| { |
| "pair": "diff_seed", |
| "models": [ |
| { |
| "checkpoint": "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/outputs/iterative_image_diff_seed_1to16/20260928-022436/7pct/seed_42/checkpoints/14000.ckpt", |
| "kind": "image", |
| "backend": "unet", |
| "training_seed": 42, |
| "horizon": 64, |
| "weights": "model", |
| "checked_stat_files": [ |
| "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/outputs/iterative_image_diff_seed_1to16/20260928-022436/7pct/seed_42/StackCube-v1_image_data_stat.json" |
| ], |
| "inference_passes": 1 |
| }, |
| { |
| "checkpoint": "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/outputs/iterative_image_diff_seed_1to16/20260928-022436/7pct/seed_43/checkpoints/14000.ckpt", |
| "kind": "image", |
| "backend": "unet", |
| "training_seed": 43, |
| "horizon": 64, |
| "weights": "model", |
| "checked_stat_files": [ |
| "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/outputs/iterative_image_diff_seed_1to16/20260928-022436/7pct/seed_43/StackCube-v1_image_data_stat.json" |
| ], |
| "inference_passes": 1 |
| } |
| ], |
| "arguments": { |
| "pair": "diff_seed", |
| "checkpoint_a": "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/outputs/iterative_image_diff_seed_1to16/20260928-022436/7pct/seed_42/checkpoints/14000.ckpt", |
| "checkpoint_b": "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/outputs/iterative_image_diff_seed_1to16/20260928-022436/7pct/seed_43/checkpoints/14000.ckpt", |
| "image_h5": "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/render/image_idm_four_checkpoints_200/full/rollout_recon_rgb_224.h5", |
| "pointcloud_h5": "/vepfs-mlp2/c20250508/242303041/caiyi/Structured-4D-Model/render/image_idm_four_checkpoints_200/full/rollout_pointclouds.h5", |
| "output_dir": "outputs/rollout200_correlation_new_pair/diff_seed", |
| "obs_gap": 64, |
| "compare_steps": 32, |
| "step_stride": 32, |
| "batch_size": 8, |
| "seeds": [ |
| 42 |
| ], |
| "device": "cuda:0" |
| }, |
| "num_chunks": 739, |
| "num_trajectories": 200, |
| "metrics": { |
| "arm": { |
| "means": { |
| "disagreement": 0.0005110082774893017, |
| "error_a": 0.0015615346796819813, |
| "error_b": 0.0013162845208851365, |
| "mean_error": 0.0014389096002835591 |
| }, |
| "correlations": { |
| "error_a": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.23542427610134747, |
| "spearman": 0.3473395637295436 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.07576466847669316, |
| "spearman": 0.242137553438836 |
| } |
| }, |
| "error_b": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.2147787625057661, |
| "spearman": 0.3209700971310266 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.13779001753878045, |
| "spearman": 0.23213830345758646 |
| } |
| }, |
| "mean_error": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.23391033523067586, |
| "spearman": 0.4026476061856768 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.11032725153203346, |
| "spearman": 0.3485847146178655 |
| } |
| } |
| } |
| }, |
| "gripper": { |
| "means": { |
| "disagreement": 0.21740834481415425, |
| "error_a": 0.6728408226527905, |
| "error_b": 0.693212512155207, |
| "mean_error": 0.6830266674039988 |
| }, |
| "correlations": { |
| "error_a": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.12845551455251783, |
| "spearman": 0.2957166655326608 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.2439457132653402, |
| "spearman": 0.28794269856746424 |
| } |
| }, |
| "error_b": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.2334855678562151, |
| "spearman": 0.3996873912361575 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.3533019175580762, |
| "spearman": 0.36611415285382143 |
| } |
| }, |
| "mean_error": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.1872857687022101, |
| "spearman": 0.4451640170837177 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.30746580742399815, |
| "spearman": 0.3900187504687618 |
| } |
| } |
| } |
| }, |
| "overall": { |
| "means": { |
| "disagreement": 0.02762317534457242, |
| "error_a": 0.08547144567632055, |
| "error_b": 0.08780331297517537, |
| "mean_error": 0.08663737932574794 |
| }, |
| "correlations": { |
| "error_a": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.12520100494304187, |
| "spearman": 0.08208141142231376 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.23710443712135543, |
| "spearman": 0.27073426835670894 |
| } |
| }, |
| "error_b": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.2304800738523941, |
| "spearman": 0.13174933076910042 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.3467749122732793, |
| "spearman": 0.3421120528013201 |
| } |
| }, |
| "mean_error": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.18409170607904385, |
| "spearman": 0.20032796253793647 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.3006881885689036, |
| "spearman": 0.3650686267156679 |
| } |
| } |
| } |
| } |
| }, |
| "per_seed": [ |
| { |
| "seed": 42, |
| "metrics": { |
| "arm": { |
| "means": { |
| "disagreement": 0.0005110082774893017, |
| "error_a": 0.0015615346796819813, |
| "error_b": 0.0013162845208851365, |
| "mean_error": 0.0014389096002835591 |
| }, |
| "correlations": { |
| "error_a": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.23542427610134747, |
| "spearman": 0.3473395637295436 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.07576466847669316, |
| "spearman": 0.242137553438836 |
| } |
| }, |
| "error_b": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.2147787625057661, |
| "spearman": 0.3209700971310266 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.13779001753878045, |
| "spearman": 0.23213830345758646 |
| } |
| }, |
| "mean_error": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.23391033523067586, |
| "spearman": 0.4026476061856768 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.11032725153203346, |
| "spearman": 0.3485847146178655 |
| } |
| } |
| } |
| }, |
| "gripper": { |
| "means": { |
| "disagreement": 0.21740834481415425, |
| "error_a": 0.6728408226527905, |
| "error_b": 0.693212512155207, |
| "mean_error": 0.6830266674039988 |
| }, |
| "correlations": { |
| "error_a": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.12845551455251783, |
| "spearman": 0.2957166655326608 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.2439457132653402, |
| "spearman": 0.28794269856746424 |
| } |
| }, |
| "error_b": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.2334855678562151, |
| "spearman": 0.3996873912361575 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.3533019175580762, |
| "spearman": 0.36611415285382143 |
| } |
| }, |
| "mean_error": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.1872857687022101, |
| "spearman": 0.4451640170837177 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.30746580742399815, |
| "spearman": 0.3900187504687618 |
| } |
| } |
| } |
| }, |
| "overall": { |
| "means": { |
| "disagreement": 0.02762317534457242, |
| "error_a": 0.08547144567632055, |
| "error_b": 0.08780331297517537, |
| "mean_error": 0.08663737932574794 |
| }, |
| "correlations": { |
| "error_a": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.12520100494304187, |
| "spearman": 0.08208141142231376 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.23710443712135543, |
| "spearman": 0.27073426835670894 |
| } |
| }, |
| "error_b": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.2304800738523941, |
| "spearman": 0.13174933076910042 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.3467749122732793, |
| "spearman": 0.3421120528013201 |
| } |
| }, |
| "mean_error": { |
| "chunk": { |
| "n": 739, |
| "pearson": 0.18409170607904385, |
| "spearman": 0.20032796253793647 |
| }, |
| "trajectory": { |
| "n": 200, |
| "pearson": 0.3006881885689036, |
| "spearman": 0.3650686267156679 |
| } |
| } |
| } |
| } |
| } |
| } |
| ], |
| "protocol": { |
| "action_space": "normalized abs_joint; shared StackCube DATA_STAT", |
| "target": "executed rollout actions, not expert optimal actions", |
| "observations": "stored real start/end observations; not online WM target images", |
| "rng": "idm-rollout-chunk-v1; same root/trajectory/step/modality streams across checkpoints; shared image noise for diff_seed", |
| "sampling": "mean scalar MSE/disagreement across inference seeds; ACT zero latent evaluated once", |
| "trajectory": "equal mean of chunk metrics within each trajectory", |
| "selection": "fixed user checkpoints; no fitting or checkpoint selection on this rollout set", |
| "caveat": "chunks within a trajectory are dependent; only 20 trajectory samples" |
| } |
| } |
|
|
