mjkolf3"}
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug.log b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..d92fa14ef50d0e9b596849c58579b0499c711efb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug.log
@@ -0,0 +1,26 @@
+2026-08-16 19:15:42,504 INFO    MainThread:94 [wandb_setup.py:_flush():81] Current SDK version is 0.22.3
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Configure stats pid to 94
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /home/hantzhan/.config/wandb/settings
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/wandb/settings
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():706] Logging user logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug.log
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:setup_run_log_directory():707] Logging internal logs to /mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/logs/debug-internal.log
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:init():833] calling init triggers
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:init():838] wandb.init called with sweep_config: {}
+config: {'mode': 'train', 'out_dir': '/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/runs/plain/side', 'paired_latents': '/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow/side/paired_latents_train.npz', 'image_view': 'side', 'seed': 20260526, 'val_fraction': 0.05, 'max_val': 128, 'val_overlap_train': True, 'shape_scale': 0.0, 'image_scale': 0.0, 'hidden_size': 512, 'depth': 12, 'num_heads': 8, 'head_dim': 64, 'mlp_ratio': 4.0, 'dropout': 0.0, 'pos_embed': '2d', 'attention_backend': 'sdpa', 'gradient_checkpointing': False, 'cond_dropout': 0.0, 'batch_size': 128, 'num_workers': 2, 'steps': 10000, 'lr': 0.0001, 'weight_decay': 0.01, 'log_every': 100, 'eval_every': 500, 'eval_noise_seed': 20260712, 'save_every': 5000, 'resume': True, 'sample_count': 8, 'sample_steps': 50, 'sample_seed': 20260712, 'sample_checkpoint': 'model_best.pt', 'guidance_scale': 1.0, 'decode_mesh': False, 'shape_resolution': 512, 'shape_max_batch': 32768, 'wandb': True, 'wandb_project': 'drivaer-qwen3v-input-cond', 'wandb_entity': 'kanydao-epfl', 'wandb_run_name': 'drivaer3v-inputcond-plain-side-0816c', 'wandb_tags': 'drivaer,qwen3v,input-conditional,channel-concat,side', 'wandb_mode': 'online', '_wandb': {}}
+2026-08-16 19:15:42,505 INFO    MainThread:94 [wandb_init.py:init():881] starting backend
+2026-08-16 19:15:42,714 INFO    MainThread:94 [wandb_init.py:init():884] sending inform_init request
+2026-08-16 19:15:42,719 INFO    MainThread:94 [wandb_init.py:init():892] backend started and connected
+2026-08-16 19:15:42,720 INFO    MainThread:94 [wandb_init.py:init():962] updated telemetry
+2026-08-16 19:15:42,735 INFO    MainThread:94 [wandb_init.py:init():986] communicating run to backend with 90.0 second timeout
+2026-08-16 19:15:43,362 INFO    MainThread:94 [wandb_init.py:init():1033] starting run threads in backend
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_console_start():2506] atexit reg
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_redirect():2354] redirect: wrap_raw
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_redirect():2423] Wrapping output streams.
+2026-08-16 19:15:43,827 INFO    MainThread:94 [wandb_run.py:_redirect():2446] Redirects installed.
+2026-08-16 19:15:43,832 INFO    MainThread:94 [wandb_init.py:init():1073] run started, returning control to user process
+2026-08-16 20:03:21,358 INFO    MainThread:94 [wandb_run.py:_finish():2272] finishing run kanydao-epfl/drivaer-qwen3v-input-cond/3mjkolf3
+2026-08-16 20:03:21,366 INFO    MainThread:94 [wandb_run.py:_atexit_cleanup():2471] got exitcode: 0
+2026-08-16 20:03:21,366 INFO    MainThread:94 [wandb_run.py:_restore():2453] restore
+2026-08-16 20:03:21,366 INFO    MainThread:94 [wandb_run.py:_restore():2459] restore done
+2026-08-16 20:03:22,523 INFO    MainThread:94 [wandb_run.py:_footer_sync_info():3835] logging synced files
diff --git a/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/run-3mjkolf3.wandb b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/run-3mjkolf3.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..e0ca6409958f4be72deea9a5ad03014f63f53b40
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/runs/plain/side/wandb/wandb/run-20260816_191542-3mjkolf3/run-3mjkolf3.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6d59129ec30a9d22e79d5a57cfb24daaed124c44ce9b5ec0b5f5fbaec12819f
+size 236671
diff --git a/anchorgen_baselines_20260923/input_cond/seiler_cond_backbone.py b/anchorgen_baselines_20260923/input_cond/seiler_cond_backbone.py
new file mode 100644
index 0000000000000000000000000000000000000000..e8f10278c57dca4a2eaa886c4289b62cb7d64b2d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/seiler_cond_backbone.py
@@ -0,0 +1,325 @@
+#!/usr/bin/env python3
+"""Seiler FlowMatchingVecSet backbone as an INPUT-CONDITIONAL flow.
+
+This is the ControlNet-style *simple channel-concat* counterpart of the
+unconditional joint model in ``seiler_joint_backbone.SeilerJointFlowTransformer``.
+
+Difference from the joint model, and ONLY this:
+  * The joint model diffuses ``[z_shape(256,8) | z_sketch(256,8)]`` as one 512-token
+    state and denoises BOTH segments (co-generation), then image->shape control is
+    an inverse/guidance operation.
+  * This model diffuses ONLY ``z_shape(256,8)`` and receives the clean sketch latent
+    ``z_sketch(256,8)`` concatenated on the CHANNEL axis of the input tokens
+    (``proj_in`` maps 8+8=16 -> hidden).  It directly learns ``p(z_shape | z_sketch)``.
+
+Everything else is kept identical to Seiler so the two models are comparable:
+the transformer block topology (self-attn + native context-attn per block), 8x64
+attention at width 512, GEGLU, three AdaLayerNorms, LayerScale, the timestep
+embedding, the native ``null_emb`` context, and the zero-initialised output
+projection.  The building blocks below are copied verbatim from
+``seiler_joint_backbone.py`` so this file is self-contained under ``input_cond/``.
+"""
+
+from __future__ import annotations
+
+import math
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.utils.checkpoint import checkpoint
+
+
+class PositionalEmbedding(nn.Module):
+    def __init__(self, num_channels: int, max_positions: int = 10000, endpoint: bool = False):
+        super().__init__()
+        self.num_channels = num_channels
+        self.max_positions = max_positions
+        self.endpoint = endpoint
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        freqs = torch.arange(
+            0,
+            self.num_channels // 2,
+            dtype=torch.float32,
+            device=x.device,
+        )
+        freqs = freqs / (self.num_channels // 2 - (1 if self.endpoint else 0))
+        freqs = (1 / self.max_positions) ** freqs
+        phases = x.float()[:, None] * freqs[None]
+        return torch.cat([phases.cos(), phases.sin()], dim=1).to(x.dtype)
+
+
+class CrossAttention(nn.Module):
+    def __init__(
+        self,
+        query_dim: int,
+        context_dim: int | None = None,
+        heads: int = 8,
+        dim_head: int = 64,
+        dropout: float = 0.0,
+        backend: str = "sdpa",
+    ):
+        super().__init__()
+        context_dim = query_dim if context_dim is None else context_dim
+        inner_dim = heads * dim_head
+        self.heads = heads
+        self.dim_head = dim_head
+        self.scale = dim_head**-0.5
+        self.dropout = dropout
+        self.backend = backend
+        self.to_q = nn.Linear(query_dim, inner_dim, bias=False)
+        self.to_k = nn.Linear(context_dim, inner_dim, bias=False)
+        self.to_v = nn.Linear(context_dim, inner_dim, bias=False)
+        self.to_out = nn.Sequential(nn.Linear(inner_dim, query_dim), nn.Dropout(dropout))
+
+    def forward(self, x: torch.Tensor, context: torch.Tensor | None = None) -> torch.Tensor:
+        context = x if context is None else context
+        batch = x.shape[0]
+        q = self.to_q(x).view(batch, -1, self.heads, self.dim_head).transpose(1, 2)
+        k = self.to_k(context).view(batch, -1, self.heads, self.dim_head).transpose(1, 2)
+        v = self.to_v(context).view(batch, -1, self.heads, self.dim_head).transpose(1, 2)
+        if self.backend == "sdpa":
+            out = F.scaled_dot_product_attention(
+                q,
+                k,
+                v,
+                dropout_p=self.dropout if self.training else 0.0,
+            )
+        elif self.backend == "einsum":
+            scores = torch.einsum("bhid,bhjd->bhij", q, k) * self.scale
+            weights = scores.softmax(dim=-1)
+            out = torch.einsum("bhij,bhjd->bhid", weights, v)
+        else:
+            raise ValueError(f"Unsupported attention backend {self.backend!r}")
+        out = out.transpose(1, 2).contiguous().view(batch, x.shape[1], -1)
+        return self.to_out(out)
+
+
+class LayerScale(nn.Module):
+    def __init__(self, dim: int, init_values: float = 1e-5):
+        super().__init__()
+        self.gamma = nn.Parameter(init_values * torch.ones(dim))
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return x * self.gamma
+
+
+class GEGLU(nn.Module):
+    def __init__(self, dim_in: int, dim_out: int):
+        super().__init__()
+        self.proj = nn.Linear(dim_in, dim_out * 2)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        value, gate = self.proj(x).chunk(2, dim=-1)
+        return value * F.gelu(gate)
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim: int, mult: float = 4.0, dropout: float = 0.0):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        self.net = nn.Sequential(
+            GEGLU(dim, inner_dim),
+            nn.Dropout(dropout),
+            nn.Linear(inner_dim, dim),
+        )
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return self.net(x)
+
+
+class AdaLayerNorm(nn.Module):
+    def __init__(self, dim: int):
+        super().__init__()
+        self.linear = nn.Linear(dim, dim * 2)
+        self.layernorm = nn.LayerNorm(dim, elementwise_affine=False)
+
+    def forward(self, x: torch.Tensor, timestep: torch.Tensor) -> torch.Tensor:
+        scale, shift = self.linear(timestep).chunk(2, dim=2)
+        return self.layernorm(x) * (1 + scale) + shift
+
+
+class SeilerTransformerBlock(nn.Module):
+    def __init__(
+        self,
+        dim: int,
+        heads: int,
+        dim_head: int,
+        mlp_ratio: float,
+        dropout: float,
+        attention_backend: str,
+    ):
+        super().__init__()
+        self.attn1 = CrossAttention(dim, heads=heads, dim_head=dim_head, dropout=dropout, backend=attention_backend)
+        self.attn2 = CrossAttention(
+            dim,
+            context_dim=dim,
+            heads=heads,
+            dim_head=dim_head,
+            dropout=dropout,
+            backend=attention_backend,
+        )
+        self.ff = FeedForward(dim, mult=mlp_ratio, dropout=dropout)
+        self.norm1 = AdaLayerNorm(dim)
+        self.norm2 = AdaLayerNorm(dim)
+        self.norm3 = AdaLayerNorm(dim)
+        self.ls1 = LayerScale(dim)
+        self.ls2 = LayerScale(dim)
+        self.ls3 = LayerScale(dim)
+
+    def forward(self, x: torch.Tensor, timestep: torch.Tensor, context: torch.Tensor) -> torch.Tensor:
+        x = x + self.ls1(self.attn1(self.norm1(x, timestep)))
+        x = x + self.ls2(self.attn2(self.norm2(x, timestep), context=context))
+        x = x + self.ls3(self.ff(self.norm3(x, timestep)))
+        return x
+
+
+def _one_dim_sincos(embed_dim: int, positions: torch.Tensor) -> torch.Tensor:
+    half = embed_dim // 2
+    frequencies = torch.exp(
+        torch.arange(half, dtype=torch.float32) * (-math.log(10000.0) / max(half - 1, 1))
+    )
+    values = positions.float()[:, None] * frequencies[None]
+    embedding = torch.cat([values.sin(), values.cos()], dim=-1)
+    if embedding.shape[-1] < embed_dim:
+        embedding = F.pad(embedding, (0, embed_dim - embedding.shape[-1]))
+    return embedding[:, :embed_dim]
+
+
+def sequence_1d_sincos(embed_dim: int, tokens: int) -> torch.Tensor:
+    return _one_dim_sincos(embed_dim, torch.arange(tokens))
+
+
+def image_2d_sincos(embed_dim: int, tokens: int) -> torch.Tensor:
+    grid = int(round(math.sqrt(tokens)))
+    if grid * grid != tokens:
+        raise ValueError(f"Image token count {tokens} is not square")
+    y, x = torch.meshgrid(torch.arange(grid), torch.arange(grid), indexing="ij")
+    half = embed_dim // 2
+    if half % 2:
+        half -= 1
+    embedding = torch.cat(
+        [
+            _one_dim_sincos(half, y.reshape(-1)),
+            _one_dim_sincos(half, x.reshape(-1)),
+        ],
+        dim=-1,
+    )
+    if embedding.shape[-1] < embed_dim:
+        embedding = F.pad(embedding, (0, embed_dim - embedding.shape[-1]))
+    return embedding[:, :embed_dim]
+
+
+class SeilerCondFlowTransformer(nn.Module):
+    """Input-conditional flow: diffuse z_shape(256,8), condition on clean
+    z_sketch(256,8) via channel concat (proj_in maps 16 -> hidden)."""
+
+    architecture_name = "seiler_flow_matching_vecset_cond_channelconcat"
+
+    def __init__(
+        self,
+        hidden_size: int = 512,
+        depth: int = 12,
+        num_heads: int = 8,
+        head_dim: int = 64,
+        mlp_ratio: float = 4.0,
+        dropout: float = 0.0,
+        tokens: int = 256,
+        channels: int = 8,
+        cond_channels: int = 8,
+        pos_embed: str = "2d",
+        attention_backend: str = "sdpa",
+        gradient_checkpointing: bool = False,
+    ):
+        super().__init__()
+        if hidden_size != num_heads * head_dim:
+            raise ValueError(f"hidden_size must equal num_heads*head_dim: {hidden_size} != {num_heads}*{head_dim}")
+        self.tokens = tokens
+        self.channels = channels
+        self.cond_channels = cond_channels
+        self.hidden_size = hidden_size
+        self.depth = depth
+        self.num_heads = num_heads
+        self.head_dim = head_dim
+        self.mlp_ratio = mlp_ratio
+        self.dropout = dropout
+        self.pos_embed = pos_embed
+        self.attention_backend = attention_backend
+        self.gradient_checkpointing = gradient_checkpointing
+
+        # The ONLY structural change vs the joint model: input channels = 8 (noisy
+        # shape) + 8 (clean sketch condition) = 16.  ControlNet-style simple concat.
+        self.proj_in = nn.Linear(channels + cond_channels, hidden_size, bias=False)
+        self.map_noise = PositionalEmbedding(256)
+        self.map_layer0 = nn.Linear(256, hidden_size)
+        self.map_layer1 = nn.Linear(hidden_size, hidden_size)
+        # Single learned token-type embedding (the joint model used two, one per
+        # modality; here there is a single diffused modality: shape).
+        self.token_type = nn.Parameter(torch.randn(1, hidden_size) * 0.02)
+        # Keep Seiler's native cond_dim=1 context parameterization.  The flow is
+        # conditioned through the concatenated input channels, not through this
+        # attention context, so forward uses null_emb exactly like the joint model.
+        self.cond_proj = nn.Linear(1, hidden_size)
+        self.null_emb = nn.Parameter(torch.randn(1, 1, hidden_size))
+
+        if pos_embed == "2d":
+            token_pos = image_2d_sincos(hidden_size, tokens)
+        elif pos_embed == "1d":
+            token_pos = sequence_1d_sincos(hidden_size, tokens)
+        elif pos_embed == "none":
+            token_pos = torch.zeros(tokens, hidden_size)
+        else:
+            raise ValueError(f"Unsupported pos_embed={pos_embed!r}")
+        self.register_buffer("token_pos", token_pos.unsqueeze(0), persistent=False)
+
+        self.transformer_blocks = nn.ModuleList(
+            [
+                SeilerTransformerBlock(
+                    hidden_size,
+                    num_heads,
+                    head_dim,
+                    mlp_ratio,
+                    dropout,
+                    attention_backend,
+                )
+                for _ in range(depth)
+            ]
+        )
+        self.norm = nn.LayerNorm(hidden_size)
+        self.proj_out = nn.Linear(hidden_size, channels, bias=False)
+        nn.init.zeros_(self.proj_out.weight)
+
+    def forward(self, x: torch.Tensor, t: torch.Tensor, cond: torch.Tensor) -> torch.Tensor:
+        """x: noisy shape latent (B, tokens, channels).
+        t: (B,) flow time in [0,1].
+        cond: clean sketch latent (B, tokens, cond_channels).  Pass zeros for the
+        unconditional/CFG branch."""
+        if x.ndim != 3 or tuple(x.shape[1:]) != (self.tokens, self.channels):
+            raise ValueError(f"Expected shape tokens (B,{self.tokens},{self.channels}), got {tuple(x.shape)}")
+        if cond is None:
+            cond = x.new_zeros((x.shape[0], self.tokens, self.cond_channels))
+        if tuple(cond.shape[1:]) != (self.tokens, self.cond_channels):
+            raise ValueError(f"Expected cond tokens (B,{self.tokens},{self.cond_channels}), got {tuple(cond.shape)}")
+        timestep = self.map_noise(t)[:, None]
+        timestep = F.silu(self.map_layer0(timestep))
+        timestep = F.silu(self.map_layer1(timestep))
+        joint_in = torch.cat([x, cond.to(x.dtype)], dim=-1)
+        hidden = self.proj_in(joint_in) + self.token_type.unsqueeze(0).to(x.dtype) + self.token_pos.to(x.dtype)
+        context = self.null_emb.expand(x.shape[0], -1, -1).to(x.dtype)
+        for block in self.transformer_blocks:
+            if self.gradient_checkpointing and self.training and torch.is_grad_enabled():
+                hidden = checkpoint(block, hidden, timestep, context, use_reentrant=False)
+            else:
+                hidden = block(hidden, timestep, context)
+        return self.proj_out(self.norm(hidden))
+
+
+def parameter_count(model: nn.Module) -> int:
+    return sum(parameter.numel() for parameter in model.parameters())
+
+
+# For reference/comparison with the joint model (82,317,824).  The conditional
+# model differs by: proj_in 16->512 instead of 8->512 (+8*512), single token_type
+# instead of 2 modality rows (-512).  Net +3584 params; computed exactly at init.
+SEILER_REFERENCE_PARAMETER_COUNT = 82_316_800
diff --git a/anchorgen_baselines_20260923/input_cond/submit_input_cond_3view.sh b/anchorgen_baselines_20260923/input_cond/submit_input_cond_3view.sh
new file mode 100644
index 0000000000000000000000000000000000000000..cbc198b0e6f9e26b990623c921c6045c30661b59
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/submit_input_cond_3view.sh
@@ -0,0 +1,61 @@
+#!/usr/bin/env bash
+# Submit input-conditional flow training jobs to RunAI, 1 GPU each, reusing the
+# frozen paired latents.  Two comparison FAMILIES:
+#   plain -> cond_dropout 0.0  (pure conditional baseline, NO CFG)   runs/plain/<view>
+#   cfg   -> cond_dropout 0.1  (CFG-capable, guidance-scale sweep)   runs/cfg/<view>
+# jiaohu env contract + cuda11.8 image + cvlab-scratch PVC (same as vecset/joint).
+#
+# Memory-first scheduling: query live pools BEFORE running and set POOL.
+#   POOL=a100-40g ./submit_input_cond_3view.sh                     # both families, 3 views = 6 jobs
+#   POOL=a100-40g FAMILIES="plain" VIEWS="side" ./submit_input_cond_3view.sh
+set -euo pipefail
+
+CD=/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design
+ROOT=$CD/input_cond
+JF=/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/seiler_format/joint_flow
+JIAOHU=$CD/fcso/submit_scripts/jiaohu_checked.py
+
+PROJECT=cvlab-hantzhan
+IMAGE=ic-registry.epfl.ch/cvlab/haoqi/cuda11.8-cudnn8-ubuntu22.04
+POOL=${POOL:-a100-40g}
+STAMP=${STAMP:-0816c}
+VIEWS=${VIEWS:-"side front rear"}
+FAMILIES=${FAMILIES:-"plain cfg"}
+
+drop_for() { case "$1" in plain) echo 0.0;; cfg) echo 0.1;; *) echo 0.0;; esac; }
+
+test -f "$JIAOHU" || { echo "MISSING jiaohu_checked.py: $JIAOHU" >&2; exit 2; }
+test -f "$ROOT/train_cond_flow.py" || { echo "MISSING code in $ROOT" >&2; exit 2; }
+
+for FAM in $FAMILIES; do
+  DROP=$(drop_for "$FAM")
+  for VIEW in $VIEWS; do
+    PAIRED=$JF/$VIEW/paired_latents_train.npz
+    OUT=$ROOT/runs/$FAM/$VIEW
+    JOB=inputcond-$FAM-$VIEW-$POOL-$STAMP
+    RUNNAME=drivaer3v-inputcond-$FAM-$VIEW-$STAMP
+    test -s "$PAIRED" || { echo "MISSING paired latents: $PAIRED" >&2; exit 2; }
+    mkdir -p "$OUT"
+
+    # Keep RUN_CMD trivial; all logic in pod_entrypoint.sh (resume-safe).
+    RUN_CMD="set -euo pipefail; cd ${ROOT}; test \"\$(pwd -P)\" = \"${ROOT}\"; bash pod_entrypoint.sh ${VIEW} ${PAIRED} ${OUT} ${DROP} ${RUNNAME}"
+
+    echo "== submit $JOB  (family=$FAM drop=$DROP view=$VIEW pool=$POOL)"
+    runai submit "$JOB" -p "$PROJECT" \
+      -i "$IMAGE" \
+      --gpu 1 \
+      --pvc cvlab-scratch:/mnt/cvlab/scratch \
+      --large-shm \
+      --node-pools "$POOL" \
+      -e CLUSTER_USER=hantzhan \
+      -e CLUSTER_USER_ID=306442 \
+      -e CLUSTER_GROUP_NAME=CVLAB-StaffU \
+      -e CLUSTER_GROUP_ID=11166 \
+      -e ENV=dora \
+      -e RUN_CMD="$RUN_CMD" \
+      --command -- python "$JIAOHU"
+  done
+done
+
+echo "== submitted families=[$FAMILIES] views=[$VIEWS] stamp=$STAMP"
+echo "== monitor: runai list jobs -p $PROJECT | grep inputcond-.*$STAMP"
diff --git a/anchorgen_baselines_20260923/input_cond/test_cond_flow.py b/anchorgen_baselines_20260923/input_cond/test_cond_flow.py
new file mode 100644
index 0000000000000000000000000000000000000000..a54ad06a18d7bb3a727dab7f2c54a27df494d43a
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/test_cond_flow.py
@@ -0,0 +1,86 @@
+#!/usr/bin/env python3
+"""CPU smoke test for the input-conditional flow (no GPU, no submit).
+
+Validates, on the REAL side paired latents:
+  1. backbone forward: (B,256,8)+(B,256,8) cond -> (B,256,8) velocity;
+  2. a few conditional flow-matching steps reduce the loss;
+  3. euler_sample_shape produces finite z_shape from a given sketch latent.
+
+Run (login node CPU is fine):
+    $DORA test_cond_flow.py [--paired-latents <npz>]
+"""
+from __future__ import annotations
+
+import argparse
+from pathlib import Path
+
+import numpy as np
+import torch
+
+from seiler_cond_backbone import SeilerCondFlowTransformer, parameter_count
+import train_cond_flow as tcf
+
+DEFAULT_PAIRED = (
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/"
+    "drivaernet_qwen3v_joint_training_registered_20260713/full_registered_v1/"
+    "seiler_format/joint_flow/side/paired_latents_train.npz"
+)
+
+
+def main() -> None:
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--paired-latents", default=DEFAULT_PAIRED)
+    ap.add_argument("--records", type=int, default=48)
+    ap.add_argument("--steps", type=int, default=40)
+    args = ap.parse_args()
+
+    torch.manual_seed(0)
+    device = torch.device("cpu")
+
+    # 1. forward shape contract
+    model = SeilerCondFlowTransformer(gradient_checkpointing=False).to(device)
+    n = parameter_count(model)
+    x = torch.randn(4, 256, 8)
+    cond = torch.randn(4, 256, 8)
+    t = torch.rand(4)
+    v = model(x, t, cond)
+    assert v.shape == (4, 256, 8), v.shape
+    assert torch.isfinite(v).all()
+    print(f"[1] forward OK  params={n:,}  out={tuple(v.shape)}")
+
+    # 2. a few conditional training steps on real latents
+    z = np.load(args.paired_latents, allow_pickle=True)
+    zs = torch.from_numpy(z["z_shape"][: args.records].astype(np.float32))
+    zi = torch.from_numpy(z["z_img"][: args.records].astype(np.float32))
+    sscale = 1.0 / float(zs.std())
+    iscale = 1.0 / float(zi.std())
+    zs = zs * sscale
+    zi = zi * iscale
+    opt = torch.optim.AdamW(model.parameters(), lr=1e-3)
+    first = last = None
+    for s in range(args.steps):
+        loss = tcf.cond_flow_loss(model, zs, zi, cond_dropout=0.0)
+        opt.zero_grad(set_to_none=True)
+        loss.backward()
+        torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
+        opt.step()
+        if s == 0:
+            first = float(loss)
+        last = float(loss)
+    assert np.isfinite(last)
+    print(f"[2] train OK  loss {first:.4f} -> {last:.4f}  (drop {100*(1-last/first):.1f}%)")
+    assert last < first, "loss did not decrease"
+
+    # 3. conditional sampling from a given sketch latent
+    model.eval()
+    z_shape_scaled = tcf.euler_sample_shape(model, zi[:2], steps=10, seed=0, guidance_scale=1.0)
+    z_shape = z_shape_scaled / sscale
+    assert z_shape.shape == (2, 256, 8)
+    assert torch.isfinite(z_shape).all()
+    print(f"[3] sample OK  z_shape={tuple(z_shape.shape)}  std={float(z_shape.std()):.4f}")
+
+    print("SMOKE PASS")
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/train_cond_flow.py b/anchorgen_baselines_20260923/input_cond/train_cond_flow.py
new file mode 100644
index 0000000000000000000000000000000000000000..ba71d4203b5a5de35e462e082efbd58e993184a7
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/train_cond_flow.py
@@ -0,0 +1,565 @@
+#!/usr/bin/env python3
+"""Train an INPUT-CONDITIONAL Seiler flow over car shape latents.
+
+Data contract (reused verbatim from the joint pipeline, no re-preparation):
+  z_shape : Seiler VecSetVAE latent tokens, (N, 256, 8)   <- diffused target
+  z_img   : SD-VAE + bottleneck sketch latent, (N, 256, 8) <- CLEAN condition
+
+Flow-matching objective (same as the joint model, but shape-only and conditional):
+  x0 ~ N(0, I), t ~ U(0, 1)
+  xt = (1 - t) * x0 + t * z_shape
+  u  = z_shape - x0
+  loss = MSE( v_theta(xt, t, cond=z_sketch), u )
+
+The sketch enters the model as extra INPUT CHANNELS (ControlNet-style simple
+concat): the backbone sees ``[xt(256,8) | z_sketch(256,8)] = (256,16)`` and
+predicts the (256,8) shape velocity.  This directly learns ``p(z_shape|z_sketch)``.
+
+Everything else (AdamW lr1e-4/wd0.01, cosine-to-5%, grad-clip 1.0, batch 128,
+10k steps, val-every-500 best checkpoint, scale = 1/train_std) is copied from
+``train_joint_flow.py`` so the conditional and joint models are comparable.
+
+Modes:
+  train   -- train the conditional flow, write model_best.pt / model_last.pt.
+  sample  -- given stored sketch latents (val split by default), Euler-sample
+             z_shape and save cond_samples.pt (unscaled z_shape + z_img + uids).
+             Add --decode-mesh to also decode R512 meshes via the frozen VecSet
+             decoder (best-effort import of the sketchonly worktree helpers).
+"""
+
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+import math
+import os
+import random
+import subprocess
+import sys
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Iterable
+
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.utils.data import DataLoader, Dataset
+
+from seiler_cond_backbone import SeilerCondFlowTransformer, parameter_count
+
+SHAPE_TOKENS = 256
+CHANNELS = 8
+VIEWS = ("front", "side", "rear")
+
+# Frozen VecSet shape decoder (only needed for --decode-mesh in sample mode).
+SKETCHONLY_WORKTREE = (
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/"
+    "pairwise-diffusion-drivaer3v-sketchonly-0713h/experiments/seiler_joint_diffusion"
+)
+SEILER_EXPDIR = (
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/outputs/customized-design/"
+    "drivaernet_qwen3v_joint_training_clean_20260712/seiler_format/shape_codec/"
+    "vecset_vae_256_8_customized"
+)
+SMOOTHSDF_ROOT = (
+    "/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/seiler_joint_anchor/"
+    "third_party/seiler_smoothsdf"
+)
+
+
+def parse_args() -> argparse.Namespace:
+    p = argparse.ArgumentParser()
+    p.add_argument("--mode", choices=["train", "sample"], default="train")
+    p.add_argument("--out-dir", required=True)
+    p.add_argument("--paired-latents", required=True,
+                   help="Path to the existing paired_latents_train.npz to REUSE (z_shape + z_img).")
+    p.add_argument("--image-view", choices=VIEWS, required=True)
+    # split / scaling (kept identical to the joint run)
+    p.add_argument("--seed", type=int, default=20260526)
+    p.add_argument("--val-fraction", type=float, default=0.05)
+    p.add_argument("--max-val", type=int, default=128)
+    p.add_argument("--val-overlap-train", action=argparse.BooleanOptionalAction, default=True)
+    p.add_argument("--shape-scale", type=float, default=0.0, help="0 means 1/train_std.")
+    p.add_argument("--image-scale", type=float, default=0.0, help="0 means 1/train_std.")
+    # model
+    p.add_argument("--hidden-size", type=int, default=512)
+    p.add_argument("--depth", type=int, default=12)
+    p.add_argument("--num-heads", type=int, default=8)
+    p.add_argument("--head-dim", type=int, default=64)
+    p.add_argument("--mlp-ratio", type=float, default=4.0)
+    p.add_argument("--dropout", type=float, default=0.0)
+    p.add_argument("--pos-embed", choices=["none", "1d", "2d"], default="2d")
+    p.add_argument("--attention-backend", choices=["sdpa", "einsum"], default="sdpa")
+    p.add_argument("--gradient-checkpointing", action=argparse.BooleanOptionalAction, default=False)
+    # conditioning
+    p.add_argument("--cond-dropout", type=float, default=0.0,
+                   help="Prob. of replacing the sketch condition with zeros (enables CFG later).")
+    # optimization
+    p.add_argument("--batch-size", type=int, default=128)
+    p.add_argument("--num-workers", type=int, default=2)
+    p.add_argument("--steps", type=int, default=10000)
+    p.add_argument("--lr", type=float, default=1e-4)
+    p.add_argument("--weight-decay", type=float, default=0.01)
+    p.add_argument("--log-every", type=int, default=100)
+    p.add_argument("--eval-every", type=int, default=500)
+    p.add_argument("--eval-noise-seed", type=int, default=20260712)
+    p.add_argument("--save-every", type=int, default=5000)
+    p.add_argument("--resume", action=argparse.BooleanOptionalAction, default=True,
+                   help="Resume from the newest model_step_*.pt in out-dir if present.")
+    # sampling
+    p.add_argument("--sample-count", type=int, default=8)
+    p.add_argument("--sample-steps", type=int, default=50)
+    p.add_argument("--sample-seed", type=int, default=20260712)
+    p.add_argument("--sample-checkpoint", default="model_best.pt")
+    p.add_argument("--guidance-scale", type=float, default=1.0, help="CFG scale (1.0 = plain conditional).")
+    p.add_argument("--decode-mesh", action="store_true", help="Decode sampled z_shape to R512 meshes.")
+    p.add_argument("--shape-resolution", type=int, default=512)
+    p.add_argument("--shape-max-batch", type=int, default=32768)
+    # wandb
+    p.add_argument("--wandb", action="store_true")
+    p.add_argument("--wandb-project", default=os.environ.get("WANDB_PROJECT", ""))
+    p.add_argument("--wandb-entity", default=os.environ.get("WANDB_ENTITY", ""))
+    p.add_argument("--wandb-run-name", default=os.environ.get("WANDB_RUN_NAME", ""))
+    p.add_argument("--wandb-tags", default=os.environ.get("WANDB_TAGS", ""))
+    p.add_argument("--wandb-mode", default=os.environ.get("WANDB_MODE", ""))
+    return p.parse_args()
+
+
+def set_seed(seed: int) -> None:
+    random.seed(seed)
+    np.random.seed(seed)
+    torch.manual_seed(seed)
+    torch.cuda.manual_seed_all(seed)
+
+
+def sha256_file(path: Path) -> str:
+    digest = hashlib.sha256()
+    with path.open("rb") as handle:
+        for chunk in iter(lambda: handle.read(1024 * 1024), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def git_provenance() -> dict[str, str]:
+    here = Path(__file__).resolve()
+
+    def run(*args: str) -> str:
+        try:
+            return subprocess.check_output(["git", "-C", str(here.parent), *args], text=True).strip()
+        except Exception:
+            return ""
+
+    return {
+        "file": str(here),
+        "commit": run("rev-parse", "HEAD"),
+        "branch": run("branch", "--show-current"),
+        "train_cond_flow_sha256": sha256_file(here),
+        "seiler_cond_backbone_sha256": sha256_file(here.with_name("seiler_cond_backbone.py")),
+    }
+
+
+def init_wandb(args: argparse.Namespace, out_dir: Path):
+    if not args.wandb:
+        return None
+    try:
+        import wandb
+    except Exception as exc:  # pragma: no cover - soft-fail so training never dies on wandb
+        print(f"[wandb] disabled (import failed: {exc})", flush=True)
+        return None
+    project = args.wandb_project or "drivaer-qwen3v-input-cond"
+    entity = args.wandb_entity or None
+    run_name = args.wandb_run_name or out_dir.name
+    tags = [tag.strip() for tag in args.wandb_tags.split(",") if tag.strip()]
+    mode = args.wandb_mode or None
+    wandb_dir = out_dir / "wandb"
+    wandb_dir.mkdir(parents=True, exist_ok=True)
+    try:
+        run = wandb.init(project=project, entity=entity, name=run_name, tags=tags or None,
+                         mode=mode, dir=str(wandb_dir), config=vars(args))
+        print(f"[wandb] project={project} run={run_name} mode={mode}", flush=True)
+        return run
+    except Exception as exc:  # pragma: no cover
+        print(f"[wandb] disabled (init failed: {exc})", flush=True)
+        return None
+
+
+# --------------------------------------------------------------------------- #
+# Data
+# --------------------------------------------------------------------------- #
+class PairedLatentDataset(Dataset):
+    def __init__(self, z_shape: np.ndarray, z_img: np.ndarray, indices: Iterable[int]):
+        self.z_shape = z_shape
+        self.z_img = z_img
+        self.indices = np.asarray(list(indices), dtype=np.int64)
+
+    def __len__(self) -> int:
+        return int(len(self.indices))
+
+    def __getitem__(self, idx: int) -> dict[str, torch.Tensor]:
+        i = int(self.indices[idx])
+        return {
+            "shape": torch.from_numpy(self.z_shape[i]),
+            "image": torch.from_numpy(self.z_img[i]),
+        }
+
+
+@dataclass
+class SplitData:
+    z_shape: np.ndarray
+    z_img: np.ndarray
+    ids: np.ndarray
+    train_idx: np.ndarray
+    val_idx: np.ndarray
+    shape_scale: float
+    image_scale: float
+
+
+def load_split_data(args: argparse.Namespace) -> SplitData:
+    path = Path(args.paired_latents)
+    if not path.exists():
+        raise FileNotFoundError(f"Missing paired dataset: {path}")
+    data = np.load(path, allow_pickle=True)
+    ids = data["ids"].astype(str)
+    z_shape = data["z_shape"].astype(np.float32)
+    z_img = data["z_img"].astype(np.float32)
+    stored_view = str(data["image_view"].item()) if "image_view" in data.files else "legacy"
+    if stored_view != args.image_view:
+        raise ValueError(f"Paired cache view mismatch: stored={stored_view!r}, requested={args.image_view!r}")
+    if z_shape.shape != (len(ids), SHAPE_TOKENS, CHANNELS):
+        raise ValueError(f"Bad cached shape latent tensor: {z_shape.shape}")
+    if z_img.shape != (len(ids), SHAPE_TOKENS, CHANNELS):
+        raise ValueError(f"Bad cached image latent tensor: {z_img.shape}")
+    if not np.isfinite(z_shape).all() or not np.isfinite(z_img).all():
+        raise ValueError(f"Non-finite values in paired cache {path}")
+
+    rng = np.random.default_rng(args.seed)
+    order = rng.permutation(len(ids))
+    n_val = max(1, int(round(len(ids) * args.val_fraction))) if len(ids) > 1 else 0
+    if args.max_val and args.max_val > 0:
+        n_val = min(n_val, args.max_val)
+    val_idx = order[:n_val]
+    train_idx = order.copy() if args.val_overlap_train else order[n_val:]
+    if len(train_idx) == 0:
+        raise RuntimeError("Empty train split.")
+
+    train_shape = z_shape[train_idx]
+    train_img = z_img[train_idx]
+    shape_scale = args.shape_scale if args.shape_scale > 0 else 1.0 / max(float(train_shape.std()), 1e-6)
+    image_scale = args.image_scale if args.image_scale > 0 else 1.0 / max(float(train_img.std()), 1e-6)
+    return SplitData(z_shape, z_img, ids, train_idx, val_idx, float(shape_scale), float(image_scale))
+
+
+def scaled_batch(batch: dict[str, torch.Tensor], sscale: float, iscale: float, device):
+    z_shape = batch["shape"].to(device, non_blocking=True).float() * sscale
+    z_img = batch["image"].to(device, non_blocking=True).float() * iscale
+    return z_shape, z_img
+
+
+# --------------------------------------------------------------------------- #
+# Model + flow loss
+# --------------------------------------------------------------------------- #
+def make_model(args: argparse.Namespace, device: torch.device) -> SeilerCondFlowTransformer:
+    model = SeilerCondFlowTransformer(
+        hidden_size=args.hidden_size,
+        depth=args.depth,
+        num_heads=args.num_heads,
+        head_dim=args.head_dim,
+        mlp_ratio=args.mlp_ratio,
+        dropout=args.dropout,
+        tokens=SHAPE_TOKENS,
+        channels=CHANNELS,
+        cond_channels=CHANNELS,
+        pos_embed=args.pos_embed,
+        attention_backend=args.attention_backend,
+        gradient_checkpointing=args.gradient_checkpointing,
+    ).to(device)
+    n = parameter_count(model)
+    print(f"[model] arch={SeilerCondFlowTransformer.architecture_name} params={n:,} "
+          f"hidden={args.hidden_size} depth={args.depth} heads={args.num_heads}x{args.head_dim} "
+          f"pos={args.pos_embed} cond=channel_concat(16->{args.hidden_size})", flush=True)
+    return model
+
+
+def cond_flow_loss(model, z_shape, z_img, cond_dropout: float = 0.0, generator=None):
+    bsz = z_shape.shape[0]
+    x0 = torch.randn(z_shape.shape, device=z_shape.device, dtype=z_shape.dtype, generator=generator)
+    t = torch.rand(bsz, device=z_shape.device, generator=generator)
+    view = (bsz,) + (1,) * (z_shape.ndim - 1)
+    xt = (1.0 - t.view(view)) * x0 + t.view(view) * z_shape
+    ut = z_shape - x0
+    cond = z_img
+    if cond_dropout > 0.0:
+        keep = (torch.rand(bsz, device=z_shape.device, generator=generator) >= cond_dropout).view(view)
+        cond = cond * keep.to(cond.dtype)
+    pred = model(xt, t, cond)
+    loss = ((pred - ut) ** 2).mean()
+    return loss
+
+
+@torch.no_grad()
+def evaluate(model, loader, data: SplitData, args, device, max_batches: int = 16) -> dict[str, float]:
+    if len(loader.dataset) == 0:
+        return {"shape": float("nan")}
+    model.eval()
+    gen = torch.Generator(device=device)
+    gen.manual_seed(args.eval_noise_seed)
+    vals = []
+    for bi, batch in enumerate(loader):
+        if bi >= max_batches:
+            break
+        z_shape, z_img = scaled_batch(batch, data.shape_scale, data.image_scale, device)
+        loss = cond_flow_loss(model, z_shape, z_img, cond_dropout=0.0, generator=gen)
+        vals.append(float(loss.cpu()))
+    return {"shape": float(np.mean(vals)) if vals else float("nan")}
+
+
+def checkpoint_payload(model, args, data: SplitData, step: int, val, opt=None, sched=None, best=None):
+    payload = {
+        "state_dict": model.state_dict(),
+        "step": step,
+        "val": val,
+        "best_val": best,
+        "model": {
+            "architecture": SeilerCondFlowTransformer.architecture_name,
+            "parameter_count": parameter_count(model),
+            "conditioning": "sketch_latent_channel_concat_controlnet_simple",
+            "hidden_size": args.hidden_size,
+            "depth": args.depth,
+            "num_heads": args.num_heads,
+            "head_dim": args.head_dim,
+            "mlp_ratio": args.mlp_ratio,
+            "dropout": args.dropout,
+            "pos_embed": args.pos_embed,
+            "tokens": SHAPE_TOKENS,
+            "channels": CHANNELS,
+            "cond_channels": CHANNELS,
+            "cond_dropout": args.cond_dropout,
+            "attention_backend": args.attention_backend,
+        },
+        "scales": {"shape_scale": data.shape_scale, "image_scale": data.image_scale},
+        "data_contract": {
+            "image_view": args.image_view,
+            "paired_latents": str(Path(args.paired_latents).resolve()),
+            "paired_latents_sha256": sha256_file(Path(args.paired_latents)),
+            "seed": args.seed,
+        },
+        "code_provenance": git_provenance(),
+    }
+    if opt is not None:
+        payload["optimizer"] = opt.state_dict()
+    if sched is not None:
+        payload["scheduler"] = sched.state_dict()
+    return payload
+
+
+def save_checkpoint(out_dir: Path, name: str, payload: dict) -> None:
+    torch.save(payload, out_dir / name)
+
+
+def find_resume_ckpt(out_dir: Path) -> Path | None:
+    cands = sorted(out_dir.glob("model_step_*.pt"))
+    return cands[-1] if cands else None
+
+
+def train(args: argparse.Namespace) -> None:
+    out_dir = Path(args.out_dir)
+    out_dir.mkdir(parents=True, exist_ok=True)
+    set_seed(args.seed)
+    data = load_split_data(args)
+    (out_dir / "run_config.json").write_text(json.dumps(vars(args), indent=2))
+    (out_dir / "code_provenance.json").write_text(json.dumps(git_provenance(), indent=2))
+
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    torch.backends.cuda.matmul.allow_tf32 = True
+    torch.backends.cudnn.allow_tf32 = True
+    try:
+        torch.set_float32_matmul_precision("medium")
+    except Exception:
+        pass
+    print(f"[train] device={device} view={args.image_view} train={len(data.train_idx)} val={len(data.val_idx)}", flush=True)
+    print(f"[train] scales shape={data.shape_scale:.6f} image={data.image_scale:.6f}", flush=True)
+
+    train_loader = DataLoader(
+        PairedLatentDataset(data.z_shape, data.z_img, data.train_idx),
+        batch_size=args.batch_size, shuffle=True, num_workers=args.num_workers,
+        drop_last=True, pin_memory=True,
+    )
+    val_loader = DataLoader(
+        PairedLatentDataset(data.z_shape, data.z_img, data.val_idx),
+        batch_size=args.batch_size, shuffle=False, num_workers=args.num_workers,
+        drop_last=False, pin_memory=True,
+    )
+    model = make_model(args, device)
+    opt = torch.optim.AdamW(model.parameters(), lr=args.lr, weight_decay=args.weight_decay)
+    sched = torch.optim.lr_scheduler.CosineAnnealingLR(opt, T_max=max(1, args.steps), eta_min=args.lr * 0.05)
+
+    start_step = 0
+    best = float("inf")
+    if args.resume:
+        ck = find_resume_ckpt(out_dir)
+        if ck is not None:
+            try:
+                blob = torch.load(ck, map_location=device)
+                model.load_state_dict(blob["state_dict"])
+                if "optimizer" in blob:
+                    opt.load_state_dict(blob["optimizer"])
+                if "scheduler" in blob:
+                    sched.load_state_dict(blob["scheduler"])
+                start_step = int(blob.get("step", 0))
+                if isinstance(blob.get("best_val"), (int, float)) and math.isfinite(blob["best_val"]):
+                    best = float(blob["best_val"])
+                print(f"[resume] loaded {ck.name} step={start_step} best={best:.6f}", flush=True)
+            except Exception as exc:
+                print(f"[resume] failed ({exc}); starting fresh", flush=True)
+                start_step = 0
+
+    wandb_run = init_wandb(args, out_dir)
+    history = []
+    t0 = time.time()
+    iterator = iter(train_loader)
+
+    for step in range(start_step + 1, args.steps + 1):
+        try:
+            batch = next(iterator)
+        except StopIteration:
+            iterator = iter(train_loader)
+            batch = next(iterator)
+        model.train()
+        z_shape, z_img = scaled_batch(batch, data.shape_scale, data.image_scale, device)
+        loss = cond_flow_loss(model, z_shape, z_img, cond_dropout=args.cond_dropout)
+        opt.zero_grad(set_to_none=True)
+        loss.backward()
+        torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
+        opt.step()
+        sched.step()
+
+        if step == 1 or step % args.log_every == 0:
+            row = {"step": step, "train_shape": float(loss.detach().cpu()),
+                   "lr": float(sched.get_last_lr()[0]), "elapsed_sec": time.time() - t0}
+            history.append(row)
+            print(f"[train] step={step:06d} loss={row['train_shape']:.6f} lr={row['lr']:.2e}", flush=True)
+            if wandb_run is not None:
+                wandb_run.log({"train/shape": row["train_shape"], "train/lr": row["lr"],
+                               "time/elapsed_sec": row["elapsed_sec"]}, step=step)
+
+        if step == 1 or step % args.eval_every == 0 or step == args.steps:
+            val = evaluate(model, val_loader, data, args, device)
+            print(f"[val] step={step:06d} shape={val['shape']:.6f}", flush=True)
+            if wandb_run is not None:
+                wandb_run.log({"val/shape": val["shape"]}, step=step)
+            if math.isfinite(val["shape"]) and val["shape"] < best:
+                best = val["shape"]
+                save_checkpoint(out_dir, "model_best.pt",
+                                checkpoint_payload(model, args, data, step, val, best=best))
+            history.append({"step": step, "val_shape": val["shape"]})
+
+        if step % args.save_every == 0:
+            save_checkpoint(out_dir, f"model_step_{step:06d}.pt",
+                            checkpoint_payload(model, args, data, step,
+                                               {"shape": float(loss.detach().cpu())},
+                                               opt=opt, sched=sched, best=best))
+
+    save_checkpoint(out_dir, "model_last.pt",
+                    checkpoint_payload(model, args, data, args.steps,
+                                       {"shape": float(loss.detach().cpu())},
+                                       opt=opt, sched=sched, best=best))
+    (out_dir / "train_history.json").write_text(json.dumps(history, indent=2))
+    summary = {
+        "image_view": args.image_view,
+        "best_val_shape": best,
+        "steps": args.steps,
+        "train_samples": int(len(data.train_idx)),
+        "val_samples": int(len(data.val_idx)),
+        "shape_scale": data.shape_scale,
+        "image_scale": data.image_scale,
+        "conditioning": "sketch_latent_channel_concat",
+    }
+    (out_dir / "summary.json").write_text(json.dumps(summary, indent=2))
+    if wandb_run is not None:
+        wandb_run.summary.update(summary)
+        wandb_run.finish()
+    print(f"[train] done best_val_shape={best:.6f} out={out_dir}", flush=True)
+
+
+# --------------------------------------------------------------------------- #
+# Conditional sampling (given sketch -> shape latent -> optional mesh)
+# --------------------------------------------------------------------------- #
+@torch.no_grad()
+def euler_sample_shape(model, z_sketch_scaled, steps: int, seed: int, guidance_scale: float = 1.0):
+    """z_sketch_scaled: (B,256,8) scaled sketch condition. Returns scaled z_shape (B,256,8)."""
+    device = z_sketch_scaled.device
+    gen = torch.Generator(device=device)
+    gen.manual_seed(seed)
+    x = torch.randn((z_sketch_scaled.shape[0], SHAPE_TOKENS, CHANNELS), device=device, generator=gen)
+    dt = 1.0 / steps
+    zero_cond = torch.zeros_like(z_sketch_scaled)
+    for i in range(steps):
+        t = torch.full((x.shape[0],), i * dt, device=device)
+        v = model(x, t, z_sketch_scaled)
+        if guidance_scale != 1.0:
+            v_u = model(x, t, zero_cond)
+            v = v_u + guidance_scale * (v - v_u)
+        x = x + v * dt
+    return x
+
+
+def sample(args: argparse.Namespace) -> None:
+    out_dir = Path(args.out_dir)
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    ckpt = torch.load(out_dir / args.sample_checkpoint, map_location=device)
+    scales = ckpt["scales"]
+    sscale, iscale = float(scales["shape_scale"]), float(scales["image_scale"])
+    model = make_model(args, device)
+    model.load_state_dict(ckpt["state_dict"])
+    model.eval()
+
+    data = load_split_data(args)
+    idx = data.val_idx[: args.sample_count]
+    if len(idx) == 0:
+        idx = data.train_idx[: args.sample_count]
+    uids = [str(data.ids[i]) for i in idx]
+    z_img = torch.from_numpy(data.z_img[idx]).float().to(device) * iscale
+
+    z_shape_scaled = euler_sample_shape(model, z_img, args.sample_steps, args.sample_seed, args.guidance_scale)
+    z_shape = (z_shape_scaled / sscale).cpu()
+    sample_dir = out_dir / "cond_samples"
+    sample_dir.mkdir(parents=True, exist_ok=True)
+    torch.save({"z_shape": z_shape, "z_img": torch.from_numpy(data.z_img[idx]).float(),
+                "uids": uids, "scales": scales, "guidance_scale": args.guidance_scale},
+               sample_dir / "cond_samples.pt")
+    print(f"[sample] view={args.image_view} n={len(uids)} -> {sample_dir/'cond_samples.pt'}", flush=True)
+    print(f"[sample] uids={uids}", flush=True)
+
+    if args.decode_mesh:
+        _decode_meshes(z_shape, uids, sample_dir, device, args)
+
+
+def _decode_meshes(z_shape, uids, sample_dir, device, args):
+    """Best-effort R512 decode of sampled shape latents via the frozen VecSet decoder."""
+    sys.path.insert(0, SKETCHONLY_WORKTREE)
+    from types import SimpleNamespace
+    import train_joint_flow as tjf  # noqa: E402
+    shape_dec = tjf.load_shape_decoder(
+        SimpleNamespace(smoothsdf_root=SMOOTHSDF_ROOT, seiler_expdir=SEILER_EXPDIR, load_epoch=2000), device)
+    for i, uid in enumerate(uids):
+        z = z_shape[i].to(device)
+        sdf = tjf.decode_sdf_grid(shape_dec, z, args.shape_resolution, args.shape_max_batch, device)
+        mesh, _ = tjf.sdf_grid_to_mesh(sdf, args.shape_resolution)
+        out = sample_dir / f"cond_sample_{i:02d}_{uid}.obj"
+        mesh.export(str(out))
+        print(f"[decode] {uid} -> {out}", flush=True)
+
+
+def main() -> None:
+    args = parse_args()
+    if args.mode == "train":
+        train(args)
+    elif args.mode == "sample":
+        sample(args)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/README.md b/anchorgen_baselines_20260923/input_cond/vis/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..43bbb6db1da156ca1bb00b5ffb66543804924044
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/README.md
@@ -0,0 +1,33 @@
+# input_cond/vis — metric-vs-drag-reduction curves
+
+Plots how each shape-similarity metric changes across drag-reduction buckets, for the methods:
+- **joint (sketch-as-LOSS)** — `deform/drag/runs/full_dataset_bigbkt_0817`
+- **input_cond plain (sketch-as-CONDITION)** — `input_cond/results/full_dataset_bigbkt_0817`
+- **7 CFG variants** (constant/linear/exponential/rectified_pp/zero_star/APG/SMC, w=2) —
+  `input_cond/cfg/cfg_results/full_dataset_cfg_0817`
+
+## Scripts & outputs
+- `plot_bucket_metric_curves.py` → `figs/` : the 2-method view (joint vs plain), 8 metrics per scope
+  + `headline_F1_CDl1_by_group.png`. Input = `anchors/loss_vs_condition_deepbucket_0817/DEEPBUCKET_METRICS.csv`.
+- `plot_all_methods_curves.py` → `figs_allmethods/` : **all 9 methods overlaid** (joint = bold black
+  solid, plain = bold red dashed, 7 CFG = thin distinct colors). Per scope
+  `allmethods_{ALL,length_side,width_front,height_rear}.png` (8 metrics) + `allmethods_headline_F1_CDl1.png`
+  (F@1% & CD-L1 across the 4 scopes). Inputs = the DEEPBUCKET_METRICS.csv (joint+plain) merged with
+  `input_cond/cfg/cfg_results/full_dataset_cfg_0817/SUMMARY_cfg_variants.csv` (7 CFG variants).
+
+Style adapted from the codex Table2 figure11 four-curve overlay
+(`.../outputs/pairwise-diffusion/make_figure11_fourcurve_overlay_20260618.py`).
+
+## Regenerate (CPU only)
+```
+dora/bin/python plot_bucket_metric_curves.py            # 2-method
+dora/bin/python plot_all_methods_curves.py              # all 9 methods
+```
+(If haas001 is down, run via the `rcp` jumphost = haas034; same scratch mount.)
+
+## Reading
+↓ metrics (CD-L1/L2, HD95): lower = closer to target. ↑ metrics (F-scores, normal-consistency):
+higher = closer. x-axis = actual drag reduction % at buckets 5/10/15/20.
+Result: joint(LOSS) is clearly best and stays ~flat; plain(CONDITION) erodes fastest; the 7 CFG
+variants sit BETWEEN — slightly above plain (best: APG) but well below joint (gap not closed).
+`C_D,p` = pressure-only Transolver surrogate, not total CFD drag.
diff --git a/anchorgen_baselines_20260923/input_cond/vis/__pycache__/plot_all_methods_curves.cpython-310.pyc b/anchorgen_baselines_20260923/input_cond/vis/__pycache__/plot_all_methods_curves.cpython-310.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..b1949da17eabc3be5055ae5384b1d1b2d90f3e41
Binary files /dev/null and b/anchorgen_baselines_20260923/input_cond/vis/__pycache__/plot_all_methods_curves.cpython-310.pyc differ
diff --git a/anchorgen_baselines_20260923/input_cond/vis/build_mindrag_curves.py b/anchorgen_baselines_20260923/input_cond/vis/build_mindrag_curves.py
new file mode 100644
index 0000000000000000000000000000000000000000..1fda1382916800d55266d4c385657f69298cf9b5
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/build_mindrag_curves.py
@@ -0,0 +1,98 @@
+#!/usr/bin/env python3
+"""Aggregate + plot the min-drag baseline comparison (shape retention vs SOURCE) over
+compare/{gd,flowcorrect,ictm}. These legs are pure min_drag (no sketch target) -> metrics
+are metrics_vs_source, so this is a SEPARATE figure from the sketch (vs-target) figs_allmethods.
+Scope = ALL (100 sources): min_drag has no view semantics, so no per-view split.
+Writes compare/MINDRAG_SUMMARY.csv and figs_allmethods/mindrag_{ALL,headline}.png.
+"""
+import json, glob, csv
+from collections import defaultdict
+from pathlib import Path
+import matplotlib; matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+from matplotlib.ticker import MaxNLocator
+
+CORR = Path("/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/correct")
+COMPARE = CORR / "compare"
+OUTDIR = CORR.parent / "input_cond/vis/figs_allmethods"
+METRICS = [("cd_l1_norm","CD-L1 (norm)  down"),("cd_l2_norm","CD-L2 (norm)  down"),
+  ("hd95_norm","HD95 (norm)  down"),("normal_consistency","Normal consistency  up"),
+  ("fscore_1pct","F-score @1%  up"),("fscore_0p5pct","F-score @0.5%  up"),
+  ("surface_area_ratio","Surface-area ratio"),("volume_ratio_abs","Volume ratio")]
+MK=[m[0] for m in METRICS]
+BUCKETS=["5pct","10pct","15pct","20pct"]   # unified standard: cap at 20% drag reduction
+STYLE={"gd":("GD (min-drag)","#000000","-",2.6,"o",10),
+       "flowcorrect":("Flow-Corrected","#d62728","--",2.4,"s",9),
+       "ictm":("ICTM","#1f77b4","-.",2.2,"^",8)}
+ORDER=[m for m in ["gd","flowcorrect","ictm"] if (COMPARE/m).is_dir()]
+
+def mean(xs):
+    xs=[x for x in xs if x is not None]
+    return sum(xs)/len(xs) if xs else None
+
+def load_leg(method):
+    agg=defaultdict(list)
+    for jp in glob.glob(str(COMPARE/method/"*"/"drag_result.json")):
+        b=json.load(open(jp)).get("buckets",{})
+        for bk,cell in b.items():
+            m=cell.get("metrics_vs_source")
+            if not m: continue
+            row={k:m.get(k) for k in MK}; row["drag"]=cell.get("drag_drop_actual")
+            agg[bk].append(row)
+    return agg
+
+def build():
+    OUTDIR.mkdir(parents=True,exist_ok=True); data={}; rows=[]
+    for method in ORDER:
+        agg=load_leg(method); data[method]={}
+        for bk in BUCKETS:
+            lst=agg.get(bk,[])
+            if not lst: continue
+            cell={k:mean([r[k] for r in lst]) for k in MK}
+            cell["drag"]=mean([r["drag"] for r in lst]); cell["n"]=len(lst)
+            data[method][bk]=cell
+            rows.append([method,"ALL",bk,len(lst)]+[cell[k] for k in MK]+[cell["drag"]])
+    with open(COMPARE/"MINDRAG_SUMMARY.csv","w",newline="") as f:
+        w=csv.writer(f); w.writerow(["method","group","bucket","n_matched"]+MK+["drag_drop_actual"]); w.writerows(rows)
+    print("legs:",ORDER,"| rows:",len(rows))
+    return data
+
+def curve(data,method,mk):
+    # UNIFIED: x = nominal bucket (5/10/15/20), capped 20%.
+    pts=[]
+    for bk in BUCKETS:
+        c=data.get(method,{}).get(bk)
+        if c and c.get(mk) is not None:
+            pts.append((float(bk.replace("pct","")),c[mk]))
+    return zip(*pts) if pts else ([],[])
+
+def plot_grid(data):
+    fig,axes=plt.subplots(2,4,figsize=(20,9))
+    for ax,(mk,title) in zip(axes.flat,METRICS):
+        for method in ORDER:
+            lab,col,ls,lw,mrk,z=STYLE[method]; xs,ys=curve(data,method,mk)
+            if xs: ax.plot(xs,ys,color=col,linestyle=ls,lw=lw,marker=mrk,ms=7,label=lab,zorder=z)
+        ax.set_title(title,fontsize=11); ax.set_xlabel("drag-reduction bucket (%)",fontsize=9)
+        ax.set_xticks([5,10,15,20]); ax.set_xlim(3.5,21.5); ax.grid(alpha=0.3)
+    h,l=axes.flat[4].get_legend_handles_labels()
+    fig.legend(h,l,loc="lower center",ncol=3,fontsize=12,frameon=False,bbox_to_anchor=(0.5,-0.01))
+    fig.suptitle("min-drag baselines: shape retention vs SOURCE (capped 20%)\n"
+                 "higher bucket => more shape change (CD/HD up, F down); NOTE vs-source, NOT vs-target",fontsize=13)
+    fig.tight_layout(rect=[0,0.05,1,0.94])
+    p=OUTDIR/"mindrag_ALL.png"; fig.savefig(p,dpi=130,bbox_inches="tight"); plt.close(fig); print("wrote",p)
+
+def plot_headline(data):
+    mets=[("fscore_1pct","F-score @1%  up"),("cd_l1_norm","CD-L1 (norm)  down")]
+    fig,axes=plt.subplots(1,2,figsize=(12,5))
+    for ax,(mk,title) in zip(axes,mets):
+        for method in ORDER:
+            lab,col,ls,lw,mrk,z=STYLE[method]; xs,ys=curve(data,method,mk)
+            if xs: ax.plot(xs,ys,color=col,linestyle=ls,lw=lw,marker=mrk,ms=7,label=lab,zorder=z)
+        ax.set_title(title,fontsize=12); ax.set_xlabel("drag-reduction bucket (%)",fontsize=10)
+        ax.set_xticks([5,10,15,20]); ax.set_xlim(3.5,21.5); ax.grid(alpha=0.3)
+    axes[0].legend(fontsize=10,frameon=False)
+    fig.suptitle("Headline (min-drag, vs source): retention vs drag reduction",fontsize=13)
+    fig.tight_layout(); p=OUTDIR/"mindrag_headline_F1_CDl1.png"; fig.savefig(p,dpi=130,bbox_inches="tight"); plt.close(fig); print("wrote",p)
+
+if __name__=="__main__":
+    d=build(); plot_grid(d); plot_headline(d); print("done ->",OUTDIR)
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs/headline_F1_CDl1_by_group.png b/anchorgen_baselines_20260923/input_cond/vis/figs/headline_F1_CDl1_by_group.png
new file mode 100644
index 0000000000000000000000000000000000000000..e7da0edad6aebc32952e77f82a22b494a3c14409
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs/headline_F1_CDl1_by_group.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:253d6c4389064c25eb86c990c52adcede8ddb30d479368dabb651a6b07e651b5
+size 246454
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_ALL.png b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_ALL.png
new file mode 100644
index 0000000000000000000000000000000000000000..f60ef3448efec17c3a346841498ea08f76033c5c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_ALL.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6814d5cf7fcbc8dc4e0fd46eed7ae003369aa7a52a81b1c031dfd7c4d7ff05fd
+size 262528
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_height_rear.png b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_height_rear.png
new file mode 100644
index 0000000000000000000000000000000000000000..8a83fd94628133a7b4a2850e696def91a97c06f2
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_height_rear.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d2cabcb47da4a43816eaac16322dd1be075290763b18749066db29e838dd985
+size 275909
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_length_side.png b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_length_side.png
new file mode 100644
index 0000000000000000000000000000000000000000..97de06b750d75c0c4e2ae8117a65b0368227ed68
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_length_side.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ddb1c32ec299a215d79ac8f4f1ec976defc5db840fffbeb11121d852190080c
+size 280758
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_width_front.png b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_width_front.png
new file mode 100644
index 0000000000000000000000000000000000000000..cdc636305bd3f3c37e29452aa9bd72ddc8d34b28
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs/metric_curves_width_front.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:240a4277db68908ba07e5f74ef8a49bf89f470063af9f3656975e4c23c2c7813
+size 253360
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_ALL.png b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_ALL.png
new file mode 100644
index 0000000000000000000000000000000000000000..1553d23e88d6f6a2116e1ba8adf2a840098a74d3
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_ALL.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adfa1ab6a1aeaaf792524b18b77f53c062fb97f249ecdda66e455fc819bb8d37
+size 540604
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_headline_F1_CDl1.png b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_headline_F1_CDl1.png
new file mode 100644
index 0000000000000000000000000000000000000000..41ba952944f6997fd0c15241eb0af4de85150a76
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_headline_F1_CDl1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1634dbc5641bd52688835d17db6c352357f32752bab13b5de37d2da7bd89b45
+size 489196
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_height_rear.png b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_height_rear.png
new file mode 100644
index 0000000000000000000000000000000000000000..421a1d69cf8e711ca8fd19471d9e1c5b5bba238f
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_height_rear.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2526a61340f72eb9aa4693e6b1fb372804594961b03cb34a726b53afe2eb9a6
+size 520574
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_length_side.png b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_length_side.png
new file mode 100644
index 0000000000000000000000000000000000000000..ea1a75f372ea6013b86c9554117d3b5ed5e1833c
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_length_side.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b469d62b018a15873d43aaa9c8a31450238b66f323db66f77c6d260fee13900
+size 469608
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_width_front.png b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_width_front.png
new file mode 100644
index 0000000000000000000000000000000000000000..4f1f3af84a6d955be7973e6912add972f8c07b37
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/allmethods_width_front.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b00a402f68cc5366a476539b611e1e5bfd23b0304a04fbb49f97f6f38d049b7
+size 477355
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/mindrag_ALL.png b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/mindrag_ALL.png
new file mode 100644
index 0000000000000000000000000000000000000000..2ca7d9950e744d05bf7bebdcf4bda3317d083dad
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/mindrag_ALL.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbbc5160b22820f80da9cbe64b335f49dc2bded65ed95f455b88086c775bb5c4
+size 316851
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/mindrag_headline_F1_CDl1.png b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/mindrag_headline_F1_CDl1.png
new file mode 100644
index 0000000000000000000000000000000000000000..0bd5c90d50882eb483af72cf6561f374a3687841
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_allmethods/mindrag_headline_F1_CDl1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eef3e69d002dbdc4a90edc058ff95d86330da83d2efb5098a0e681bb5999c7ad
+size 105472
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves.csv b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves.csv
new file mode 100644
index 0000000000000000000000000000000000000000..3ca4995588a95fa38133edfececc09c52b27cfd8
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves.csv
@@ -0,0 +1,154 @@
+method,group,bucket,n_matched,mean_relative_dimension_error_pct
+GD,length_side,5pct,200,4.638739238392718
+GD,length_side,10pct,200,4.723667310824381
+GD,length_side,15pct,200,4.769723983499253
+GD,length_side,20pct,200,4.822460138634531
+GD,width_front,5pct,200,0.5898013217680448
+GD,width_front,10pct,200,0.7909842092371047
+GD,width_front,15pct,200,1.0697630406901215
+GD,width_front,20pct,200,1.3657165063534094
+GD,height_rear,5pct,200,2.733569264730406
+GD,height_rear,10pct,200,4.740976066500629
+GD,height_rear,15pct,200,6.516690461850722
+GD,height_rear,20pct,200,7.896957355834079
+D-Flow,length_side,5pct,200,4.62264844659976
+D-Flow,length_side,10pct,200,4.653515064326668
+D-Flow,length_side,15pct,200,4.699033640415748
+D-Flow,length_side,20pct,200,4.756435679339194
+D-Flow,width_front,5pct,200,0.682055787672079
+D-Flow,width_front,10pct,200,0.8433307220764644
+D-Flow,width_front,15pct,200,0.9477787361480352
+D-Flow,width_front,20pct,200,1.0321491625516754
+D-Flow,height_rear,5pct,200,2.3560945438003116
+D-Flow,height_rear,10pct,200,3.7601366359229553
+D-Flow,height_rear,15pct,200,5.341327617926736
+D-Flow,height_rear,20pct,200,6.68820545556569
+ICTM,length_side,5pct,200,4.646922875011165
+ICTM,length_side,10pct,200,4.702541434468275
+ICTM,length_side,15pct,200,4.8012557879006765
+ICTM,width_front,5pct,200,0.7878283232099649
+ICTM,width_front,10pct,200,0.7840204841680627
+ICTM,width_front,15pct,200,0.859104834448715
+ICTM,height_rear,5pct,200,2.89142329861002
+ICTM,height_rear,10pct,200,3.6226141475608227
+ICTM,height_rear,15pct,200,4.780638945807413
+FCSO,length_side,5pct,200,4.638739238392718
+FCSO,length_side,10pct,200,4.724241462056359
+FCSO,length_side,15pct,200,4.778377040527718
+FCSO,length_side,20pct,200,4.849634552949196
+FCSO,width_front,5pct,200,0.5898013217680448
+FCSO,width_front,10pct,200,0.7863814801061324
+FCSO,width_front,15pct,200,1.0154960274575915
+FCSO,width_front,20pct,200,1.2443765199112502
+FCSO,height_rear,5pct,200,2.733569264730406
+FCSO,height_rear,10pct,200,4.659693489901276
+FCSO,height_rear,15pct,200,5.653020557810555
+FCSO,height_rear,20pct,200,7.028051244160138
+Input Cond.,length_side,5pct,200,4.572127242002286
+Input Cond.,length_side,10pct,200,4.559670639950384
+Input Cond.,length_side,15pct,200,4.514100106443038
+Input Cond.,length_side,20pct,200,4.508651437104069
+Input Cond.,width_front,5pct,200,0.5529936525851532
+Input Cond.,width_front,10pct,200,0.645581120055788
+Input Cond.,width_front,15pct,200,0.7631221888821544
+Input Cond.,width_front,20pct,200,0.8641188652229972
+Input Cond.,height_rear,5pct,200,2.689291613984983
+Input Cond.,height_rear,10pct,200,4.153155759552313
+Input Cond.,height_rear,15pct,200,5.448434443148554
+Input Cond.,height_rear,20pct,200,6.623997863592268
+Constant,length_side,5pct,200,4.22141223820976
+Constant,length_side,10pct,200,4.18220066327113
+Constant,length_side,15pct,200,4.160628555593914
+Constant,length_side,20pct,200,4.153419807627724
+Constant,width_front,5pct,200,0.5142903268832221
+Constant,width_front,10pct,200,0.5742844504434114
+Constant,width_front,15pct,200,0.7173787228474775
+Constant,width_front,20pct,200,0.9025517911051345
+Constant,height_rear,5pct,200,2.7054919717332613
+Constant,height_rear,10pct,200,4.141868759085017
+Constant,height_rear,15pct,200,5.520176871659808
+Constant,height_rear,20pct,200,6.668186547795844
+Linear Schedule,length_side,5pct,200,3.8802688093212088
+Linear Schedule,length_side,10pct,200,3.7986941471191527
+Linear Schedule,length_side,15pct,200,3.7469923989122553
+Linear Schedule,length_side,20pct,200,3.7189796345192394
+Linear Schedule,width_front,5pct,200,0.48858262780819683
+Linear Schedule,width_front,10pct,200,0.5462004555983728
+Linear Schedule,width_front,15pct,200,0.6824325898335057
+Linear Schedule,width_front,20pct,200,0.8455353895916682
+Linear Schedule,height_rear,5pct,200,2.668963026799929
+Linear Schedule,height_rear,10pct,200,4.106596494959191
+Linear Schedule,height_rear,15pct,200,5.3066296649307745
+Linear Schedule,height_rear,20pct,200,6.452514382866958
+Exponential Schedule,length_side,5pct,200,4.05208663506553
+Exponential Schedule,length_side,10pct,200,3.9929681071615537
+Exponential Schedule,length_side,15pct,200,3.9550804676927505
+Exponential Schedule,length_side,20pct,200,3.9425997637274977
+Exponential Schedule,width_front,5pct,200,0.5002037809624201
+Exponential Schedule,width_front,10pct,200,0.5622919152803233
+Exponential Schedule,width_front,15pct,200,0.6965385280772971
+Exponential Schedule,width_front,20pct,200,0.8704513055912427
+Exponential Schedule,height_rear,5pct,200,2.6502397286704538
+Exponential Schedule,height_rear,10pct,200,4.130623248418728
+Exponential Schedule,height_rear,15pct,200,5.488679713364077
+Exponential Schedule,height_rear,20pct,200,6.706124934545133
+APG,length_side,5pct,200,3.7730778799283216
+APG,length_side,10pct,200,3.689310341036669
+APG,length_side,15pct,200,3.6454587804554364
+APG,length_side,20pct,200,3.6200453730944187
+APG,width_front,5pct,200,0.48271895854412283
+APG,width_front,10pct,200,0.5407601107122492
+APG,width_front,15pct,200,0.6904723998841397
+APG,width_front,20pct,200,0.8446596322646376
+APG,height_rear,5pct,200,2.589533130203143
+APG,height_rear,10pct,200,4.067843526032437
+APG,height_rear,15pct,200,5.335533193543509
+APG,height_rear,20pct,200,6.469818892139943
+CFG-Zero\star$,length_side,5pct,200,4.2726731807412435
+CFG-Zero\star$,length_side,10pct,200,4.2456646013126225
+CFG-Zero\star$,length_side,15pct,200,4.218670181495122
+CFG-Zero\star$,length_side,20pct,200,4.208424555848474
+CFG-Zero\star$,width_front,5pct,200,0.5147300620453644
+CFG-Zero\star$,width_front,10pct,200,0.5783316075789975
+CFG-Zero\star$,width_front,15pct,200,0.7226088677651538
+CFG-Zero\star$,width_front,20pct,200,0.9012952435477682
+CFG-Zero\star$,height_rear,5pct,200,2.6600071620015022
+CFG-Zero\star$,height_rear,10pct,200,4.143296107089038
+CFG-Zero\star$,height_rear,15pct,200,5.363457528358266
+CFG-Zero\star$,height_rear,20pct,200,6.506557466487664
+Rectified-CFG++,length_side,5pct,200,4.3356337059901096
+Rectified-CFG++,length_side,10pct,200,4.316888147083044
+Rectified-CFG++,length_side,15pct,200,4.303219204476561
+Rectified-CFG++,length_side,20pct,200,4.3117581229388495
+Rectified-CFG++,width_front,5pct,200,0.5210456722409869
+Rectified-CFG++,width_front,10pct,200,0.5845810972575797
+Rectified-CFG++,width_front,15pct,200,0.72929692141871
+Rectified-CFG++,width_front,20pct,200,0.8925244554530863
+Rectified-CFG++,height_rear,5pct,200,2.7601954350678115
+Rectified-CFG++,height_rear,10pct,200,4.253770826262294
+Rectified-CFG++,height_rear,15pct,200,5.627184700792903
+Rectified-CFG++,height_rear,20pct,200,6.850964772651529
+SMC,length_side,5pct,200,5.1434277351957185
+SMC,length_side,10pct,200,5.164640003063832
+SMC,length_side,15pct,200,5.241386690797884
+SMC,length_side,20pct,200,5.315085496796575
+SMC,width_front,5pct,200,0.5920627722799593
+SMC,width_front,10pct,200,0.6823378011221785
+SMC,width_front,15pct,200,0.8735693687175271
+SMC,width_front,20pct,200,1.0969691786502849
+SMC,height_rear,5pct,200,2.8112681393446444
+SMC,height_rear,10pct,200,4.385276673755711
+SMC,height_rear,15pct,200,5.767510103746747
+SMC,height_rear,20pct,200,7.064855556179485
+AnchorGen (Ours),length_side,5pct,200,4.255898602939848
+AnchorGen (Ours),length_side,10pct,200,3.8226806514623415
+AnchorGen (Ours),length_side,15pct,200,3.194318474655063
+AnchorGen (Ours),length_side,20pct,200,2.6465781255522223
+AnchorGen (Ours),width_front,5pct,200,0.5513395934202093
+AnchorGen (Ours),width_front,10pct,200,0.5805593525835294
+AnchorGen (Ours),width_front,15pct,200,0.5722220312243858
+AnchorGen (Ours),width_front,20pct,200,0.5756057792764591
+AnchorGen (Ours),height_rear,5pct,200,2.6347102762204697
+AnchorGen (Ours),height_rear,10pct,200,3.660091962119319
+AnchorGen (Ours),height_rear,15pct,200,4.241351633496111
+AnchorGen (Ours),height_rear,20pct,200,4.784604109311413
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper.pdf b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..636b6831eb2f51ff1ffb4bb64f5eddb365b680ae
Binary files /dev/null and b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper.pdf differ
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper.png
new file mode 100644
index 0000000000000000000000000000000000000000..4dcbb8caeef11bc2efc7addeca0bc0e9ed3f0bdb
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:046dfd726f88ab9c0165c25789ee177c5c96c3488fc5bba30cce22578cc8c540
+size 655575
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper_legend2row.pdf b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper_legend2row.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..ebf521f5da0d90a6404f1e19e5b5e4057e8429d7
Binary files /dev/null and b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper_legend2row.pdf differ
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper_legend2row.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper_legend2row.png
new file mode 100644
index 0000000000000000000000000000000000000000..b301357bf53fe268e9a01e213054b773c8c28676
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/dimension_error_curves_paper_legend2row.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2efc1142c794d5b732fb2c183b30bed32df376f46ed4117a6357154208a5d1ef
+size 738150
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/error_curve_file.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/error_curve_file.png
new file mode 100644
index 0000000000000000000000000000000000000000..ade8786b4de8283ae744b2dfa0e78840306f70f1
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/error_curve_file.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3c3369583a226b24ef191d7edc1ac24d1a012f3c8904f500172c8aa6090fcf2
+size 509393
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_height_rear.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_height_rear.png
new file mode 100644
index 0000000000000000000000000000000000000000..32518009d6ef407c184e103e0808c6d5d6051358
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_height_rear.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:846ae9b5b005638cfe0b4372f3b547e7a68d9fbc0d75bcb0fd4e74da0a5c7935
+size 314217
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_length_side.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_length_side.png
new file mode 100644
index 0000000000000000000000000000000000000000..1cc9fd220c48265db47d0755eb550fc1adbe12a3
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_length_side.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c015b5854504d236511a431b120665b1864bda9f8563e341ab9643110cb6fd02
+size 329379
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_ALL.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_ALL.png
new file mode 100644
index 0000000000000000000000000000000000000000..03a029582bda8c34530ee5b709989118d958a9c2
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_ALL.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33021f8aea6e6b43fb520310e0f81c9bbd4375024d2f28d3e3eb6da17bffdb09
+size 231803
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_height_rear.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_height_rear.png
new file mode 100644
index 0000000000000000000000000000000000000000..a39625a2ca16e717283d3388de80d2be1e94252d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_height_rear.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8f67292a7870183deb837f6d2208d0ebbd4a3277809b915c4d6bf71dc5285c2
+size 227765
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_length_side.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_length_side.png
new file mode 100644
index 0000000000000000000000000000000000000000..12eb312685b4b6a1b62255a5cbb209fb45026750
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_length_side.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2221e00ab6b57f6365058650469b5088553a884b90ec6d7d8ed4112a369b74a2
+size 235845
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_width_front.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_width_front.png
new file mode 100644
index 0000000000000000000000000000000000000000..3a24ab42a8707edb49499c6340db74375aa1cf73
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_sgo_filtered_width_front.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b38b4ed1cb940fa0409734df9281226a7904d176a80e456f20112d5b18c340b
+size 204664
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_width_front.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_width_front.png
new file mode 100644
index 0000000000000000000000000000000000000000..e9dd28b5a94fa6c9928e31262609849641bc4d84
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/final_width_front.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac3837601fa5875eedd8c8c82a7b302df0f309454907042cd2984040b664c970
+size 292368
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_curve_file.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_curve_file.png
new file mode 100644
index 0000000000000000000000000000000000000000..c449fe8cd050bed066498cc984d9fede493a79e8
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_curve_file.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60137a89dbdad9f964427667df2db63cc38d62d5796eebd7d6b6c9a32fb9705c
+size 672247
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_similarity_curves_paper_legend2row.pdf b/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_similarity_curves_paper_legend2row.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..209a74a989ea9807e5d73ccce7d70c95d829139c
Binary files /dev/null and b/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_similarity_curves_paper_legend2row.pdf differ
diff --git a/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_similarity_curves_paper_legend2row.png b/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_similarity_curves_paper_legend2row.png
new file mode 100644
index 0000000000000000000000000000000000000000..bd96b75428fba395d86db5d76daa96012daf536b
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/figs_final/shape_similarity_curves_paper_legend2row.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49e6dcecb71c3bde788532f90e2620942498302a7dcdd93bf0d27ea43c938cc4
+size 1060474
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_all_methods_curves.py b/anchorgen_baselines_20260923/input_cond/vis/plot_all_methods_curves.py
new file mode 100644
index 0000000000000000000000000000000000000000..7ded118dc2d8e2b7cf8d436b372dbd8d14cc8199
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_all_methods_curves.py
@@ -0,0 +1,179 @@
+#!/usr/bin/env python3
+"""All-method curves: joint(LOSS) + input_cond plain(CONDITION) + 7 CFG variants
++ ICTM min-drag baseline, all vs the SAME 600 matched targets, vs drag reduction.
+
+Merges tidy CSVs:
+  - DEEPBUCKET_METRICS.csv    (methods: joint_LOSS, input_cond_CONDITION)
+  - SUMMARY_cfg_variants.csv  (7 CFG variants)
+  - ICTM_VS_TARGET_SUMMARY.csv (ICTM min-drag scored vs the 600 targets; see
+                                correct/eval_mindrag_vs_target.py). ICTM is target-agnostic,
+                                so it is a lower-bound baseline: pure drag reduction, no target pull.
+Style follows the codex figure11 four-curve overlay. joint = bold black solid, plain = bold red
+dashed, CFG variants = thin distinct-color lines, ICTM = bold orange dotted (baseline).
+Per scope (ALL + 3 groups): 8-metric grid; plus a compact headline (F@1% & CD-L1) across scopes.
+"""
+from __future__ import annotations
+import argparse, csv
+from collections import defaultdict
+from pathlib import Path
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+from matplotlib.ticker import MaxNLocator
+
+METRICS = [
+    ("cd_l1_norm", "CD-L1 (norm)  down"), ("cd_l2_norm", "CD-L2 (norm)  down"),
+    ("hd95_norm", "HD95 (norm)  down"), ("normal_consistency", "Normal consistency  up"),
+    ("fscore_1pct", "F-score @1%  up"), ("fscore_0p5pct", "F-score @0.5%  up"),
+    ("surface_area_ratio", "Surface-area ratio"), ("volume_ratio_abs", "Volume ratio"),
+]
+GROUPS = ["ALL", "length_side", "width_front", "height_rear"]
+BUCKET_ORDER = ["5pct", "10pct", "15pct", "20pct"]
+MK = [m[0] for m in METRICS]
+
+# CFG variant palette (distinct); order = best->worst by -20% F@1%
+CFG_ORDER = ["APG", "linear", "exponential", "constant", "zero_star", "rectified_pp", "SMC"]
+CFG_COLORS = ["#1f77b4", "#2ca02c", "#9467bd", "#8c564b", "#17becf", "#bcbd22", "#e377c2"]
+# style registry: method_key -> (label, color, linestyle, lw, marker, zorder)
+STYLE = {
+    "joint_LOSS": ("joint (LOSS)", "#000000", "-", 2.8, "o", 10),
+    "input_cond_CONDITION": ("input_cond plain (CONDITION)", "#d62728", "--", 2.6, "s", 9),
+    "gd_mindrag": ("GD (min-drag, no target)", "#7570b3", ":", 2.6, "P", 8),
+    "flowcorrect_mindrag": ("Flow-Corrected (min-drag, no target)", "#e7298a", ":", 2.6, "X", 8),
+    "fmg_mindrag": ("FMG (drag-CFG, no target)", "#1b9e77", "-.", 2.8, "h", 8),
+    "ictm_mindrag": ("ICTM (min-drag, no target)", "#ff7f0e", ":", 2.6, "^", 8),
+    "dflow_mindrag": ("D-Flow (min-drag, no target)", "#008080", ":", 2.6, "v", 8),
+    "sgo_mindrag": ("SGO/SDS (min-drag, no target)", "#555555", ":", 2.6, "D", 8),
+}
+for v, c in zip(CFG_ORDER, CFG_COLORS):
+    STYLE["cfg:" + v] = ("cfg:%s" % v, c, "-", 1.4, ".", 5)
+METHOD_ORDER = (["joint_LOSS", "input_cond_CONDITION",
+                 "gd_mindrag", "flowcorrect_mindrag", "ictm_mindrag", "dflow_mindrag", "fmg_mindrag"]
+                + ["cfg:" + v for v in CFG_ORDER])
+
+
+def num(x):
+    try:
+        return float(x)
+    except Exception:
+        return None
+
+
+def load(deep_csv, cfg_csv, mindrag_csvs):
+    data = defaultdict(dict)  # method -> {(group,bucket): {metric:val, drag:val}}
+    for r in csv.DictReader(open(deep_csv)):
+        m = r["method"]
+        cell = {k: num(r.get(k)) for k in MK}
+        cell["drag"] = num(r.get("drag_drop_actual"))
+        data[m][(r["group"], r["bucket"])] = cell
+    for r in csv.DictReader(open(cfg_csv)):
+        m = "cfg:" + r["variant"]
+        cell = {k: num(r.get(k)) for k in MK}
+        cell["drag"] = num(r.get("drag_drop"))
+        data[m][(r["group"], r["bucket"])] = cell
+    # min-drag legs (ictm/dflow/sgo) computed vs the same 600 targets -> {method_key: csv_path}
+    for method_key, path in mindrag_csvs.items():
+        if path and Path(path).exists():
+            for r in csv.DictReader(open(path)):
+                cell = {k: num(r.get(k)) for k in MK}
+                cell["drag"] = num(r.get("drag_drop_actual"))
+                data[method_key][(r["group"], r["bucket"])] = cell
+    return data
+
+
+def curve(data, method, scope, mk):
+    # UNIFIED standard: x = nominal drag-reduction bucket (5/10/15/20), capped at 20%, so every
+    # method is compared at the SAME operating points on the SAME matched cases. (min-drag legs
+    # may overshoot the nominal drag — that is a caption note, not an x-axis distortion.)
+    d = data.get(method, {})
+    pts = []
+    for b in BUCKET_ORDER:
+        c = d.get((scope, b))
+        if c and c.get(mk) is not None:
+            pts.append((float(b.replace("pct", "")), c[mk]))
+    return zip(*pts) if pts else ([], [])
+
+
+def plot_scope(data, scope, out_png):
+    fig, axes = plt.subplots(2, 4, figsize=(20, 9))
+    for ax, (mk, title) in zip(axes.flat, METRICS):
+        for method in METHOD_ORDER:
+            if method not in data:
+                continue
+            label, color, ls, lw, mrk, z = STYLE[method]
+            xs, ys = curve(data, method, scope, mk)
+            if xs:
+                ax.plot(xs, ys, color=color, linestyle=ls, lw=lw, marker=mrk,
+                        ms=6 if lw > 2 else 4, label=label, zorder=z)
+        ax.set_title(title, fontsize=11)
+        ax.set_xlabel("drag-reduction bucket (%)", fontsize=9)
+        ax.set_xticks([5, 10, 15, 20]); ax.set_xlim(3.5, 21.5)
+        ax.grid(alpha=0.3)
+    handles, labels = axes.flat[4].get_legend_handles_labels()
+    fig.legend(handles, labels, loc="lower center", ncol=5, fontsize=9, frameon=False,
+               bbox_to_anchor=(0.5, -0.01))
+    fig.suptitle("Similarity to target vs drag reduction (capped 20%%) — %s" % scope, fontsize=14)
+    fig.tight_layout(rect=[0, 0.05, 1, 0.96])
+    out_png.parent.mkdir(parents=True, exist_ok=True)
+    fig.savefig(out_png, dpi=130, bbox_inches="tight"); plt.close(fig)
+    print("wrote", out_png)
+
+
+def plot_headline(data, out_png):
+    metrics = [("fscore_1pct", "F-score @1%  up"), ("cd_l1_norm", "CD-L1 (norm)  down")]
+    fig, axes = plt.subplots(2, 4, figsize=(20, 9), sharex=True)
+    for row, (mk, title) in enumerate(metrics):
+        for col, scope in enumerate(GROUPS):
+            ax = axes[row, col]
+            for method in METHOD_ORDER:
+                if method not in data:
+                    continue
+                label, color, ls, lw, mrk, z = STYLE[method]
+                xs, ys = curve(data, method, scope, mk)
+                if xs:
+                    ax.plot(xs, ys, color=color, linestyle=ls, lw=lw, marker=mrk,
+                            ms=6 if lw > 2 else 4, label=label if (row == 0 and col == 0) else None, zorder=z)
+            if row == 0:
+                ax.set_title(scope, fontsize=12)
+            if col == 0:
+                ax.set_ylabel(title, fontsize=11)
+            if row == 1:
+                ax.set_xlabel("drag-reduction bucket (%)", fontsize=9)
+            ax.set_xticks([5, 10, 15, 20]); ax.set_xlim(3.5, 21.5)
+            ax.grid(alpha=0.3)
+    handles, labels = axes[0, 0].get_legend_handles_labels()
+    fig.legend(handles, labels, loc="lower center", ncol=5, fontsize=9, frameon=False, bbox_to_anchor=(0.5, -0.01))
+    fig.suptitle("Headline: target similarity vs drag reduction (capped 20%)", fontsize=14)
+    fig.tight_layout(rect=[0, 0.05, 1, 0.96])
+    fig.savefig(out_png, dpi=130, bbox_inches="tight"); plt.close(fig)
+    print("wrote", out_png)
+
+
+def main():
+    HERE = Path(__file__).resolve().parent
+    CD = HERE.parent.parent
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--deep-csv", default=str(CD / "anchors/loss_vs_condition_deepbucket_0817/DEEPBUCKET_METRICS.csv"))
+    ap.add_argument("--cfg-csv", default=str(CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/SUMMARY_cfg_variants.csv"))
+    ap.add_argument("--ictm-csv", default=str(CD / "correct/compare/ICTM_VS_TARGET_SUMMARY.csv"))
+    ap.add_argument("--dflow-csv", default=str(CD / "correct/compare/DFLOW_VS_TARGET_SUMMARY.csv"))
+    ap.add_argument("--sgo-csv", default=str(CD / "correct/compare/SGO_VS_TARGET_SUMMARY.csv"))
+    ap.add_argument("--gd-csv", default=str(CD / "correct/compare/GD_VS_TARGET_SUMMARY.csv"))
+    ap.add_argument("--flowcorrect-csv", default=str(CD / "correct/compare/FLOWCORRECT_VS_TARGET_SUMMARY.csv"))
+    ap.add_argument("--fmg-csv", default=str(CD / "correct/compare/FMG_VS_TARGET_SUMMARY.csv"))
+    ap.add_argument("--out-dir", default=str(HERE / "figs_allmethods"))
+    args = ap.parse_args()
+    data = load(args.deep_csv, args.cfg_csv,
+                {"ictm_mindrag": args.ictm_csv, "dflow_mindrag": args.dflow_csv,
+                 "sgo_mindrag": args.sgo_csv, "gd_mindrag": args.gd_csv,
+                 "flowcorrect_mindrag": args.flowcorrect_csv, "fmg_mindrag": args.fmg_csv})
+    print("methods loaded:", [m for m in METHOD_ORDER if m in data])
+    out = Path(args.out_dir)
+    for scope in GROUPS:
+        plot_scope(data, scope, out / ("allmethods_%s.png" % scope))
+    plot_headline(data, out / "allmethods_headline_F1_CDl1.png")
+    print("done ->", out)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_all_methods_curves.py.bak_pre_ictm b/anchorgen_baselines_20260923/input_cond/vis/plot_all_methods_curves.py.bak_pre_ictm
new file mode 100644
index 0000000000000000000000000000000000000000..8e82710e39728dc6b0e301372bf6ff2e725fc1cd
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_all_methods_curves.py.bak_pre_ictm
@@ -0,0 +1,141 @@
+#!/usr/bin/env python3
+"""All-method curves: joint(LOSS) + input_cond plain(CONDITION) + 7 CFG variants, vs drag reduction.
+
+Merges two tidy CSVs:
+  - DEEPBUCKET_METRICS.csv  (methods: joint_LOSS, input_cond_CONDITION)
+  - SUMMARY_cfg_variants.csv (7 CFG variants: APG/linear/exponential/constant/zero_star/rectified_pp/SMC)
+Style follows the codex figure11 four-curve overlay. joint = bold black solid, plain = bold red dashed,
+CFG variants = thin distinct-color lines, so it's easy to see which CFG sits closest to joint.
+Per scope (ALL + 3 groups): 8-metric grid; plus a compact headline (F@1% & CD-L1) across scopes.
+"""
+from __future__ import annotations
+import argparse, csv
+from collections import defaultdict
+from pathlib import Path
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+from matplotlib.ticker import MaxNLocator
+
+METRICS = [
+    ("cd_l1_norm", "CD-L1 (norm)  ↓"), ("cd_l2_norm", "CD-L2 (norm)  ↓"),
+    ("hd95_norm", "HD95 (norm)  ↓"), ("normal_consistency", "Normal consistency  ↑"),
+    ("fscore_1pct", "F-score @1%  ↑"), ("fscore_0p5pct", "F-score @0.5%  ↑"),
+    ("surface_area_ratio", "Surface-area ratio"), ("volume_ratio_abs", "Volume ratio"),
+]
+GROUPS = ["ALL", "length_side", "width_front", "height_rear"]
+BUCKET_ORDER = ["5pct", "10pct", "15pct", "20pct"]
+MK = [m[0] for m in METRICS]
+
+# CFG variant palette (distinct); order = best->worst by -20% F@1%
+CFG_ORDER = ["APG", "linear", "exponential", "constant", "zero_star", "rectified_pp", "SMC"]
+CFG_COLORS = ["#1f77b4", "#2ca02c", "#9467bd", "#8c564b", "#17becf", "#bcbd22", "#e377c2"]
+# style registry: method_key -> (label, color, linestyle, lw, marker, zorder)
+STYLE = {
+    "joint_LOSS": ("joint (LOSS)", "#000000", "-", 2.8, "o", 10),
+    "input_cond_CONDITION": ("input_cond plain (CONDITION)", "#d62728", "--", 2.6, "s", 9),
+}
+for v, c in zip(CFG_ORDER, CFG_COLORS):
+    STYLE["cfg:" + v] = ("cfg:%s" % v, c, "-", 1.4, ".", 5)
+METHOD_ORDER = ["joint_LOSS", "input_cond_CONDITION"] + ["cfg:" + v for v in CFG_ORDER]
+
+
+def load(deep_csv, cfg_csv):
+    data = defaultdict(dict)  # method -> {(group,bucket): {metric:val, drag:val}}
+
+    def num(x):
+        try:
+            return float(x)
+        except Exception:
+            return None
+    for r in csv.DictReader(open(deep_csv)):
+        m = r["method"]
+        cell = {k: num(r.get(k)) for k in MK}
+        cell["drag"] = num(r.get("drag_drop_actual"))
+        data[m][(r["group"], r["bucket"])] = cell
+    for r in csv.DictReader(open(cfg_csv)):
+        m = "cfg:" + r["variant"]
+        cell = {k: num(r.get(k)) for k in MK}
+        cell["drag"] = num(r.get("drag_drop"))
+        data[m][(r["group"], r["bucket"])] = cell
+    return data
+
+
+def curve(data, method, scope, mk):
+    d = data.get(method, {})
+    pts = []
+    for b in BUCKET_ORDER:
+        c = d.get((scope, b))
+        if c and c.get(mk) is not None and c.get("drag") is not None:
+            pts.append((c["drag"] * 100.0, c[mk]))
+    return zip(*pts) if pts else ([], [])
+
+
+def plot_scope(data, scope, out_png):
+    fig, axes = plt.subplots(2, 4, figsize=(20, 9))
+    for ax, (mk, title) in zip(axes.flat, METRICS):
+        for method in METHOD_ORDER:
+            label, color, ls, lw, mrk, z = STYLE[method]
+            xs, ys = curve(data, method, scope, mk)
+            if xs:
+                ax.plot(xs, ys, color=color, linestyle=ls, lw=lw, marker=mrk,
+                        ms=6 if lw > 2 else 4, label=label, zorder=z)
+        ax.set_title(title, fontsize=11)
+        ax.set_xlabel("drag reduction $C_{D,p}$ (%)", fontsize=9)
+        ax.grid(alpha=0.3); ax.xaxis.set_major_locator(MaxNLocator(nbins=5))
+    handles, labels = axes.flat[4].get_legend_handles_labels()
+    fig.legend(handles, labels, loc="lower center", ncol=5, fontsize=9, frameon=False,
+               bbox_to_anchor=(0.5, -0.01))
+    fig.suptitle("Similarity to target vs drag reduction — %s  (joint / plain / 7 CFG variants)" % scope,
+                 fontsize=14)
+    fig.tight_layout(rect=[0, 0.05, 1, 0.96])
+    out_png.parent.mkdir(parents=True, exist_ok=True)
+    fig.savefig(out_png, dpi=130, bbox_inches="tight"); plt.close(fig)
+    print("wrote", out_png)
+
+
+def plot_headline(data, out_png):
+    metrics = [("fscore_1pct", "F-score @1%  ↑"), ("cd_l1_norm", "CD-L1 (norm)  ↓")]
+    fig, axes = plt.subplots(2, 4, figsize=(20, 9), sharex=True)
+    for row, (mk, title) in enumerate(metrics):
+        for col, scope in enumerate(GROUPS):
+            ax = axes[row, col]
+            for method in METHOD_ORDER:
+                label, color, ls, lw, mrk, z = STYLE[method]
+                xs, ys = curve(data, method, scope, mk)
+                if xs:
+                    ax.plot(xs, ys, color=color, linestyle=ls, lw=lw, marker=mrk,
+                            ms=6 if lw > 2 else 4, label=label if (row == 0 and col == 0) else None, zorder=z)
+            if row == 0:
+                ax.set_title(scope, fontsize=12)
+            if col == 0:
+                ax.set_ylabel(title, fontsize=11)
+            if row == 1:
+                ax.set_xlabel("drag reduction (%)", fontsize=9)
+            ax.grid(alpha=0.3)
+    handles, labels = axes[0, 0].get_legend_handles_labels()
+    fig.legend(handles, labels, loc="lower center", ncol=5, fontsize=9, frameon=False, bbox_to_anchor=(0.5, -0.01))
+    fig.suptitle("Headline: target similarity vs drag reduction — joint / plain / 7 CFG variants", fontsize=14)
+    fig.tight_layout(rect=[0, 0.05, 1, 0.96])
+    fig.savefig(out_png, dpi=130, bbox_inches="tight"); plt.close(fig)
+    print("wrote", out_png)
+
+
+def main():
+    HERE = Path(__file__).resolve().parent
+    CD = HERE.parent.parent
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--deep-csv", default=str(CD / "anchors/loss_vs_condition_deepbucket_0817/DEEPBUCKET_METRICS.csv"))
+    ap.add_argument("--cfg-csv", default=str(CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/SUMMARY_cfg_variants.csv"))
+    ap.add_argument("--out-dir", default=str(HERE / "figs_allmethods"))
+    args = ap.parse_args()
+    data = load(args.deep_csv, args.cfg_csv)
+    out = Path(args.out_dir)
+    for scope in GROUPS:
+        plot_scope(data, scope, out / ("allmethods_%s.png" % scope))
+    plot_headline(data, out / "allmethods_headline_F1_CDl1.png")
+    print("done ->", out)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_bucket_metric_curves.py b/anchorgen_baselines_20260923/input_cond/vis/plot_bucket_metric_curves.py
new file mode 100644
index 0000000000000000000000000000000000000000..3e1b297b69565f5617c625e0d15e30e16be2c2b6
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_bucket_metric_curves.py
@@ -0,0 +1,130 @@
+#!/usr/bin/env python3
+"""Metric-vs-drag-reduction curves: joint (sketch-as-LOSS) vs input_cond (sketch-as-CONDITION).
+
+Style adapted from the codex Table2 figure11 four-curve overlay
+(`.../outputs/pairwise-diffusion/make_figure11_fourcurve_overlay_20260618.py`):
+one subplot per metric, methods overlaid as curves, colorblind palette, Agg backend.
+
+Input = the frozen tidy CSV produced by freeze_deepbucket_compare.py
+(`method,group,bucket,n_matched,<8 metrics>,drag_drop_actual`). Makes one figure per scope
+(ALL + 3 groups); x-axis = actual drag reduction (%), two curves (joint solid / input_cond dashed).
+"""
+from __future__ import annotations
+
+import argparse
+import csv
+from collections import defaultdict
+from pathlib import Path
+
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+from matplotlib.ticker import MaxNLocator
+
+# method key -> (legend label, color, linestyle, marker)
+METHODS = [
+    ("joint_LOSS", "joint  (sketch-as-LOSS)", "#0072B2", "-", "o"),
+    ("input_cond_CONDITION", "input_cond  (sketch-as-CONDITION)", "#D55E00", "--", "s"),
+]
+# metric key -> (axis title, "down" means lower=better)
+METRICS = [
+    ("cd_l1_norm", "CD-L1 (norm)  ↓", True),
+    ("cd_l2_norm", "CD-L2 (norm)  ↓", True),
+    ("hd95_norm", "HD95 (norm)  ↓", True),
+    ("normal_consistency", "Normal consistency  ↑", False),
+    ("fscore_1pct", "F-score @1%  ↑", False),
+    ("fscore_0p5pct", "F-score @0.5%  ↑", False),
+    ("surface_area_ratio", "Surface-area ratio", False),
+    ("volume_ratio_abs", "Volume ratio", False),
+]
+GROUPS = ["ALL", "length_side", "width_front", "height_rear"]
+BUCKET_ORDER = ["5pct", "10pct", "15pct", "20pct"]
+
+
+def load(csv_path):
+    metric_keys = [m[0] for m in METRICS] + ["drag_drop_actual"]
+    data = defaultdict(dict)   # (method,group) -> {bucket: {metric:val}}
+    n_matched = {}
+    for r in csv.DictReader(open(csv_path)):
+        cell = {}
+        for k in metric_keys:
+            v = r.get(k, "")
+            cell[k] = float(v) if v not in ("", "None", None) else None
+        data[(r["method"], r["group"])][r["bucket"]] = cell
+        n_matched[r["group"]] = r.get("n_matched", "?")
+    return data, n_matched
+
+
+def plot_scope(data, n_matched, scope, out_png):
+    fig, axes = plt.subplots(2, 4, figsize=(19, 8.5))
+    for ax, (mk, title, down) in zip(axes.flat, METRICS):
+        for meth, label, color, ls, mrk in METHODS:
+            d = data.get((meth, scope), {})
+            pts = [(d[b]["drag_drop_actual"] * 100.0, d[b][mk]) for b in BUCKET_ORDER
+                   if b in d and d[b].get(mk) is not None and d[b].get("drag_drop_actual") is not None]
+            if not pts:
+                continue
+            xs, ys = zip(*pts)
+            ax.plot(xs, ys, color=color, linestyle=ls, marker=mrk, label=label, lw=2.2, ms=7)
+        ax.set_title(title, fontsize=11)
+        ax.set_xlabel("drag reduction  $C_{D,p}$  (%)", fontsize=9)
+        ax.grid(alpha=0.3)
+        ax.xaxis.set_major_locator(MaxNLocator(nbins=5))
+    axes.flat[0].legend(loc="best", fontsize=9, framealpha=0.9)
+    fig.suptitle("Shape similarity to target vs drag reduction  —  %s  (matched n=%s)"
+                 % (scope, n_matched.get(scope, "?")), fontsize=14)
+    fig.tight_layout(rect=[0, 0, 1, 0.96])
+    out_png.parent.mkdir(parents=True, exist_ok=True)
+    fig.savefig(out_png, dpi=130)
+    plt.close(fig)
+    print("wrote", out_png)
+
+
+def plot_headline_gap(data, out_png):
+    """One compact figure: F@1% and cd_l1_norm, ALL + 3 groups columns, both methods."""
+    cols = GROUPS
+    metrics = [("fscore_1pct", "F-score @1%  ↑", False), ("cd_l1_norm", "CD-L1 (norm)  ↓", True)]
+    fig, axes = plt.subplots(2, 4, figsize=(19, 8), sharex=True)
+    for row, (mk, title, _down) in enumerate(metrics):
+        for col, scope in enumerate(cols):
+            ax = axes[row, col]
+            for meth, label, color, ls, mrk in METHODS:
+                d = data.get((meth, scope), {})
+                pts = [(d[b]["drag_drop_actual"] * 100.0, d[b][mk]) for b in BUCKET_ORDER
+                       if b in d and d[b].get(mk) is not None]
+                if pts:
+                    xs, ys = zip(*pts)
+                    ax.plot(xs, ys, color=color, linestyle=ls, marker=mrk, lw=2.2, ms=7,
+                            label=label if (row == 0 and col == 0) else None)
+            if row == 0:
+                ax.set_title(scope, fontsize=12)
+            if col == 0:
+                ax.set_ylabel(title, fontsize=11)
+            if row == 1:
+                ax.set_xlabel("drag reduction (%)", fontsize=9)
+            ax.grid(alpha=0.3)
+    axes[0, 0].legend(loc="best", fontsize=8.5)
+    fig.suptitle("Headline: target similarity vs drag reduction (LOSS vs CONDITION)", fontsize=14)
+    fig.tight_layout(rect=[0, 0, 1, 0.96])
+    fig.savefig(out_png, dpi=130)
+    plt.close(fig)
+    print("wrote", out_png)
+
+
+def main():
+    HERE = Path(__file__).resolve().parent
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--csv", default=str(HERE.parent.parent /
+                    "anchors/loss_vs_condition_deepbucket_0817/DEEPBUCKET_METRICS.csv"))
+    ap.add_argument("--out-dir", default=str(HERE / "figs"))
+    args = ap.parse_args()
+    data, n_matched = load(args.csv)
+    out = Path(args.out_dir)
+    for scope in GROUPS:
+        plot_scope(data, n_matched, scope, out / ("metric_curves_%s.png" % scope))
+    plot_headline_gap(data, out / "headline_F1_CDl1_by_group.png")
+    print("done -> %s" % out)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves.py b/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves.py
new file mode 100644
index 0000000000000000000000000000000000000000..01bfa5dc5a0b4284596b12b60c432a5e050230c3
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves.py
@@ -0,0 +1,277 @@
+#!/usr/bin/env python3
+"""Plot target-relative controlled-dimension error at 5/10/15/20% drag.
+
+The metric is shape-level:
+    100 * abs(D_pred - D_target) / D_target
+where D is x-extent (Side/Length), y-extent (Front/Width), or z-extent
+(Rear/Height). Each plotted point is the mean over the fixed matched pairs.
+
+FMG and SGO are excluded. ICTM is shown only at 5/10/15%, where coverage is
+complete; its incomplete 20% success subset is intentionally not plotted.
+"""
+from __future__ import annotations
+
+import csv
+import json
+from collections import defaultdict
+from pathlib import Path
+
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+
+import plot_all_methods_curves as P
+
+
+HERE = Path(__file__).resolve().parent
+CD = HERE.parent.parent
+PAIR_CSV = CD / "cases/full_dataset/pairs.csv"
+JOINT_ROOT = CD / "deform/drag/runs/full_dataset_bigbkt_0817"
+OUT_DIR = HERE / "figs_final"
+
+BUCKETS = ["5pct", "10pct", "15pct", "20pct"]
+GROUPS = [
+    ("length_side", "bbox_x_ratio", "Side View: Length Error"),
+    ("width_front", "bbox_y_ratio", "Front View: Width Error"),
+    ("height_rear", "bbox_z_ratio", "Rear View: Height Error"),
+]
+
+SOURCE_METHODS = {
+    "gd_mindrag": CD / "correct/compare/gd",
+    "dflow_mindrag": CD / "correct/compare/dflow",
+    "ictm_mindrag": CD / "correct/compare/ictm",
+    "flowcorrect_mindrag": CD / "correct/compare/flowcorrect",
+}
+PAIR_METHODS = {
+    "input_cond_CONDITION": CD / "input_cond/results/full_dataset_bigbkt_0817",
+    "cfg:constant": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_constant_s2p0",
+    "cfg:linear": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_linear_s2p0",
+    "cfg:exponential": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_exponential_s2p0",
+    "cfg:APG": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_APG_s2p0",
+    "cfg:zero_star": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_zero_star_s2p0",
+    "cfg:rectified_pp": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_rectified_pp_s2p0",
+    "cfg:SMC": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_SMC_s2p0",
+    "joint_LOSS": JOINT_ROOT,
+}
+METHODS = [
+    "gd_mindrag", "dflow_mindrag", "ictm_mindrag",
+    "flowcorrect_mindrag", "input_cond_CONDITION", "cfg:constant",
+    "cfg:linear", "cfg:exponential", "cfg:APG", "cfg:zero_star",
+    "cfg:rectified_pp", "cfg:SMC", "joint_LOSS",
+]
+LABELS = {
+    "gd_mindrag": "GD",
+    "dflow_mindrag": "D-Flow",
+    "ictm_mindrag": "ICTM",
+    "flowcorrect_mindrag": "FCSO",
+    "input_cond_CONDITION": "Input Cond.",
+    "cfg:constant": "Constant",
+    "cfg:linear": "Linear Schedule",
+    "cfg:exponential": "Exponential Schedule",
+    "cfg:APG": "APG",
+    "cfg:zero_star": r"CFG-Zero\star$",
+    "cfg:rectified_pp": "Rectified-CFG++",
+    "cfg:SMC": "SMC",
+    "joint_LOSS": "AnchorGen (Ours)",
+}
+
+
+def load_json(path: Path):
+    with path.open() as f:
+        return json.load(f)
+
+
+def pair_json_path(root: Path, row):
+    return (
+        root
+        / row["group"]
+        / row["source_uid"]
+        / (row["direction"] + "__" + row["target_uid"])
+        / "drag_result.json"
+    )
+
+
+def build_curves():
+    with PAIR_CSV.open(newline="") as f:
+        rows = list(csv.DictReader(f))
+
+    group_key = {group: ratio_key for group, ratio_key, _ in GROUPS}
+    raw = defaultdict(list)
+    source_cache = {}
+    missing = []
+
+    for row in rows:
+        group = row["group"]
+        ratio_key = group_key[group]
+        joint_path = pair_json_path(JOINT_ROOT, row)
+        if not joint_path.is_file():
+            missing.append(("joint_LOSS", group, row["source_uid"], "all"))
+            continue
+
+        joint = load_json(joint_path)
+        source_to_target = joint.get("k0_source_vs_target_metrics", {}).get(ratio_key)
+        if source_to_target is None:
+            missing.append(("source_to_target", group, row["source_uid"], ratio_key))
+            continue
+        source_to_target = float(source_to_target)
+
+        for method, root in PAIR_METHODS.items():
+            data = joint if method == "joint_LOSS" else None
+            if data is None:
+                path = pair_json_path(root, row)
+                if not path.is_file():
+                    missing.append((method, group, row["source_uid"], "all"))
+                    continue
+                data = load_json(path)
+            for bucket in BUCKETS:
+                ratio = (
+                    data.get("buckets", {})
+                    .get(bucket, {})
+                    .get("metrics", {})
+                    .get(ratio_key)
+                )
+                if ratio is not None:
+                    raw[(method, group, bucket)].append(abs(float(ratio) - 1.0) * 100.0)
+
+        for method, root in SOURCE_METHODS.items():
+            cache_key = (method, row["source_uid"])
+            if cache_key not in source_cache:
+                path = root / row["source_uid"] / "drag_result.json"
+                source_cache[cache_key] = load_json(path) if path.is_file() else None
+            data = source_cache[cache_key]
+            if data is None:
+                missing.append((method, group, row["source_uid"], "all"))
+                continue
+            for bucket in BUCKETS:
+                if method == "ictm_mindrag" and bucket == "20pct":
+                    continue
+                pred_to_source = (
+                    data.get("buckets", {})
+                    .get(bucket, {})
+                    .get("metrics_vs_source", {})
+                    .get(ratio_key)
+                )
+                if pred_to_source is not None:
+                    pred_to_target = float(pred_to_source) * source_to_target
+                    raw[(method, group, bucket)].append(abs(pred_to_target - 1.0) * 100.0)
+
+    curves = {}
+    output_rows = []
+    for method in METHODS:
+        for group, _, _ in GROUPS:
+            for bucket in BUCKETS:
+                values = raw.get((method, group, bucket), [])
+                if not values:
+                    continue
+                mean_error = sum(values) / len(values)
+                curves[(method, group, bucket)] = mean_error
+                output_rows.append(
+                    {
+                        "method": LABELS[method],
+                        "group": group,
+                        "bucket": bucket,
+                        "n_matched": len(values),
+                        "mean_relative_dimension_error_pct": mean_error,
+                    }
+                )
+
+    OUT_DIR.mkdir(parents=True, exist_ok=True)
+    csv_path = OUT_DIR / "dimension_error_curves.csv"
+    with csv_path.open("w", newline="") as f:
+        writer = csv.DictWriter(
+            f,
+            fieldnames=[
+                "method", "group", "bucket", "n_matched",
+                "mean_relative_dimension_error_pct",
+            ],
+        )
+        writer.writeheader()
+        writer.writerows(output_rows)
+
+    print("pairs:", len(rows), "missing:", len(missing))
+    if missing:
+        print("missing sample:", missing[:10])
+    print("wrote", csv_path)
+    return curves
+
+
+def plot(curves):
+    fig, axes = plt.subplots(2, 2, figsize=(13.8, 9.4), sharex=True)
+    controlled_groups = [group for group, _, _ in GROUPS]
+    panels = [
+        ("ALL", "Overall: Average Dimension Error"),
+        ("width_front", "Front View: Width Error"),
+        ("height_rear", "Rear View: Height Error"),
+        ("length_side", "Side View: Length Error"),
+    ]
+
+    for ax, (panel_group, title) in zip(axes.flat, panels):
+        for method in METHODS:
+            points = []
+            for bucket in BUCKETS:
+                if panel_group == "ALL":
+                    group_values = [
+                        curves.get((method, group, bucket))
+                        for group in controlled_groups
+                    ]
+                    if any(value is None for value in group_values):
+                        value = None
+                    else:
+                        value = sum(group_values) / len(group_values)
+                else:
+                    value = curves.get((method, panel_group, bucket))
+
+                if value is not None:
+                    points.append((float(bucket.replace("pct", "")), value))
+
+            if not points:
+                continue
+            _, color, linestyle, linewidth, marker, zorder = P.STYLE[method]
+            xs, ys = zip(*points)
+            ax.plot(
+                xs,
+                ys,
+                color=color,
+                linestyle=linestyle,
+                linewidth=linewidth,
+                marker=marker,
+                markersize=6 if linewidth > 2 else 4,
+                label=LABELS[method],
+                zorder=zorder,
+            )
+
+        ax.set_title(title + " (lower is better)", fontsize=11)
+        ax.set_xlabel("Target drag reduction (%)", fontsize=10)
+        ax.set_ylabel("Mean relative dimension error (%)", fontsize=9)
+        ax.set_xticks([5, 10, 15, 20])
+        ax.set_xlim(3.5, 21.5)
+        ax.grid(alpha=0.25)
+
+    handles, labels = axes.flat[0].get_legend_handles_labels()
+    fig.legend(
+        handles,
+        labels,
+        loc="lower center",
+        ncol=7,
+        fontsize=8.4,
+        frameon=False,
+        bbox_to_anchor=(0.5, -0.005),
+    )
+    fig.suptitle(
+        "Target-relative controlled-dimension error across drag-reduction levels",
+        fontsize=14,
+    )
+    fig.tight_layout(rect=[0, 0.12, 1, 0.955])
+
+    out_png = OUT_DIR / "final_dimension_errors.png"
+    fig.savefig(out_png, dpi=180, bbox_inches="tight")
+    plt.close(fig)
+    print("wrote", out_png)
+
+
+def main():
+    plot(build_curves())
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves_paper.py b/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves_paper.py
new file mode 100644
index 0000000000000000000000000000000000000000..ec3987eb3e2252c86bd04bb8e34451b17689aa41
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves_paper.py
@@ -0,0 +1,327 @@
+#!/usr/bin/env python3
+"""Plot target-relative controlled-dimension error at 5/10/15/20% drag.
+
+The metric is shape-level:
+    100 * abs(D_pred - D_target) / D_target
+where D is x-extent (Side/Length), y-extent (Front/Width), or z-extent
+(Rear/Height). Each plotted point is the mean over the fixed matched pairs.
+
+FMG and SGO are excluded. ICTM is shown only at 5/10/15%, where coverage is
+complete; its incomplete 20% success subset is intentionally not plotted.
+"""
+from __future__ import annotations
+
+import csv
+import json
+from collections import defaultdict
+from pathlib import Path
+
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+
+import plot_all_methods_curves as P
+
+
+HERE = Path(__file__).resolve().parent
+CD = HERE.parent.parent
+PAIR_CSV = CD / "cases/full_dataset/pairs.csv"
+JOINT_ROOT = CD / "deform/drag/runs/full_dataset_bigbkt_0817"
+OUT_DIR = HERE / "figs_final"
+
+BUCKETS = ["5pct", "10pct", "15pct", "20pct"]
+GROUPS = [
+    ("length_side", "bbox_x_ratio", "Side View: Length Error"),
+    ("width_front", "bbox_y_ratio", "Front View: Width Error"),
+    ("height_rear", "bbox_z_ratio", "Rear View: Height Error"),
+]
+
+SOURCE_METHODS = {
+    "gd_mindrag": CD / "correct/compare/gd",
+    "dflow_mindrag": CD / "correct/compare/dflow",
+    "ictm_mindrag": CD / "correct/compare/ictm",
+    "flowcorrect_mindrag": CD / "correct/compare/flowcorrect",
+}
+PAIR_METHODS = {
+    "input_cond_CONDITION": CD / "input_cond/results/full_dataset_bigbkt_0817",
+    "cfg:constant": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_constant_s2p0",
+    "cfg:linear": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_linear_s2p0",
+    "cfg:exponential": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_exponential_s2p0",
+    "cfg:APG": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_APG_s2p0",
+    "cfg:zero_star": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_zero_star_s2p0",
+    "cfg:rectified_pp": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_rectified_pp_s2p0",
+    "cfg:SMC": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_SMC_s2p0",
+    "joint_LOSS": JOINT_ROOT,
+}
+METHODS = [
+    "gd_mindrag", "dflow_mindrag", "ictm_mindrag",
+    "flowcorrect_mindrag", "input_cond_CONDITION", "cfg:constant",
+    "cfg:linear", "cfg:exponential", "cfg:APG", "cfg:zero_star",
+    "cfg:rectified_pp", "cfg:SMC", "joint_LOSS",
+]
+LABELS = {
+    "gd_mindrag": "GD",
+    "dflow_mindrag": "D-Flow",
+    "ictm_mindrag": "ICTM",
+    "flowcorrect_mindrag": "FCSO",
+    "input_cond_CONDITION": "Input Cond.",
+    "cfg:constant": "Constant",
+    "cfg:linear": "Linear Schedule",
+    "cfg:exponential": "Exponential Schedule",
+    "cfg:APG": "APG",
+    "cfg:zero_star": r"CFG-Zero$^\star$",
+    "cfg:rectified_pp": "Rectified-CFG++",
+    "cfg:SMC": "SMC",
+    "joint_LOSS": "AnchorGen (Ours)",
+}
+
+plt.rcParams.update(
+    {
+        "font.size": 15,
+        "font.weight": "bold",
+        "axes.labelsize": 17,
+        "axes.labelweight": "bold",
+        "axes.titlesize": 18,
+        "axes.titleweight": "bold",
+        "xtick.labelsize": 15,
+        "ytick.labelsize": 15,
+        "legend.fontsize": 13,
+    }
+)
+
+
+def load_json(path: Path):
+    with path.open() as f:
+        return json.load(f)
+
+
+def pair_json_path(root: Path, row):
+    return (
+        root
+        / row["group"]
+        / row["source_uid"]
+        / (row["direction"] + "__" + row["target_uid"])
+        / "drag_result.json"
+    )
+
+
+def build_curves():
+    with PAIR_CSV.open(newline="") as f:
+        rows = list(csv.DictReader(f))
+
+    group_key = {group: ratio_key for group, ratio_key, _ in GROUPS}
+    raw = defaultdict(list)
+    source_cache = {}
+    missing = []
+
+    for row in rows:
+        group = row["group"]
+        ratio_key = group_key[group]
+        joint_path = pair_json_path(JOINT_ROOT, row)
+        if not joint_path.is_file():
+            missing.append(("joint_LOSS", group, row["source_uid"], "all"))
+            continue
+
+        joint = load_json(joint_path)
+        source_to_target = joint.get("k0_source_vs_target_metrics", {}).get(ratio_key)
+        if source_to_target is None:
+            missing.append(("source_to_target", group, row["source_uid"], ratio_key))
+            continue
+        source_to_target = float(source_to_target)
+
+        for method, root in PAIR_METHODS.items():
+            data = joint if method == "joint_LOSS" else None
+            if data is None:
+                path = pair_json_path(root, row)
+                if not path.is_file():
+                    missing.append((method, group, row["source_uid"], "all"))
+                    continue
+                data = load_json(path)
+            for bucket in BUCKETS:
+                ratio = (
+                    data.get("buckets", {})
+                    .get(bucket, {})
+                    .get("metrics", {})
+                    .get(ratio_key)
+                )
+                if ratio is not None:
+                    raw[(method, group, bucket)].append(abs(float(ratio) - 1.0) * 100.0)
+
+        for method, root in SOURCE_METHODS.items():
+            cache_key = (method, row["source_uid"])
+            if cache_key not in source_cache:
+                path = root / row["source_uid"] / "drag_result.json"
+                source_cache[cache_key] = load_json(path) if path.is_file() else None
+            data = source_cache[cache_key]
+            if data is None:
+                missing.append((method, group, row["source_uid"], "all"))
+                continue
+            for bucket in BUCKETS:
+                if method == "ictm_mindrag" and bucket == "20pct":
+                    continue
+                pred_to_source = (
+                    data.get("buckets", {})
+                    .get(bucket, {})
+                    .get("metrics_vs_source", {})
+                    .get(ratio_key)
+                )
+                if pred_to_source is not None:
+                    pred_to_target = float(pred_to_source) * source_to_target
+                    raw[(method, group, bucket)].append(abs(pred_to_target - 1.0) * 100.0)
+
+    curves = {}
+    output_rows = []
+    for method in METHODS:
+        for group, _, _ in GROUPS:
+            for bucket in BUCKETS:
+                values = raw.get((method, group, bucket), [])
+                if not values:
+                    continue
+                mean_error = sum(values) / len(values)
+                curves[(method, group, bucket)] = mean_error
+                output_rows.append(
+                    {
+                        "method": LABELS[method],
+                        "group": group,
+                        "bucket": bucket,
+                        "n_matched": len(values),
+                        "mean_relative_dimension_error_pct": mean_error,
+                    }
+                )
+
+    OUT_DIR.mkdir(parents=True, exist_ok=True)
+    csv_path = OUT_DIR / "dimension_error_curves.csv"
+    with csv_path.open("w", newline="") as f:
+        writer = csv.DictWriter(
+            f,
+            fieldnames=[
+                "method", "group", "bucket", "n_matched",
+                "mean_relative_dimension_error_pct",
+            ],
+        )
+        writer.writeheader()
+        writer.writerows(output_rows)
+
+    print("pairs:", len(rows), "missing:", len(missing))
+    if missing:
+        print("missing sample:", missing[:10])
+    print("wrote", csv_path)
+    return curves
+
+
+def load_cached_curves():
+    """Load the frozen aggregate CSV for layout-only re-rendering."""
+    csv_path = OUT_DIR / "dimension_error_curves.csv"
+    label_to_method = {label: method for method, label in LABELS.items()}
+    # The frozen CSV predates the corrected math-text label.
+    label_to_method[r"CFG-Zero\star$"] = "cfg:zero_star"
+
+    curves = {}
+    with csv_path.open(newline="") as f:
+        for row in csv.DictReader(f):
+            method = label_to_method[row["method"]]
+            curves[(method, row["group"], row["bucket"])] = float(
+                row["mean_relative_dimension_error_pct"]
+            )
+    return curves
+
+
+def plot(curves):
+    fig, axes = plt.subplots(2, 2, figsize=(15.8, 10.8), sharex=True)
+    controlled_groups = [group for group, _, _ in GROUPS]
+    panels = [
+        ("ALL", "Average Error"),
+        ("width_front", "Width Error"),
+        ("height_rear", "Height Error"),
+        ("length_side", "Length Error"),
+    ]
+
+    for panel_index, (ax, (panel_group, title)) in enumerate(
+        zip(axes.flat, panels)
+    ):
+        row_index, column_index = divmod(panel_index, 2)
+        for method in METHODS:
+            points = []
+            for bucket in BUCKETS:
+                if panel_group == "ALL":
+                    group_values = [
+                        curves.get((method, group, bucket))
+                        for group in controlled_groups
+                    ]
+                    if any(value is None for value in group_values):
+                        value = None
+                    else:
+                        value = sum(group_values) / len(group_values)
+                else:
+                    value = curves.get((method, panel_group, bucket))
+
+                if value is not None:
+                    points.append((float(bucket.replace("pct", "")), value))
+
+            if not points:
+                continue
+            _, color, linestyle, linewidth, marker, zorder = P.STYLE[method]
+            xs, ys = zip(*points)
+            ax.plot(
+                xs,
+                ys,
+                color=color,
+                linestyle=linestyle,
+                linewidth=linewidth * 1.15,
+                marker=marker,
+                markersize=7.5 if linewidth > 2 else 5.5,
+                label=LABELS[method],
+                zorder=zorder,
+            )
+
+        ax.set_title(title, pad=11, fontweight="bold")
+        if row_index == 1:
+            ax.set_xlabel("Target Drag Reduction (%)", labelpad=9)
+        else:
+            ax.set_xlabel("")
+        if column_index == 0:
+            ax.set_ylabel("Relative Error (%)", labelpad=10)
+        else:
+            ax.set_ylabel("")
+        ax.set_xticks([5, 10, 15, 20])
+        ax.set_xlim(3.5, 21.5)
+        ax.grid(alpha=0.25)
+        ax.tick_params(axis="both", which="major", labelsize=15, width=1.4)
+        for tick_label in [*ax.get_xticklabels(), *ax.get_yticklabels()]:
+            tick_label.set_fontweight("bold")
+
+    handles, labels = axes.flat[0].get_legend_handles_labels()
+    fig.legend(
+        handles,
+        labels,
+        loc="lower center",
+        ncol=5,
+        prop={"size": 13, "weight": "bold"},
+        frameon=False,
+        bbox_to_anchor=(0.5, -0.01),
+        columnspacing=1.5,
+        handletextpad=0.6,
+    )
+    fig.suptitle(
+        "Target-Relative Controlled-Dimension Error Across Drag-Reduction Levels",
+        fontsize=21,
+        fontweight="bold",
+        y=0.985,
+    )
+    fig.tight_layout(rect=[0, 0.18, 1, 0.945], h_pad=2.0, w_pad=2.2)
+
+    out_png = OUT_DIR / "dimension_error_curves_paper.png"
+    out_pdf = OUT_DIR / "dimension_error_curves_paper.pdf"
+    fig.savefig(out_png, dpi=220, bbox_inches="tight")
+    fig.savefig(out_pdf, bbox_inches="tight")
+    plt.close(fig)
+    print("wrote", out_png)
+    print("wrote", out_pdf)
+
+
+def main():
+    plot(load_cached_curves())
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves_paper_legend2row.py b/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves_paper_legend2row.py
new file mode 100644
index 0000000000000000000000000000000000000000..f9da1c043c7dc2af7aca5db4eb6b483f31db2e84
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_dimension_error_curves_paper_legend2row.py
@@ -0,0 +1,332 @@
+#!/usr/bin/env python3
+"""Plot target-relative controlled-dimension error at 5/10/15/20% drag.
+
+The metric is shape-level:
+    100 * abs(D_pred - D_target) / D_target
+where D is x-extent (Side/Length), y-extent (Front/Width), or z-extent
+(Rear/Height). Each plotted point is the mean over the fixed matched pairs.
+
+FMG and SGO are excluded. ICTM is shown only at 5/10/15%, where coverage is
+complete; its incomplete 20% success subset is intentionally not plotted.
+"""
+from __future__ import annotations
+
+import csv
+import json
+from collections import defaultdict
+from pathlib import Path
+
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+from matplotlib.ticker import FormatStrFormatter
+
+import plot_all_methods_curves as P
+
+
+HERE = Path(__file__).resolve().parent
+CD = HERE.parent.parent
+PAIR_CSV = CD / "cases/full_dataset/pairs.csv"
+JOINT_ROOT = CD / "deform/drag/runs/full_dataset_bigbkt_0817"
+OUT_DIR = HERE / "figs_final"
+
+BUCKETS = ["5pct", "10pct", "15pct", "20pct"]
+GROUPS = [
+    ("length_side", "bbox_x_ratio", "Side View: Length Error"),
+    ("width_front", "bbox_y_ratio", "Front View: Width Error"),
+    ("height_rear", "bbox_z_ratio", "Rear View: Height Error"),
+]
+
+SOURCE_METHODS = {
+    "gd_mindrag": CD / "correct/compare/gd",
+    "dflow_mindrag": CD / "correct/compare/dflow",
+    "ictm_mindrag": CD / "correct/compare/ictm",
+    "flowcorrect_mindrag": CD / "correct/compare/flowcorrect",
+}
+PAIR_METHODS = {
+    "input_cond_CONDITION": CD / "input_cond/results/full_dataset_bigbkt_0817",
+    "cfg:constant": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_constant_s2p0",
+    "cfg:linear": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_linear_s2p0",
+    "cfg:exponential": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_exponential_s2p0",
+    "cfg:APG": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_APG_s2p0",
+    "cfg:zero_star": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_zero_star_s2p0",
+    "cfg:rectified_pp": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_rectified_pp_s2p0",
+    "cfg:SMC": CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/cfg_SMC_s2p0",
+    "joint_LOSS": JOINT_ROOT,
+}
+METHODS = [
+    "gd_mindrag", "dflow_mindrag", "ictm_mindrag",
+    "flowcorrect_mindrag", "input_cond_CONDITION", "cfg:constant",
+    "cfg:linear", "cfg:exponential", "cfg:APG", "cfg:zero_star",
+    "cfg:rectified_pp", "cfg:SMC", "joint_LOSS",
+]
+LABELS = {
+    "gd_mindrag": "GD",
+    "dflow_mindrag": "D-Flow",
+    "ictm_mindrag": "ICTM",
+    "flowcorrect_mindrag": "FCSO",
+    "input_cond_CONDITION": "Input Cond.",
+    "cfg:constant": "Constant",
+    "cfg:linear": "Linear Schedule",
+    "cfg:exponential": "Exponential Schedule",
+    "cfg:APG": "APG",
+    "cfg:zero_star": r"CFG-Zero$^\star$",
+    "cfg:rectified_pp": "Rectified-CFG++",
+    "cfg:SMC": "SMC",
+    "joint_LOSS": "AnchorGen (Ours)",
+}
+
+plt.rcParams.update(
+    {
+        "font.size": 15,
+        "font.weight": "bold",
+        "axes.labelsize": 17,
+        "axes.labelweight": "bold",
+        "axes.titlesize": 18,
+        "axes.titleweight": "bold",
+        "xtick.labelsize": 15,
+        "ytick.labelsize": 15,
+        "legend.fontsize": 13,
+    }
+)
+
+
+def load_json(path: Path):
+    with path.open() as f:
+        return json.load(f)
+
+
+def pair_json_path(root: Path, row):
+    return (
+        root
+        / row["group"]
+        / row["source_uid"]
+        / (row["direction"] + "__" + row["target_uid"])
+        / "drag_result.json"
+    )
+
+
+def build_curves():
+    with PAIR_CSV.open(newline="") as f:
+        rows = list(csv.DictReader(f))
+
+    group_key = {group: ratio_key for group, ratio_key, _ in GROUPS}
+    raw = defaultdict(list)
+    source_cache = {}
+    missing = []
+
+    for row in rows:
+        group = row["group"]
+        ratio_key = group_key[group]
+        joint_path = pair_json_path(JOINT_ROOT, row)
+        if not joint_path.is_file():
+            missing.append(("joint_LOSS", group, row["source_uid"], "all"))
+            continue
+
+        joint = load_json(joint_path)
+        source_to_target = joint.get("k0_source_vs_target_metrics", {}).get(ratio_key)
+        if source_to_target is None:
+            missing.append(("source_to_target", group, row["source_uid"], ratio_key))
+            continue
+        source_to_target = float(source_to_target)
+
+        for method, root in PAIR_METHODS.items():
+            data = joint if method == "joint_LOSS" else None
+            if data is None:
+                path = pair_json_path(root, row)
+                if not path.is_file():
+                    missing.append((method, group, row["source_uid"], "all"))
+                    continue
+                data = load_json(path)
+            for bucket in BUCKETS:
+                ratio = (
+                    data.get("buckets", {})
+                    .get(bucket, {})
+                    .get("metrics", {})
+                    .get(ratio_key)
+                )
+                if ratio is not None:
+                    raw[(method, group, bucket)].append(abs(float(ratio) - 1.0) * 100.0)
+
+        for method, root in SOURCE_METHODS.items():
+            cache_key = (method, row["source_uid"])
+            if cache_key not in source_cache:
+                path = root / row["source_uid"] / "drag_result.json"
+                source_cache[cache_key] = load_json(path) if path.is_file() else None
+            data = source_cache[cache_key]
+            if data is None:
+                missing.append((method, group, row["source_uid"], "all"))
+                continue
+            for bucket in BUCKETS:
+                if method == "ictm_mindrag" and bucket == "20pct":
+                    continue
+                pred_to_source = (
+                    data.get("buckets", {})
+                    .get(bucket, {})
+                    .get("metrics_vs_source", {})
+                    .get(ratio_key)
+                )
+                if pred_to_source is not None:
+                    pred_to_target = float(pred_to_source) * source_to_target
+                    raw[(method, group, bucket)].append(abs(pred_to_target - 1.0) * 100.0)
+
+    curves = {}
+    output_rows = []
+    for method in METHODS:
+        for group, _, _ in GROUPS:
+            for bucket in BUCKETS:
+                values = raw.get((method, group, bucket), [])
+                if not values:
+                    continue
+                mean_error = sum(values) / len(values)
+                curves[(method, group, bucket)] = mean_error
+                output_rows.append(
+                    {
+                        "method": LABELS[method],
+                        "group": group,
+                        "bucket": bucket,
+                        "n_matched": len(values),
+                        "mean_relative_dimension_error_pct": mean_error,
+                    }
+                )
+
+    OUT_DIR.mkdir(parents=True, exist_ok=True)
+    csv_path = OUT_DIR / "dimension_error_curves.csv"
+    with csv_path.open("w", newline="") as f:
+        writer = csv.DictWriter(
+            f,
+            fieldnames=[
+                "method", "group", "bucket", "n_matched",
+                "mean_relative_dimension_error_pct",
+            ],
+        )
+        writer.writeheader()
+        writer.writerows(output_rows)
+
+    print("pairs:", len(rows), "missing:", len(missing))
+    if missing:
+        print("missing sample:", missing[:10])
+    print("wrote", csv_path)
+    return curves
+
+
+def load_cached_curves():
+    """Load the frozen aggregate CSV for layout-only re-rendering."""
+    csv_path = OUT_DIR / "dimension_error_curves.csv"
+    label_to_method = {label: method for method, label in LABELS.items()}
+    # The frozen CSV predates the corrected math-text label.
+    label_to_method[r"CFG-Zero\star$"] = "cfg:zero_star"
+
+    curves = {}
+    with csv_path.open(newline="") as f:
+        for row in csv.DictReader(f):
+            method = label_to_method[row["method"]]
+            curves[(method, row["group"], row["bucket"])] = float(
+                row["mean_relative_dimension_error_pct"]
+            )
+    return curves
+
+
+def plot(curves):
+    fig, axes = plt.subplots(2, 2, figsize=(17.2, 11.2), sharex=True)
+    controlled_groups = [group for group, _, _ in GROUPS]
+    panels = [
+        ("ALL", "Average Error"),
+        ("width_front", "Width Error"),
+        ("height_rear", "Height Error"),
+        ("length_side", "Length Error"),
+    ]
+
+    for panel_index, (ax, (panel_group, title)) in enumerate(
+        zip(axes.flat, panels)
+    ):
+        row_index, column_index = divmod(panel_index, 2)
+        for method in METHODS:
+            points = []
+            for bucket in BUCKETS:
+                if panel_group == "ALL":
+                    group_values = [
+                        curves.get((method, group, bucket))
+                        for group in controlled_groups
+                    ]
+                    if any(value is None for value in group_values):
+                        value = None
+                    else:
+                        value = sum(group_values) / len(group_values)
+                else:
+                    value = curves.get((method, panel_group, bucket))
+
+                if value is not None:
+                    points.append((float(bucket.replace("pct", "")), value))
+
+            if not points:
+                continue
+            _, color, linestyle, linewidth, marker, zorder = P.STYLE[method]
+            xs, ys = zip(*points)
+            ax.plot(
+                xs,
+                ys,
+                color=color,
+                linestyle=linestyle,
+                linewidth=linewidth * 1.15,
+                marker=marker,
+                markersize=7.5 if linewidth > 2 else 5.5,
+                label=LABELS[method],
+                zorder=zorder,
+            )
+
+        ax.set_title(title, pad=11, fontweight="bold")
+        if row_index == 1:
+            ax.set_xlabel("Target Drag Reduction (%)", labelpad=9)
+        else:
+            ax.set_xlabel("")
+        if column_index == 0:
+            ax.set_ylabel("Relative Error (%)", labelpad=10)
+        else:
+            ax.set_ylabel("")
+        ax.set_xticks([5, 10, 15, 20])
+        ax.set_xlim(3.5, 21.5)
+        ax.yaxis.set_major_formatter(FormatStrFormatter("%.1f"))
+        ax.grid(alpha=0.25)
+        ax.tick_params(axis="both", which="major", labelsize=15, width=1.4)
+        for tick_label in [*ax.get_xticklabels(), *ax.get_yticklabels()]:
+            tick_label.set_fontweight("bold")
+
+    handles, labels = axes.flat[0].get_legend_handles_labels()
+    fig.legend(
+        handles,
+        labels,
+        loc="lower center",
+        ncol=7,
+        prop={"size": 17, "weight": "bold"},
+        frameon=False,
+        bbox_to_anchor=(0.5, -0.005),
+        columnspacing=0.8,
+        handletextpad=0.4,
+        handlelength=1.55,
+        labelspacing=0.65,
+        markerscale=1.35,
+    )
+    fig.suptitle(
+        "Target-Relative Controlled-Dimension Error Across Drag-Reduction Levels",
+        fontsize=21,
+        fontweight="bold",
+        y=0.985,
+    )
+    fig.tight_layout(rect=[0, 0.15, 1, 0.945], h_pad=2.0, w_pad=2.2)
+
+    out_png = OUT_DIR / "dimension_error_curves_paper_legend2row.png"
+    out_pdf = OUT_DIR / "dimension_error_curves_paper_legend2row.pdf"
+    fig.savefig(out_png, dpi=220, bbox_inches="tight")
+    fig.savefig(out_pdf, bbox_inches="tight")
+    plt.close(fig)
+    print("wrote", out_png)
+    print("wrote", out_pdf)
+
+
+def main():
+    plot(load_cached_curves())
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_final_curves.py b/anchorgen_baselines_20260923/input_cond/vis/plot_final_curves.py
new file mode 100644
index 0000000000000000000000000000000000000000..04be352449951cdc0e3b98a8a60a4b3dcfd14269
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_final_curves.py
@@ -0,0 +1,94 @@
+#!/usr/bin/env python3
+"""Paper-ready ALL-group comparison using the six metrics in the main table.
+
+All plotted values are scaled by 100. SGO is intentionally excluded because its
+off-manifold geometric collapse makes the deep-bucket curves non-comparable.
+Output: input_cond/vis/figs_final/final_ALL.png.
+"""
+from pathlib import Path
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+import plot_all_methods_curves as P
+
+HERE = Path(__file__).resolve().parent
+CD = HERE.parent.parent
+
+DEEP = CD / "anchors/loss_vs_condition_deepbucket_0817/DEEPBUCKET_METRICS.csv"
+CFG = CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/SUMMARY_cfg_variants.csv"
+MINDRAG = {
+    "gd_mindrag": CD / "correct/compare/GD_VS_TARGET_SUMMARY.csv",
+    "flowcorrect_mindrag": CD / "correct/compare/FLOWCORRECT_VS_TARGET_SUMMARY.csv",
+    "fmg_mindrag": CD / "correct/compare/FMG_VS_TARGET_SUMMARY.csv", 
+    "ictm_mindrag": CD / "correct/compare/ICTM_VS_TARGET_SUMMARY.csv",
+    "dflow_mindrag": CD / "correct/compare/DFLOW_VS_TARGET_SUMMARY.csv",
+    "sgo_mindrag": CD / "correct/compare/SGO_VS_TARGET_SUMMARY.csv",
+}
+
+METHODS = [
+    "gd_mindrag", "dflow_mindrag", "ictm_mindrag",
+    "flowcorrect_mindrag", "input_cond_CONDITION", "cfg:constant", "cfg:linear",
+    "cfg:exponential", "cfg:APG", "cfg:zero_star", "cfg:rectified_pp", "cfg:SMC",
+    "joint_LOSS",
+]
+LABELS = {
+    "gd_mindrag": "GD", "fmg_mindrag": "FMG", "dflow_mindrag": "D-Flow",
+    "ictm_mindrag": "ICTM", "flowcorrect_mindrag": "FCSO",
+    "input_cond_CONDITION": "Input Cond.", "cfg:constant": "Constant",
+    "cfg:linear": "Linear Schedule", "cfg:exponential": "Exponential Schedule",
+    "cfg:APG": "APG", "cfg:zero_star": r"CFG-Zero$^\star$",
+    "cfg:rectified_pp": "Rectified-CFG++", "cfg:SMC": "SMC",
+    "joint_LOSS": "AnchorGen (Ours)",
+}
+METRICS = [
+    ("fscore_1pct", r"F@1% $\uparrow$"),
+    ("fscore_0p5pct", r"F@0.5% $\uparrow$"),
+    ("cd_l1_norm", r"CD-L1 $\downarrow$"),
+    ("volume_ratio_abs", r"Vol-Ratio $\uparrow$"),
+    ("normal_consistency", r"Normal-C $\uparrow$"),
+    ("hd95_norm", r"HD95 $\downarrow$"),
+]
+
+
+def plot_scope(data, scope, out_png):
+    fig, axes = plt.subplots(2, 3, figsize=(15.5, 8.6), sharex=True)
+    for ax, (mk, title) in zip(axes.flat, METRICS):
+        for method in METHODS:
+            if method not in data:
+                continue
+            _, color, ls, lw, mrk, z = P.STYLE[method]
+            xs, ys = P.curve(data, method, scope, mk)
+            if method == "ictm_mindrag":
+                pts = [(x, y) for x, y in zip(xs, ys) if x < 20]
+                xs, ys = zip(*pts) if pts else ([], [])
+            if xs:
+                ys = [100.0 * y for y in ys]
+                ax.plot(xs, ys, color=color, linestyle=ls, lw=lw, marker=mrk,
+                        ms=6 if lw > 2 else 4, label=LABELS[method], zorder=z)
+        ax.set_title(title, fontsize=11)
+        ax.set_ylabel(r"Metric value ($\times 100$)", fontsize=9)
+        ax.set_xticks([5, 10, 15, 20])
+        ax.set_xlim(3.5, 21.5)
+        ax.grid(alpha=0.25)
+    for ax in axes[1]:
+        ax.set_xlabel("Target drag reduction (%)", fontsize=10)
+    handles, labels = axes[0, 0].get_legend_handles_labels()
+    fig.legend(handles, labels, loc="lower center", ncol=7, fontsize=8.4,
+               frameon=False, bbox_to_anchor=(0.5, -0.015))
+    fig.suptitle("ALL group ($n=600$): target-shape similarity across drag-reduction levels",
+                 fontsize=14)
+    fig.tight_layout(rect=[0, 0.105, 1, 0.955])
+    out_png.parent.mkdir(parents=True, exist_ok=True)
+    fig.savefig(out_png, dpi=180, bbox_inches="tight")
+    plt.close(fig)
+    print("wrote", out_png)
+
+
+def main():
+    data = P.load(str(DEEP), str(CFG), {k: str(v) for k, v in MINDRAG.items()})
+    out = HERE / "figs_final"
+    plot_scope(data, "ALL", out / "final_ALL.png")
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_final_curves_paper_legend2row.py b/anchorgen_baselines_20260923/input_cond/vis/plot_final_curves_paper_legend2row.py
new file mode 100644
index 0000000000000000000000000000000000000000..2373ea2c61ea573289ee3bb44bca6e372b2f8483
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_final_curves_paper_legend2row.py
@@ -0,0 +1,147 @@
+#!/usr/bin/env python3
+"""Large-type paper-ready ALL-group comparison for the six table metrics.
+
+All plotted values are scaled by 100. SGO is intentionally excluded because its
+off-manifold geometric collapse makes the deep-bucket curves non-comparable.
+The plotting data and method ordering are unchanged from plot_final_curves.py.
+"""
+from pathlib import Path
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+from matplotlib.ticker import FormatStrFormatter
+import plot_all_methods_curves as P
+
+HERE = Path(__file__).resolve().parent
+CD = HERE.parent.parent
+
+DEEP = CD / "anchors/loss_vs_condition_deepbucket_0817/DEEPBUCKET_METRICS.csv"
+CFG = CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/SUMMARY_cfg_variants.csv"
+MINDRAG = {
+    "gd_mindrag": CD / "correct/compare/GD_VS_TARGET_SUMMARY.csv",
+    "flowcorrect_mindrag": CD / "correct/compare/FLOWCORRECT_VS_TARGET_SUMMARY.csv",
+    "fmg_mindrag": CD / "correct/compare/FMG_VS_TARGET_SUMMARY.csv", 
+    "ictm_mindrag": CD / "correct/compare/ICTM_VS_TARGET_SUMMARY.csv",
+    "dflow_mindrag": CD / "correct/compare/DFLOW_VS_TARGET_SUMMARY.csv",
+    "sgo_mindrag": CD / "correct/compare/SGO_VS_TARGET_SUMMARY.csv",
+}
+
+METHODS = [
+    "gd_mindrag", "dflow_mindrag", "ictm_mindrag",
+    "flowcorrect_mindrag", "input_cond_CONDITION", "cfg:constant", "cfg:linear",
+    "cfg:exponential", "cfg:APG", "cfg:zero_star", "cfg:rectified_pp", "cfg:SMC",
+    "joint_LOSS",
+]
+LABELS = {
+    "gd_mindrag": "GD", "fmg_mindrag": "FMG", "dflow_mindrag": "D-Flow",
+    "ictm_mindrag": "ICTM", "flowcorrect_mindrag": "FCSO",
+    "input_cond_CONDITION": "Input Cond.", "cfg:constant": "Constant",
+    "cfg:linear": "Linear Schedule", "cfg:exponential": "Exponential Schedule",
+    "cfg:APG": "APG", "cfg:zero_star": r"CFG-Zero$^\star$",
+    "cfg:rectified_pp": "Rectified-CFG++", "cfg:SMC": "SMC",
+    "joint_LOSS": "AnchorGen (Ours)",
+}
+METRICS = [
+    ("fscore_1pct", r"F@1% $\uparrow$"),
+    ("fscore_0p5pct", r"F@0.5% $\uparrow$"),
+    ("cd_l1_norm", r"CD-L1 $\downarrow$"),
+    ("volume_ratio_abs", r"Vol-Ratio $\uparrow$"),
+    ("normal_consistency", r"Normal-C $\uparrow$"),
+    ("hd95_norm", r"HD95 $\downarrow$"),
+]
+
+plt.rcParams.update(
+    {
+        "font.size": 15,
+        "font.weight": "bold",
+        "axes.labelsize": 17,
+        "axes.labelweight": "bold",
+        "axes.titlesize": 18,
+        "axes.titleweight": "bold",
+        "xtick.labelsize": 15,
+        "ytick.labelsize": 15,
+        "legend.fontsize": 17,
+    }
+)
+
+
+def plot_scope(data, scope, out_png):
+    fig, axes = plt.subplots(2, 3, figsize=(19.2, 10.8), sharex=True)
+    for panel_index, (ax, (mk, title)) in enumerate(zip(axes.flat, METRICS)):
+        row_index, column_index = divmod(panel_index, 3)
+        for method in METHODS:
+            if method not in data:
+                continue
+            _, color, ls, lw, mrk, z = P.STYLE[method]
+            xs, ys = P.curve(data, method, scope, mk)
+            if method == "ictm_mindrag":
+                pts = [(x, y) for x, y in zip(xs, ys) if x < 20]
+                xs, ys = zip(*pts) if pts else ([], [])
+            if xs:
+                ys = [100.0 * y for y in ys]
+                ax.plot(
+                    xs,
+                    ys,
+                    color=color,
+                    linestyle=ls,
+                    linewidth=lw * 1.15,
+                    marker=mrk,
+                    markersize=7.5 if lw > 2 else 5.5,
+                    label=LABELS[method],
+                    zorder=z,
+                )
+        ax.set_title(title, pad=11, fontweight="bold")
+        if row_index == 1:
+            ax.set_xlabel("Target Drag Reduction (%)", labelpad=9)
+        else:
+            ax.set_xlabel("")
+        if column_index == 0:
+            ax.set_ylabel(r"Metric Value ($\times 100$)", labelpad=10)
+        else:
+            ax.set_ylabel("")
+        ax.set_xticks([5, 10, 15, 20])
+        ax.set_xlim(3.5, 21.5)
+        ax.yaxis.set_major_formatter(FormatStrFormatter("%.1f"))
+        ax.grid(alpha=0.25)
+        ax.tick_params(axis="both", which="major", labelsize=15, width=1.4)
+        for tick_label in [*ax.get_xticklabels(), *ax.get_yticklabels()]:
+            tick_label.set_fontweight("bold")
+    handles, labels = axes[0, 0].get_legend_handles_labels()
+    fig.legend(
+        handles,
+        labels,
+        loc="lower center",
+        ncol=7,
+        prop={"size": 17, "weight": "bold"},
+        frameon=False,
+        bbox_to_anchor=(0.5, -0.005),
+        columnspacing=0.8,
+        handletextpad=0.4,
+        handlelength=1.55,
+        labelspacing=0.65,
+        markerscale=1.35,
+    )
+    fig.suptitle(
+        "ALL Group ($n=600$): Target-Shape Similarity Across Drag-Reduction Levels",
+        fontsize=21,
+        fontweight="bold",
+        y=0.985,
+    )
+    fig.tight_layout(rect=[0, 0.15, 1, 0.945], h_pad=2.0, w_pad=2.2)
+    out_png.parent.mkdir(parents=True, exist_ok=True)
+    out_pdf = out_png.with_suffix(".pdf")
+    fig.savefig(out_png, dpi=220, bbox_inches="tight")
+    fig.savefig(out_pdf, bbox_inches="tight")
+    plt.close(fig)
+    print("wrote", out_png)
+    print("wrote", out_pdf)
+
+
+def main():
+    data = P.load(str(DEEP), str(CFG), {k: str(v) for k, v in MINDRAG.items()})
+    out = HERE / "figs_final"
+    plot_scope(data, "ALL", out / "shape_similarity_curves_paper_legend2row.png")
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/vis/plot_sgo_filtered.py b/anchorgen_baselines_20260923/input_cond/vis/plot_sgo_filtered.py
new file mode 100644
index 0000000000000000000000000000000000000000..c4783cd480c97fc55c166ec58669f586f31aaa39
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/vis/plot_sgo_filtered.py
@@ -0,0 +1,143 @@
+#!/usr/bin/env python3
+"""SGO 'full vs successful-only' analysis + curves. 'Successful' = the source's bucket mesh did NOT
+collapse (vs-SOURCE drift cd_l1_norm <= 0.03; normal min-drag ~0.015, collapse 0.05-0.09 — this also
+drops the '5%->34% jump' overshoots since a collapse step blows up drift). Filter applied to SGO only.
+
+Writes compare/SGO_SUCCESS_VS_TARGET_SUMMARY.csv (method=sgo_success, same schema as the other
+*_VS_TARGET_SUMMARY.csv) and prints the ALL/20pct full-vs-successful table + success fraction.
+Plots figs_final/final_sgo_filtered_{ALL,groups}.png: suitable methods + SGO(all, gray dashed) +
+SGO(successful, gray solid), 5 retained metrics, capped 20%. Run with dora python from input_cond/vis/.
+"""
+import json, glob, csv
+from collections import defaultdict
+from pathlib import Path
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+import plot_all_methods_curves as P
+
+HERE = Path(__file__).resolve().parent
+CD = HERE.parent.parent
+SGO = CD / "correct/compare/sgo"
+DRIFT_THRESH = 0.03
+MK = ["cd_l1_norm", "cd_l2_norm", "hd95_norm", "normal_consistency",
+      "fscore_1pct", "fscore_0p5pct", "surface_area_ratio", "volume_ratio_abs"]
+BUCKETS = ["5pct", "10pct", "15pct", "20pct"]
+GROUPS = ["ALL", "length_side", "width_front", "height_rear"]
+
+
+def mean(xs):
+    xs = [x for x in xs if x is not None]
+    return sum(xs) / len(xs) if xs else None
+
+
+def build():
+    # source-bucket vs-SOURCE drift = collapse signal
+    drift = {}
+    for jp in glob.glob(str(SGO / "*/drag_result.json")):
+        d = json.load(open(jp)); src = d["source_uid"]
+        for bk, c in d.get("buckets", {}).items():
+            m = (c.get("metrics_vs_source") or {})
+            if m.get("cd_l1_norm") is not None:
+                drift[(src, bk)] = m["cd_l1_norm"]
+    agg_all = defaultdict(lambda: defaultdict(list))
+    agg_suc = defaultdict(lambda: defaultdict(list))
+    for jp in glob.glob(str(SGO / "*/vstarget__*.json")):
+        d = json.load(open(jp)); src = d["source_uid"]; grp = d["group"]
+        for bk, c in d.get("buckets", {}).items():
+            m = c.get("metrics_vs_target")
+            if not m:
+                continue
+            item = (m, c.get("drag_drop_actual"))
+            agg_all[grp][bk].append(item); agg_all["ALL"][bk].append(item)
+            dv = drift.get((src, bk))
+            if dv is not None and dv <= DRIFT_THRESH:
+                agg_suc[grp][bk].append(item); agg_suc["ALL"][bk].append(item)
+    return agg_all, agg_suc
+
+
+def rows_of(agg, label):
+    rows = []
+    for g in GROUPS:
+        for bk in BUCKETS:
+            lst = agg[g][bk]
+            if not lst:
+                continue
+            cell = {k: mean([mm[0].get(k) for mm in lst]) for k in MK}
+            drag = mean([mm[1] for mm in lst])
+            rows.append([label, g, bk, len(lst)] + [cell[k] for k in MK] + [drag])
+    return rows
+
+
+def write_csv(rows, path):
+    with open(path, "w", newline="") as f:
+        w = csv.writer(f)
+        w.writerow(["method", "group", "bucket", "n_matched"] + MK + ["drag_drop_actual"])
+        w.writerows(rows)
+    print("wrote", path)
+
+
+def main():
+    agg_all, agg_suc = build()
+    all_rows = rows_of(agg_all, "sgo")
+    suc_rows = rows_of(agg_suc, "sgo_success")
+    write_csv(suc_rows, CD / "correct/compare/SGO_SUCCESS_VS_TARGET_SUMMARY.csv")
+    write_csv(all_rows, CD / "correct/compare/SGO_VS_TARGET_SUMMARY.csv")
+
+    def get(rows, g, bk):
+        for r in rows:
+            if r[1] == g and r[2] == bk:
+                return r
+        return None
+    print("\n=== SGO ALL @20pct: FULL vs SUCCESSFUL (drift<=%.2f) ===" % DRIFT_THRESH)
+    print("%-14s n    F@1%%   F@.5%%  CD-L1   HD95    Norm-C  drag%%" % "")
+    for lab, rows in [("full", all_rows), ("successful", suc_rows)]:
+        r = get(rows, "ALL", "20pct")
+        if r:
+            g = lambda k: r[4 + MK.index(k)] * 100
+            print("%-14s %-4d %.2f  %.2f  %.3f  %.3f  %.2f  %.1f"
+                  % (lab, r[3], g("fscore_1pct"), g("fscore_0p5pct"), g("cd_l1_norm"),
+                     g("hd95_norm"), g("normal_consistency"), r[-1] * 100))
+
+    # ---- plot: suitable methods + SGO(all) + SGO(successful) ----
+    data = P.load(str(CD / "anchors/loss_vs_condition_deepbucket_0817/DEEPBUCKET_METRICS.csv"),
+                  str(CD / "input_cond/cfg/cfg_results/full_dataset_cfg_0817/SUMMARY_cfg_variants.csv"),
+                  {"gd_mindrag": str(CD / "correct/compare/GD_VS_TARGET_SUMMARY.csv"),
+                   "flowcorrect_mindrag": str(CD / "correct/compare/FLOWCORRECT_VS_TARGET_SUMMARY.csv"),
+                   "ictm_mindrag": str(CD / "correct/compare/ICTM_VS_TARGET_SUMMARY.csv")})
+    for lab, rows in [("sgo_all", all_rows), ("sgo_success", suc_rows)]:
+        for r in rows:
+            cell = {k: r[4 + MK.index(k)] for k in MK}; cell["drag"] = r[-1]
+            data[lab][(r[1], r[2])] = cell
+    P.STYLE["sgo_all"] = ("SGO (all, incl. collapsed)", "#999999", ":", 2.2, "D", 6)
+    P.STYLE["sgo_success"] = ("SGO (successful only)", "#000000", "-", 2.2, "D", 7)
+    methods = ["joint_LOSS", "input_cond_CONDITION", "gd_mindrag", "flowcorrect_mindrag",
+               "ictm_mindrag", "sgo_all", "sgo_success"]
+    METRICS = [("fscore_1pct", "F-score @1%  up"), ("fscore_0p5pct", "F-score @0.5%  up"),
+               ("cd_l1_norm", "CD-L1 (norm)  down"), ("hd95_norm", "HD95 (norm)  down"),
+               ("normal_consistency", "Normal consistency  up")]
+    out = HERE / "figs_final"
+    for scope in GROUPS:
+        n = len(METRICS)
+        fig, axes = plt.subplots(1, n, figsize=(4.2 * n, 4.4))
+        for ax, (mk, title) in zip(axes, METRICS):
+            for m in methods:
+                if m not in data:
+                    continue
+                lab, col, ls, lw, mrk, z = P.STYLE[m]
+                xs, ys = P.curve(data, m, scope, mk)
+                if xs:
+                    ax.plot(xs, ys, color=col, linestyle=ls, lw=lw, marker=mrk, ms=6, label=lab, zorder=z)
+            ax.set_title(title, fontsize=11); ax.set_xlabel("drag-reduction bucket (%)", fontsize=9)
+            ax.set_xticks([5, 10, 15, 20]); ax.set_xlim(3.5, 21.5); ax.grid(alpha=0.3)
+        h, l = axes[0].get_legend_handles_labels()
+        fig.legend(h, l, loc="lower center", ncol=7, fontsize=8, frameon=False, bbox_to_anchor=(0.5, -0.02))
+        fig.suptitle("SGO full vs successful-only (drift<=0.03) + reference methods — %s" % scope, fontsize=12)
+        fig.tight_layout(rect=[0, 0.08, 1, 0.95])
+        out.mkdir(parents=True, exist_ok=True)
+        p = out / ("final_sgo_filtered_%s.png" % scope)
+        fig.savefig(p, dpi=130, bbox_inches="tight"); plt.close(fig); print("wrote", p)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/anchorgen_baselines_20260923/input_cond/watchdog.sh b/anchorgen_baselines_20260923/input_cond/watchdog.sh
new file mode 100644
index 0000000000000000000000000000000000000000..4466e254a9c156dcb40fe8eaa07f585baf064897
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/watchdog.sh
@@ -0,0 +1,18 @@
+#!/bin/bash
+# Poll all input_cond jobs matching STAMP every 300s; append status + latest step.
+# Detached: nohup ./watchdog.sh &
+P=cvlab-hantzhan
+STAMP=${STAMP:-0816c}
+LOG=/mnt/cvlab/scratch/cvlab/home/hantzhan/code/ICLR/customized-design/input_cond/watchdog_status.log
+for i in $(seq 1 90); do
+  ts=$(date "+%F %T")
+  jobs=$(runai list jobs -p $P 2>/dev/null | grep -oE "inputcond-[a-z]+-[a-z]+-[a-z0-9-]*${STAMP}" | sort -u)
+  for J in $jobs; do
+    st=$(runai list jobs -p $P 2>/dev/null | awk -v j="$J" '$1==j{print $2}')
+    step=$(runai logs "$J" -p $P 2>/dev/null | grep -oE "step=[0-9]+ loss=[0-9.]+" | tail -1)
+    val=$(runai logs "$J" -p $P 2>/dev/null | grep -oE "\[val\] step=[0-9]+ shape=[0-9.]+" | tail -1)
+    echo "$ts $J status=$st $step $val" >> "$LOG"
+  done
+  echo "----" >> "$LOG"
+  sleep 300
+done
diff --git a/anchorgen_baselines_20260923/input_cond/watchdog_status.log b/anchorgen_baselines_20260923/input_cond/watchdog_status.log
new file mode 100644
index 0000000000000000000000000000000000000000..85d768fea5e121268fc6fc125ec0058bd4afcf4d
--- /dev/null
+++ b/anchorgen_baselines_20260923/input_cond/watchdog_status.log
@@ -0,0 +1,646 @@
+2026-08-16 20:53:15 inputcond-side-a100-40g-0816 status=Running step=000100 loss=1.279186 [val] step=000001 shape=2.006118
+2026-08-16 20:53:15 inputcond-front-a100-40g-0816 status=Running step=000100 loss=1.268984 [val] step=000001 shape=2.006057
+2026-08-16 20:53:15 inputcond-rear-a100-40g-0816 status=Running step=000100 loss=1.269827 [val] step=000001 shape=2.006045
+----
+2026-08-16 20:58:21 inputcond-side-a100-40g-0816 status=Running step=001200 loss=0.879705 [val] step=001000 shape=0.898779
+2026-08-16 20:58:21 inputcond-front-a100-40g-0816 status=Running step=001200 loss=0.877812 [val] step=001000 shape=0.900342
+2026-08-16 20:58:21 inputcond-rear-a100-40g-0816 status=Running step=001200 loss=0.876379 [val] step=001000 shape=0.897811
+----
+2026-08-16 21:03:26 inputcond-side-a100-40g-0816 status=Running step=002200 loss=0.817196 [val] step=002000 shape=0.872692
+2026-08-16 21:03:26 inputcond-front-a100-40g-0816 status=Running step=002300 loss=0.879019 [val] step=002000 shape=0.872869
+2026-08-16 21:03:26 inputcond-rear-a100-40g-0816 status=Running step=002300 loss=0.878616 [val] step=002000 shape=0.872898
+----
+2026-08-16 21:08:31 inputcond-side-a100-40g-0816 status=Running step=003300 loss=0.806843 [val] step=003000 shape=0.855979
+2026-08-16 21:08:31 inputcond-front-a100-40g-0816 status=Running step=003300 loss=0.810758 [val] step=003000 shape=0.860761
+2026-08-16 21:08:31 inputcond-rear-a100-40g-0816 status=Running step=003300 loss=0.810508 [val] step=003000 shape=0.859960
+----
+2026-08-16 21:16:42 inputcond-cfg-front-a100-40g-0816c status=Running step=000100 loss=1.235969 [val] step=000001 shape=2.006061
+2026-08-16 21:16:42 inputcond-cfg-rear-a100-40g-0816c status=Running step=000100 loss=1.228219 [val] step=000001 shape=2.006047
+2026-08-16 21:16:42 inputcond-cfg-side-a100-40g-0816c status=Running step=000200 loss=0.968851 [val] step=000001 shape=2.006123
+2026-08-16 21:16:42 inputcond-plain-front-a100-40g-0816c status=Running step=000100 loss=1.268984 [val] step=000001 shape=2.006057
+2026-08-16 21:16:42 inputcond-plain-rear-a100-40g-0816c status=Running step=000200 loss=1.032565 [val] step=000001 shape=2.006045
+2026-08-16 21:16:42 inputcond-plain-side-a100-40g-0816c status=Running step=000200 loss=1.040881 [val] step=000001 shape=2.006118
+----
+2026-08-16 21:21:53 inputcond-cfg-front-a100-40g-0816c status=Running step=001200 loss=0.863432 [val] step=001000 shape=0.893551
+2026-08-16 21:21:53 inputcond-cfg-rear-a100-40g-0816c status=Running step=001200 loss=0.863267 [val] step=001000 shape=0.896825
+2026-08-16 21:21:53 inputcond-cfg-side-a100-40g-0816c status=Running step=001300 loss=0.827217 [val] step=001000 shape=0.897834
+2026-08-16 21:21:53 inputcond-plain-front-a100-40g-0816c status=Running step=001200 loss=0.877812 [val] step=001000 shape=0.900342
+2026-08-16 21:21:53 inputcond-plain-rear-a100-40g-0816c status=Running step=001200 loss=0.876379 [val] step=001000 shape=0.897811
+2026-08-16 21:21:53 inputcond-plain-side-a100-40g-0816c status=Running step=001300 loss=0.816932 [val] step=001000 shape=0.898779
+----
+2026-08-16 21:27:03 inputcond-cfg-front-a100-40g-0816c status=Running step=002300 loss=0.817925 [val] step=002000 shape=0.868111
+2026-08-16 21:27:03 inputcond-cfg-rear-a100-40g-0816c status=Running step=002300 loss=0.820257 [val] step=002000 shape=0.870664
+2026-08-16 21:27:03 inputcond-cfg-side-a100-40g-0816c status=Running step=002400 loss=0.800772 [val] step=002000 shape=0.874796
+2026-08-16 21:27:03 inputcond-plain-front-a100-40g-0816c status=Running step=002300 loss=0.879019 [val] step=002000 shape=0.872869
+2026-08-16 21:27:03 inputcond-plain-rear-a100-40g-0816c status=Running step=002300 loss=0.878616 [val] step=002000 shape=0.872898
+2026-08-16 21:27:03 inputcond-plain-side-a100-40g-0816c status=Running step=002300 loss=0.878228 [val] step=002000 shape=0.872692
+----
+2026-08-16 21:32:13 inputcond-cfg-front-a100-40g-0816c status=Running step=003400 loss=0.807128 [val] step=003000 shape=0.856758
+2026-08-16 21:32:13 inputcond-cfg-rear-a100-40g-0816c status=Running step=003400 loss=0.808585 [val] step=003000 shape=0.858550
+2026-08-16 21:32:13 inputcond-cfg-side-a100-40g-0816c status=Running step=003500 loss=0.820824 [val] step=003500 shape=0.854158
+2026-08-16 21:32:13 inputcond-plain-front-a100-40g-0816c status=Running step=003400 loss=0.806018 [val] step=003000 shape=0.860557
+2026-08-16 21:32:13 inputcond-plain-rear-a100-40g-0816c status=Running step=003400 loss=0.804921 [val] step=003000 shape=0.859960
+2026-08-16 21:32:13 inputcond-plain-side-a100-40g-0816c status=Running step=003400 loss=0.804631 [val] step=003000 shape=0.855979
+----
+2026-08-16 21:37:24 inputcond-cfg-front-a100-40g-0816c status=Running step=004500 loss=0.775474 [val] step=004500 shape=0.842167
+2026-08-16 21:37:24 inputcond-cfg-rear-a100-40g-0816c status=Running step=004500 loss=0.777383 [val] step=004500 shape=0.845430
+2026-08-16 21:37:24 inputcond-cfg-side-a100-40g-0816c status=Running step=004600 loss=0.811783 [val] step=004500 shape=0.845604
+2026-08-16 21:37:24 inputcond-plain-front-a100-40g-0816c status=Running step=004500 loss=0.766437 [val] step=004500 shape=0.846097
+2026-08-16 21:37:24 inputcond-plain-rear-a100-40g-0816c status=Running step=004500 loss=0.768958 [val] step=004500 shape=0.845614
+2026-08-16 21:37:24 inputcond-plain-side-a100-40g-0816c status=Running step=004500 loss=0.763447 [val] step=004500 shape=0.842087
+----
+2026-08-16 21:42:34 inputcond-cfg-front-a100-40g-0816c status=Running step=005600 loss=0.809541 [val] step=005500 shape=0.837293
+2026-08-16 21:42:34 inputcond-cfg-rear-a100-40g-0816c status=Running step=005600 loss=0.812302 [val] step=005500 shape=0.839053
+2026-08-16 21:42:34 inputcond-cfg-side-a100-40g-0816c status=Running step=005700 loss=0.767015 [val] step=005500 shape=0.839235
+2026-08-16 21:42:34 inputcond-plain-front-a100-40g-0816c status=Running step=005500 loss=0.791137 [val] step=005500 shape=0.838939
+2026-08-16 21:42:34 inputcond-plain-rear-a100-40g-0816c status=Running step=005600 loss=0.740854 [val] step=005500 shape=0.840423
+2026-08-16 21:42:34 inputcond-plain-side-a100-40g-0816c status=Running step=005600 loss=0.739220 [val] step=005500 shape=0.834440
+----
+2026-08-16 21:47:44 inputcond-cfg-front-a100-40g-0816c status=Running step=006600 loss=0.759678 [val] step=006500 shape=0.832933
+2026-08-16 21:47:44 inputcond-cfg-rear-a100-40g-0816c status=Running step=006700 loss=0.793759 [val] step=006500 shape=0.834366
+2026-08-16 21:47:44 inputcond-cfg-side-a100-40g-0816c status=Running step=006800 loss=0.773349 [val] step=006500 shape=0.834818
+2026-08-16 21:47:44 inputcond-plain-front-a100-40g-0816c status=Running step=006600 loss=0.762873 [val] step=006500 shape=0.832295
+2026-08-16 21:47:44 inputcond-plain-rear-a100-40g-0816c status=Running step=006700 loss=0.804139 [val] step=006500 shape=0.833964
+2026-08-16 21:47:44 inputcond-plain-side-a100-40g-0816c status=Running step=006700 loss=0.800137 [val] step=006500 shape=0.829394
+----
+2026-08-16 21:52:54 inputcond-cfg-front-a100-40g-0816c status=Running step=007700 loss=0.719265 [val] step=007500 shape=0.829073
+2026-08-16 21:52:54 inputcond-cfg-rear-a100-40g-0816c status=Running step=007800 loss=0.776718 [val] step=007500 shape=0.831119
+2026-08-16 21:52:54 inputcond-cfg-side-a100-40g-0816c status=Running step=007900 loss=0.812744 [val] step=007500 shape=0.831571
+2026-08-16 21:52:54 inputcond-plain-front-a100-40g-0816c status=Running step=007700 loss=0.833422 [val] step=007500 shape=0.829670
+2026-08-16 21:52:54 inputcond-plain-rear-a100-40g-0816c status=Running step=007800 loss=0.786242 [val] step=007500 shape=0.832417
+2026-08-16 21:52:54 inputcond-plain-side-a100-40g-0816c status=Running step=007800 loss=0.784528 [val] step=007500 shape=0.827095
+----
+2026-08-16 21:58:04 inputcond-cfg-front-a100-40g-0816c status=Running step=008800 loss=0.777775 [val] step=008500 shape=0.826718
+2026-08-16 21:58:04 inputcond-cfg-rear-a100-40g-0816c status=Running step=008900 loss=0.736398 [val] step=008500 shape=0.828830
+2026-08-16 21:58:04 inputcond-cfg-side-a100-40g-0816c status=Running step=009100 loss=0.737742 [val] step=009000 shape=0.829065
+2026-08-16 21:58:04 inputcond-plain-front-a100-40g-0816c status=Running step=008800 loss=0.812365 [val] step=008500 shape=0.826995
+2026-08-16 21:58:04 inputcond-plain-rear-a100-40g-0816c status=Running step=008900 loss=0.756337 [val] step=008500 shape=0.829680
+2026-08-16 21:58:04 inputcond-plain-side-a100-40g-0816c status=Running step=008900 loss=0.753192 [val] step=008500 shape=0.824447
+----
+2026-08-16 22:03:15 inputcond-cfg-front-a100-40g-0816c status=Running step=009900 loss=0.724129 [val] step=009500 shape=0.825431
+2026-08-16 22:03:15 inputcond-cfg-rear-a100-40g-0816c status=Running step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:03:15 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:03:15 inputcond-plain-front-a100-40g-0816c status=Running step=009900 loss=0.753926 [val] step=009500 shape=0.825653
+2026-08-16 22:03:15 inputcond-plain-rear-a100-40g-0816c status=Running step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:03:15 inputcond-plain-side-a100-40g-0816c status=Running step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:08:25 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:08:25 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:08:25 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:08:25 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:08:25 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:08:25 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:13:36 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:13:36 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:13:36 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:13:36 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:13:36 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:13:36 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:18:47 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:18:47 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:18:47 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:18:47 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:18:47 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:18:47 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:23:57 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:23:57 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:23:57 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:23:57 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:23:57 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:23:57 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:29:07 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:29:07 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:29:07 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:29:07 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:29:07 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:29:07 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:34:18 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:34:18 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:34:18 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:34:18 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:34:18 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:34:18 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:39:28 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:39:28 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:39:28 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:39:28 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:39:28 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:39:28 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:44:38 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:44:38 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:44:38 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:44:38 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:44:38 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:44:38 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:49:48 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:49:48 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:49:48 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:49:48 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:49:48 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:49:48 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 22:54:58 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 22:54:58 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 22:54:58 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 22:54:58 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 22:54:58 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 22:54:58 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:00:09 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:00:09 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:00:09 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:00:09 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:00:09 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:00:09 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:05:20 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:05:20 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:05:20 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:05:20 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:05:20 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:05:20 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:10:30 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:10:30 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:10:30 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:10:30 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:10:30 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:10:30 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:15:40 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:15:40 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:15:40 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:15:40 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:15:40 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:15:40 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:20:50 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:20:50 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:20:50 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:20:50 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:20:50 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:20:50 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:26:02 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:26:02 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:26:02 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:26:02 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:26:02 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:26:02 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:31:13 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:31:13 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:31:13 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:31:13 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:31:13 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:31:13 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:36:23 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:36:23 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:36:23 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:36:23 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:36:23 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:36:23 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:41:33 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:41:33 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:41:33 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:41:33 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:41:33 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:41:33 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:46:44 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:46:44 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:46:44 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:46:44 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:46:44 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:46:44 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:51:54 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:51:54 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:51:54 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:51:54 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:51:54 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:51:54 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-16 23:57:05 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-16 23:57:05 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-16 23:57:05 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-16 23:57:05 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-16 23:57:05 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-16 23:57:05 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:02:16 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:02:16 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:02:16 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:02:16 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:02:16 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:02:16 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:07:26 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:07:26 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:07:26 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:07:26 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:07:26 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:07:26 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:12:37 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:12:37 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:12:37 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:12:37 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:12:37 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:12:37 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:17:48 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:17:48 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:17:48 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:17:48 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:17:48 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:17:48 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:23:00 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:23:00 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:23:00 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:23:00 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:23:00 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:23:00 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:28:10 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:28:10 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:28:10 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:28:10 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:28:10 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:28:10 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:33:21 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:33:21 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:33:21 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:33:21 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:33:21 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:33:21 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:38:31 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:38:31 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:38:31 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:38:31 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:38:31 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:38:31 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:43:41 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:43:41 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:43:41 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:43:41 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:43:41 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:43:41 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:48:51 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:48:51 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:48:51 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:48:51 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:48:51 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:48:51 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:54:01 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:54:01 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:54:01 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:54:01 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:54:01 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:54:01 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 00:59:12 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 00:59:12 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 00:59:12 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 00:59:12 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 00:59:12 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 00:59:12 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:04:22 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:04:22 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:04:22 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:04:22 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:04:22 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:04:22 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:09:33 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:09:33 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:09:33 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:09:33 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:09:33 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:09:33 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:14:44 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:14:44 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:14:44 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:14:44 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:14:44 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:14:44 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:19:55 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:19:55 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:19:55 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:19:55 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:19:55 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:19:55 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:25:05 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:25:05 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:25:05 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:25:05 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:25:05 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:25:05 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:30:16 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:30:16 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:30:16 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:30:16 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:30:16 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:30:16 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:35:28 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:35:28 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:35:28 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:35:28 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:35:28 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:35:28 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:40:38 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:40:38 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:40:38 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:40:38 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:40:38 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:40:38 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:45:49 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:45:49 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:45:49 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:45:49 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:45:49 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:45:49 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:50:59 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:50:59 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:50:59 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:50:59 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:50:59 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:50:59 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 01:56:10 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 01:56:10 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 01:56:10 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 01:56:10 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 01:56:10 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 01:56:10 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:01:21 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:01:21 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:01:21 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:01:21 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:01:21 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:01:21 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:06:32 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:06:32 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:06:32 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:06:32 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:06:32 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:06:32 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:11:42 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:11:42 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:11:42 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:11:42 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:11:42 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:11:42 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:16:53 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:16:53 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:16:53 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:16:53 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:16:53 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:16:53 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:22:03 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:22:03 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:22:03 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:22:03 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:22:03 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:22:03 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:27:13 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:27:13 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:27:13 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:27:13 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:27:13 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:27:13 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:32:24 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:32:24 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:32:24 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:32:24 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:32:24 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:32:24 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:37:34 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:37:34 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:37:34 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:37:34 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:37:34 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:37:34 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:42:46 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:42:46 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:42:46 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:42:46 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:42:46 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:42:46 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:47:57 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:47:57 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:47:57 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:47:57 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:47:57 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:47:57 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:53:07 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:53:07 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:53:07 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:53:07 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:53:07 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:53:07 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 02:58:18 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 02:58:18 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 02:58:18 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 02:58:18 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 02:58:18 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 02:58:18 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:03:29 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:03:29 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:03:29 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:03:29 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:03:29 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:03:29 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:08:39 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:08:39 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:08:39 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:08:39 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:08:39 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:08:39 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:13:49 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:13:49 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:13:49 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:13:49 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:13:49 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:13:49 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:18:59 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:18:59 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:18:59 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:18:59 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:18:59 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:18:59 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:24:10 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:24:10 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:24:10 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:24:10 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:24:10 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:24:10 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:29:21 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:29:21 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:29:21 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:29:21 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:29:21 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:29:21 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:34:31 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:34:31 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:34:31 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:34:31 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:34:31 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:34:31 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:39:42 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:39:42 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:39:42 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:39:42 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:39:42 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:39:42 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:44:52 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:44:52 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:44:52 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:44:52 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:44:52 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:44:52 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:50:03 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:50:03 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:50:03 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:50:03 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:50:03 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:50:03 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 03:55:14 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 03:55:14 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 03:55:14 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 03:55:14 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 03:55:14 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 03:55:14 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:00:24 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:00:24 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:00:24 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:00:24 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:00:24 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:00:24 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:05:36 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:05:36 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:05:36 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:05:36 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:05:36 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:05:36 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:10:45 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:10:45 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:10:45 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:10:45 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:10:45 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:10:45 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:15:56 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:15:56 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:15:56 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:15:56 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:15:56 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:15:56 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:21:06 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:21:06 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:21:06 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:21:06 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:21:06 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:21:06 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:26:16 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:26:16 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:26:16 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:26:16 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:26:16 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:26:16 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:31:27 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:31:27 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:31:27 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:31:27 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:31:27 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:31:27 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:36:38 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:36:38 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:36:38 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:36:38 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:36:38 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:36:38 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:41:47 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:41:47 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:41:47 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:41:47 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:41:47 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:41:47 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:46:57 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:46:57 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:46:57 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:46:57 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:46:57 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:46:57 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:52:08 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:52:08 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:52:08 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:52:08 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:52:08 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:52:08 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----
+2026-08-17 04:57:18 inputcond-cfg-front-a100-40g-0816c status=Succeeded step=010000 loss=0.781975 [val] step=010000 shape=0.825316
+2026-08-17 04:57:18 inputcond-cfg-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.783486 [val] step=010000 shape=0.827270
+2026-08-17 04:57:18 inputcond-cfg-side-a100-40g-0816c status=Succeeded step=010000 loss=0.784705 [val] step=010000 shape=0.827590
+2026-08-17 04:57:18 inputcond-plain-front-a100-40g-0816c status=Succeeded step=010000 loss=0.809452 [val] step=010000 shape=0.825599
+2026-08-17 04:57:18 inputcond-plain-rear-a100-40g-0816c status=Succeeded step=010000 loss=0.811191 [val] step=010000 shape=0.828091
+2026-08-17 04:57:18 inputcond-plain-side-a100-40g-0816c status=Succeeded step=010000 loss=0.809046 [val] step=010000 shape=0.822673
+----