13_pixel_arm: the best pixel arm (window model + 8-step update head, 2026-10-31), README, commands, test33 results

13_pixel_arm/README.md

@@ -0,0 +1,68 @@
+# Pixel arm: the best pixel-root memory (window model + 8-step update head; 2026-10-31)
+
+The current best pixel (root) memory on the orbit benchmark. It replaces the previous pixel arm (Bk, formerly `12_pixel_arm_bk/`,
+removed 2026-11-01): same window model, a head retrained for 8 update steps, and a lighter evaluation configuration. It matches Bk's
+PSNR at the end of the stream (−0.00 ± 0.03 dB on test33) with better LPIPS (−.010), 3.75 x less memory and about 3.5 x the speed.
+
+- `model_001000.pt`: the window model (3D-Belief trunk, proposal head, Gaussian decoder; EMA) of `outputs/joint_j1keptlr`, step 1000. The
+  same file as Bk's (sha256 unchanged); `args['decoder_from']` is repo-relative (the only change from the cluster original).
+- `upd_000800.pt`: the learned root update head (EMA, `['ema']`), retrained for 8 update steps (`outputs/opt_it8`, `command_head.json`):
+  head only, from Bk's head, 800 steps, head lr 2.5e-5, seed 3, rollouts over 400 DL3DV training orbits (`data/orbit_cache/dl3dv_train`,
+  disjoint from val33 / test33) with the 5 deg / .08 pose store, replay 8 views at a time and kept targets. Its `['args']` still carries
+  absolute cluster paths inherited from older heads; loading reads only `['ema']` and `inner_lpips`.
+- `command_window.json` / `command_j1.json`: the window model's training (J1, then Bk's +1000 steps at half lr); `command_eval_test33.json`:
+  the evaluation that produced the numbers below (shard 0 of 4).
+
+**Needs, besides these files** (run from the repo root, branch amodal-completion, commit b7f7e0a or later):
+- the decoder construction files `free_completion_20260927/checkpoints/ray/step002000` from `03_checkpoints_streaming.zip`, at
+  `outputs/free_completion_20260927/checkpoints/ray/step002000` (the snapshot's weights are loaded over it);
+- the 3D-Belief prior `3d_belief_re10k_256.pt` at `$FLC_BELIEF_CKPT` (read when the trunk is built; not in this repo);
+- for `--sky-layer`: the sky segmenter `nvidia/segformer-b0-finetuned-ade-512-512` at `checkpoints/segformer_b0_ade`;
+- the orbit caches `data/orbit_cache/dl3dv_test33` (and `dl3dv_val33`) from `11_orbit_eval_data/`.
+
+## How it is run
+
+Each 6-frame window (advancing 2 frames) is written into the root memory; the update head then refines the roots for 8 steps, fitting the
+window's frames plus a self-replay of every stored past pose that overlaps the update. A pose is stored when it is > 10 deg or > 0.15 x scene
+depth from every stored pose; its replay target is **kept**: the memory's own render right after the last window that observed that pose or
+its cell (uint8, 256 px). Lossless speed-ups (half-precision state and roots, a lean bookkeeping TSDF, a cheaper exact gradient), every render
+on a 5 x 5 Gaussian lattice per root (25 Gaussians), pruning by opacity mass, a short lifetime for the update state of roots not updated
+recently, and a sky layer (a direction-only panorama of observed sky behind the Gaussians; far views skip roots that render mostly on sky).
+
+**No past real frame is stored**: each window's frames are used for its write and update and dropped. Kept per stream (test33, ~348 MB at the
+end): roots, update-head state, bookkeeping TSDF, replay poses (+ 32 x 32 caches rendered from the memory), kept replay targets (the memory's
+own renders), and the sky panorama + masks (one fused colour per observed sky direction, the only stored content derived from real pixels).
+
+    python scripts/unified/eval_long_stream.py --snapshot 13_pixel_arm/model_001000.pt --root-updater 13_pixel_arm/upd_000800.pt \
+      --replay-chunk 8 --replay-store cell --fast-grad --state-half --roots-half --tsdf-lean --tsdf-carve 0 \
+      --lod 5 --prune-mass .4 --state-ttl 4 --tsdf-size .1 --sky-layer --sky-thr 2 --sky-far 16 \
+      --admit-angle 10 --admit-dist .15 --ru-iters 8 \
+      --data data/orbit_cache/dl3dv_test33 --clips 33 --loops 2 --async-period 0 --out <dir>
+
+For far overview renders only, `--far-cap 4 --far-depth .15` (with `--sky-far`) also caps Gaussians at 4 source pixels and skips those in
+front of the TSDF surface at that camera; render-only, the memory and near-path renders are unchanged.
+
+## Results
+
+dl3dv_test33 (33 DL3DV test orbits), walked twice (6 orbits); after every window the memory renders the first orbit's held-out frames; all
+pixels (holes count); one seed; one process per GPU (three H100, one H200). `test33_orbit.png` (orbit curves, per-scene changes, tiles) and
+`test33_orbit.json` (the numbers below).
+
+| test33, 6 orbits, all pixels | PSNR orbit 1 / 3 / 6 | orbit 6 − orbit 1 | walk 2 (6 − 3) | LPIPS at 6 | MB at the end | s per frame |
+|---|---|---|---|---|---|---|
+| **pixel arm (this)** | **17.36 / 18.08 / 18.12** | **+0.75 ± 0.10 (29 / 33 rise)** | +0.04 (19 / 33) | .394 | 348 | 2.09 |
+| frame memory (Context-as-Memory style, our window model) | 16.27 / 16.52 / 16.52 | +0.26 ± 0.25 | 0.00 | .382 | 17 | 0.04 |
+| Gen3R, frame buffer (8 scenes) | 19.29 / 17.30 / 17.43 | −1.86 | +0.13 | .301 | | |
+| SPMem TSDF (same window predictions) | 14.20 / 14.73 / 14.51 | +0.31 ± 0.20 | −0.22 | .627 | 147 | 0.05 |
+| plain point accumulation | 12.35 / 11.85 / 11.65 | −0.69 ± 0.09 | −0.20 | .753 | 78 | 0.15 |
+
+LPIPS of this arm at orbits 1 / 3 / 6: .417 / .389 / .394. Bk on the same scenes: 17.42 / 18.15 / 18.12, +0.69, 1.1-1.3 GB, 5.1 s per frame.
+dl3dv_val8 (8 scenes, the selection set): 16.44 / 17.10 / 17.08, +0.64, LPIPS .435, 367 MB, 2.37 s per frame (Bk 8.23 s, matched GPUs).
+A window takes ~4.4 s on test33 (6-frame windows advancing 2, two walks), most of it the replay of ~44 stored poses per window.
+
+**Known limits:** the mean over the first orbit's views rises for about 1.5 orbits and then stays flat; views the memory already rendered well
+slowly lose detail (0.7-1.1 dB below their own peak) while weaker views catch up. Far views show haze from a few hundred runaway roots; the
+far-view filter hides it, it is not removed. Runs are not bit-reproducible (repeat runs of the same flags differ by up to ~0.1 dB per scene).
+
+sha256 model_001000.pt: 08e30bdaac8ddb560151384df06e16bc88606cbe86915bb92d587e16e522110e
+sha256 upd_000800.pt: 7a1d6c3a76c68102f278048a8869ab589e4bab48e80250e9805eaaa428d99d5d

13_pixel_arm/command_eval_test33.json

@@ -0,0 +1,60 @@
+{
+ "argv": [
+  "scripts/unified/eval_long_stream.py",
+  "--data",
+  "data/orbit_cache/dl3dv_test33",
+  "--clips",
+  "33",
+  "--loops",
+  "2",
+  "--async-period",
+  "0",
+  "--shard",
+  "0/4",
+  "--out",
+  "outputs/long_stream/test33/bk_opt/s0of4",
+  "--overview",
+  "--save-track",
+  "ends",
+  "--snapshot",
+  "outputs/joint_j1keptlr/model_001000.pt",
+  "--root-updater",
+  "outputs/joint_j1keptlr/upd_001000.pt",
+  "--admit-angle",
+  "5",
+  "--admit-dist",
+  ".08",
+  "--replay-chunk",
+  "8",
+  "--replay-store",
+  "cell",
+  "--fast-grad",
+  "--state-half",
+  "--roots-half",
+  "--tsdf-lean",
+  "--tsdf-carve",
+  "0",
+  "--prof",
+  "--lod",
+  "5",
+  "--prune-mass",
+  ".4",
+  "--state-ttl",
+  "4",
+  "--tsdf-size",
+  ".1",
+  "--sky-layer",
+  "--sky-thr",
+  "2",
+  "--sky-far",
+  "16",
+  "--admit-angle",
+  "10",
+  "--admit-dist",
+  ".15",
+  "--ru-iters",
+  "8",
+  "--root-updater",
+  "outputs/opt_it8/upd_000800.pt"
+ ]
+}

13_pixel_arm/command_head.json

@@ -0,0 +1,81 @@
+{
+ "argv": [
+  "flc/unified/train_joint_stream.py",
+  "--train",
+  "head",
+  "--out",
+  "outputs/opt_it8",
+  "--steps",
+  "800",
+  "--save-every",
+  "200",
+  "--iters",
+  "8",
+  "--admit-angle",
+  "5",
+  "--admit-dist",
+  ".08",
+  "--replay-chunk",
+  "8",
+  "--replay-store",
+  "cell",
+  "--fast-grad",
+  "--lr-head",
+  "2.5e-5",
+  "--lr-trunk",
+  "1e-5",
+  "--snapshot",
+  "outputs/joint_j1keptlr/model_001000.pt",
+  "--root-updater",
+  "outputs/joint_j1keptlr/upd_001000.pt",
+  "--seed",
+  "3"
+ ],
+ "args": {
+  "out": "outputs/opt_it8",
+  "train": "head",
+  "snapshot": "outputs/joint_j1keptlr/model_001000.pt",
+  "root_updater": "outputs/joint_j1keptlr/upd_001000.pt",
+  "data": "/weka/projects/tshu2/yyin34/free-latent-completion/data/orbit_cache/dl3dv_train",
+  "overfit": "",
+  "exclude": "",
+  "steps": "800",
+  "rollouts": "2",
+  "iters": "8",
+  "t_grad": "6",
+  "lr_head": "2.5e-05",
+  "lr_trunk": "1e-05",
+  "warmup": "100",
+  "ema": "0.999",
+  "n_win": "3",
+  "n_other": "5",
+  "far": "12",
+  "max_ang": "150.0",
+  "see_region": "True",
+  "w_other": "0.7",
+  "lpips_w": "1.0",
+  "w_depth": "0.1",
+  "dep_clamp": "0.3",
+  "w_harm": "0.0",
+  "w_self": "0.2",
+  "ot_app_w": "0.0",
+  "ot_geo_w": "0.0",
+  "ot_n": "2",
+  "ot_frac": "0.3",
+  "log_every": "10",
+  "save_every": "200",
+  "max_steps_run": "0",
+  "seed": "3",
+  "admit": "3db",
+  "admit_angle": "5.0",
+  "admit_dist": "0.08",
+  "replay_chunk": "8",
+  "fast_grad": "True",
+  "replay_store": "cell",
+  "n_held": "0",
+  "n_held_near": "0",
+  "tsdf_n": "0",
+  "w_tsdf": "0.0",
+  "w_tsdf_dep": "0.0"
+ }
+}

13_pixel_arm/command_j1.json

@@ -0,0 +1,48 @@
+{
+ "argv": [
+  "flc/unified/train_joint_stream.py",
+  "--train",
+  "joint",
+  "--out",
+  "outputs/joint_j1fix",
+  "--steps",
+  "2000",
+  "--save-every",
+  "500"
+ ],
+ "args": {
+  "out": "outputs/joint_j1fix",
+  "train": "joint",
+  "snapshot": "/weka/projects/tshu2/yyin34/free-latent-completion/outputs/card_r1lat_v1/ema_005999.pt",
+  "root_updater": "/weka/projects/tshu2/yyin34/free-latent-completion/outputs/root_upd_v1/ema_005000.pt",
+  "data": "/weka/projects/tshu2/yyin34/free-latent-completion/data/orbit_cache/dl3dv_train",
+  "overfit": "",
+  "exclude": "",
+  "steps": "2000",
+  "rollouts": "2",
+  "iters": "16",
+  "t_grad": "6",
+  "lr_head": "5e-05",
+  "lr_trunk": "2e-05",
+  "warmup": "100",
+  "ema": "0.999",
+  "n_win": "3",
+  "n_other": "5",
+  "far": "12",
+  "max_ang": "150.0",
+  "see_region": "True",
+  "w_other": "0.7",
+  "lpips_w": "1.0",
+  "w_depth": "0.1",
+  "dep_clamp": "0.3",
+  "w_self": "0.2",
+  "ot_app_w": "0.0",
+  "ot_geo_w": "0.0",
+  "ot_n": "2",
+  "ot_frac": "0.3",
+  "log_every": "10",
+  "save_every": "500",
+  "max_steps_run": "0",
+  "seed": "0"
+ }
+}

13_pixel_arm/command_window.json

@@ -0,0 +1,72 @@
+{
+ "argv": [
+  "flc/unified/train_joint_stream.py",
+  "--train",
+  "joint",
+  "--out",
+  "outputs/joint_j1keptlr",
+  "--steps",
+  "1000",
+  "--save-every",
+  "250",
+  "--admit-angle",
+  "5",
+  "--admit-dist",
+  ".08",
+  "--replay-chunk",
+  "8",
+  "--lr-head",
+  "2.5e-5",
+  "--lr-trunk",
+  "1e-5",
+  "--replay-store",
+  "cell",
+  "--snapshot",
+  "outputs/joint_j1fix/model_001000.pt",
+  "--root-updater",
+  "outputs/joint_j1fix/upd_001000.pt",
+  "--seed",
+  "2"
+ ],
+ "args": {
+  "out": "outputs/joint_j1keptlr",
+  "train": "joint",
+  "snapshot": "outputs/joint_j1fix/model_001000.pt",
+  "root_updater": "outputs/joint_j1fix/upd_001000.pt",
+  "data": "/weka/projects/tshu2/yyin34/free-latent-completion/data/orbit_cache/dl3dv_train",
+  "overfit": "",
+  "exclude": "",
+  "steps": "1000",
+  "rollouts": "2",
+  "iters": "16",
+  "t_grad": "6",
+  "lr_head": "2.5e-05",
+  "lr_trunk": "1e-05",
+  "warmup": "100",
+  "ema": "0.999",
+  "n_win": "3",
+  "n_other": "5",
+  "far": "12",
+  "max_ang": "150.0",
+  "see_region": "True",
+  "w_other": "0.7",
+  "lpips_w": "1.0",
+  "w_depth": "0.1",
+  "dep_clamp": "0.3",
+  "w_harm": "0.0",
+  "w_self": "0.2",
+  "ot_app_w": "0.0",
+  "ot_geo_w": "0.0",
+  "ot_n": "2",
+  "ot_frac": "0.3",
+  "log_every": "10",
+  "save_every": "250",
+  "max_steps_run": "0",
+  "seed": "2",
+  "admit": "3db",
+  "admit_angle": "5.0",
+  "admit_dist": "0.08",
+  "replay_chunk": "8",
+  "replay_store": "cell"
+ }
+}

13_pixel_arm/model_001000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08e30bdaac8ddb560151384df06e16bc88606cbe86915bb92d587e16e522110e
+size 1875927516

13_pixel_arm/test33_orbit.json

@@ -0,0 +1,91 @@
+{
+ "bk": {
+  "n": 33,
+  "w2": 0.03525312019116951,
+  "w2sem": 0.029616343146641126,
+  "w2rises": 19,
+  "psnr": [
+   17.36470041005299,
+   18.080895019300055,
+   18.116148139491226
+  ],
+  "lpips6": 0.39366416317043884,
+  "change": 0.751447729438242,
+  "sem": 0.09540301630408392,
+  "rises": 29,
+  "mb": 347.5564751769557,
+  "spf": 2.0931816904571865,
+  "lat_end": 6.051709124075856
+ },
+ "frames": {
+  "n": 33,
+  "w2": 0.0,
+  "w2sem": 0.0,
+  "w2rises": 0,
+  "psnr": [
+   16.267142729447794,
+   16.52245733532537,
+   16.52245733532537
+  ],
+  "lpips6": 0.3821600458838723,
+  "change": 0.25531460587757454,
+  "sem": 0.25462612944353685,
+  "rises": 24,
+  "mb": 17.1735031821511,
+  "spf": 0.03791738243446462,
+  "lat_end": 4.209590825308679e-05
+ },
+ "gen3r": {
+  "n": 8,
+  "w2": 0.13177835941314697,
+  "w2rises": 5,
+  "psnr": [
+   19.291090428829193,
+   17.295971393585205,
+   17.427749752998352
+  ],
+  "lpips6": 0.3005409687757492,
+  "change": -1.863340675830841,
+  "sem": 0.47082713944956667,
+  "rises": 1,
+  "mb": null,
+  "spf": null,
+  "lat_end": null
+ },
+ "tsdf": {
+  "n": 33,
+  "w2": -0.21811450813645986,
+  "w2sem": 0.11167982985899921,
+  "w2rises": 13,
+  "psnr": [
+   14.20225967425853,
+   14.727262031998611,
+   14.509147523862152
+  ],
+  "lpips6": 0.6270990173021952,
+  "change": 0.3068878496036213,
+  "sem": 0.19860134132758261,
+  "rises": 19,
+  "mb": 147.3649885004217,
+  "spf": 0.054843342170983515,
+  "lat_end": 0.10048230033223235
+ },
+ "append": {
+  "n": 33,
+  "w2": -0.19977969792793646,
+  "w2sem": 0.02144698507209263,
+  "w2rises": 2,
+  "psnr": [
+   12.34671978763236,
+   11.85263101645072,
+   11.652851318522783
+  ],
+  "lpips6": 0.7525797973979603,
+  "change": -0.693868469109576,
+  "sem": 0.08872276525531778,
+  "rises": 3,
+  "mb": 78.36447837136008,
+  "spf": 0.14940210696587528,
+  "lat_end": 0.2953080643668672
+ }
+}

13_pixel_arm/upd_000800.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a1d6c3a76c68102f278048a8869ab589e4bab48e80250e9805eaaa428d99d5d
+size 2097973

SHA256SUMS

@@ -4,5 +4,5 @@ e67573c66ee8baeb892e6c32dc6a73380a68b4718d96d26e0bf9f0e738336aa9  03_checkpoints
 e0056e2d7e12fe99eeb1161c0a633d7d7b18837ee12aa0dc3ca3bbf2f2a5db50  04_eval_data.zip
 14d454ea4e2caeb2f26478bf5661270b707c03b8f50640cdb2cb86029ba0797a  05_boards.zip
 fe7959a3bd5770cbd8075d34a97185c3fdf5aa1c7395f95b37403d1dcd622f42  06_gen3r_outputs.zip
-08e30bdaac8ddb560151384df06e16bc88606cbe86915bb92d587e16e522110e  12_pixel_arm_bk/model_001000.pt
-b72d9620b484eb1c05e4e04d259af923f807e7cb04243323fd74cd2809c6f412  12_pixel_arm_bk/upd_001000.pt
+08e30bdaac8ddb560151384df06e16bc88606cbe86915bb92d587e16e522110e  13_pixel_arm/model_001000.pt
+7a1d6c3a76c68102f278048a8869ab589e4bab48e80250e9805eaaa428d99d5d  13_pixel_arm/upd_000800.pt