added DiP checkpoints

checkpoints/dip/DiP_10steps_context20_predict40/args.json

@@ -0,0 +1,66 @@
+{
+    "adam_beta2": 0.999,
+    "arch": "trans_dec",
+    "autoregressive": true,
+    "autoregressive_include_prefix": false,
+    "autoregressive_init": "data",
+    "avg_model_beta": 0.9999,
+    "batch_size": 64,
+    "cond_mask_prob": 0.1,
+    "context_len": 20,
+    "cuda": true,
+    "data_dir": "",
+    "dataset": "humanml",
+    "device": 0,
+    "diffusion_steps": 10,
+    "emb_before_mask": false,
+    "emb_policy": "add",
+    "emb_trans_dec": false,
+    "eval_batch_size": 32,
+    "eval_during_training": true,
+    "eval_num_samples": 1000,
+    "eval_rep_times": 3,
+    "eval_split": "test",
+    "external_mode": false,
+    "gen_during_training": true,
+    "gen_guidance_param": 2.5,
+    "gen_num_repetitions": 2,
+    "gen_num_samples": 3,
+    "hml_type": null,
+    "keyframe_cond_prob": 0.5,
+    "keyframe_cond_type": "",
+    "lambda_fc": 0.0,
+    "lambda_rcxyz": 0.0,
+    "lambda_target_loc": 0.0,
+    "lambda_vel": 0.0,
+    "latent_dim": 512,
+    "layers": 8,
+    "log_interval": 1000,
+    "lr": 0.0001,
+    "lr_anneal_steps": 0,
+    "mask_frames": true,
+    "multi_encoder_type": "multi",
+    "multi_target_cond": false,
+    "noise_schedule": "cosine",
+    "num_frames": 60,
+    "num_steps": 600000,
+    "overwrite": true,
+    "pos_embed_max_len": 5000,
+    "pred_len": 40,
+    "resume_checkpoint": "",
+    "sampling_mode": "none",
+    "save_dir": "save/DiP_10steps_context20_predict40",
+    "save_interval": 50000,
+    "seed": 10,
+    "sigma_small": true,
+    "spatial_condition": null,
+    "target_enc_layers": 1,
+    "target_joint_names": null,
+    "text_encoder_type": "bert",
+    "train_platform_type": "WandBPlatform",
+    "unconstrained": false,
+    "use_ema": true,
+    "use_inpainting": false,
+    "use_recon_guidance": false,
+    "weight_decay": 0.0
+}

checkpoints/dip/DiP_10steps_context20_predict40/eval_humanml_PreComHorRerAut_DifSte_10_ConLen_20_PreLen_40_000600343_gscale7.5_wo_mm.log

@@ -0,0 +1,247 @@
+==================== Replication 0 ====================
+Time: 2024-10-30 13:21:39.216134
+---> [ground truth] Matching Score: 3.2365
+---> [ground truth] R_precision: (top 1): 0.4543 (top 2): 0.6526 (top 3): 0.7612 
+---> [vald] Matching Score: 3.1560
+---> [vald] R_precision: (top 1): 0.4688 (top 2): 0.6592 (top 3): 0.7764 
+Time: 2024-10-30 13:21:43.823692
+---> [ground truth] FID: 0.0010
+---> [vald] FID: 0.3887
+Time: 2024-10-30 13:21:47.421551
+---> [ground truth] Diversity: 9.3780
+---> [vald] Diversity: 8.9154
+AUX METRICS:
+---> [vald] skate_ratio: 0.0721
+---> [vald] mean_penetration: 6.5568
+---> [vald] penetration: 0.1228
+---> [vald] floating: 24.3247
+---> [vald] skating: 0.5522
+AUX METRICS:
+---> [vald] fps: 2594.5557
+---> [vald] time_per_sample_ms: 15.4169
+!!! DONE !!!
+==================== Replication 1 ====================
+Time: 2024-10-30 13:23:16.167503
+---> [ground truth] Matching Score: 3.2482
+---> [ground truth] R_precision: (top 1): 0.4603 (top 2): 0.6550 (top 3): 0.7675 
+---> [vald] Matching Score: 3.1117
+---> [vald] R_precision: (top 1): 0.4658 (top 2): 0.6758 (top 3): 0.7715 
+Time: 2024-10-30 13:23:20.446822
+---> [ground truth] FID: 0.0012
+---> [vald] FID: 0.2311
+Time: 2024-10-30 13:23:24.204251
+---> [ground truth] Diversity: 9.1760
+---> [vald] Diversity: 9.2977
+AUX METRICS:
+---> [vald] skate_ratio: 0.0777
+---> [vald] mean_penetration: 6.1410
+---> [vald] penetration: 0.0525
+---> [vald] floating: 23.3839
+---> [vald] skating: 0.6070
+AUX METRICS:
+---> [vald] fps: 3102.4554
+---> [vald] time_per_sample_ms: 12.8930
+!!! DONE !!!
+==================== Replication 2 ====================
+Time: 2024-10-30 13:24:43.358279
+---> [ground truth] Matching Score: 3.2179
+---> [ground truth] R_precision: (top 1): 0.4612 (top 2): 0.6675 (top 3): 0.7700 
+---> [vald] Matching Score: 3.1865
+---> [vald] R_precision: (top 1): 0.4746 (top 2): 0.6680 (top 3): 0.7744 
+Time: 2024-10-30 13:24:47.581610
+---> [ground truth] FID: 0.0011
+---> [vald] FID: 0.3562
+Time: 2024-10-30 13:24:51.264851
+---> [ground truth] Diversity: 9.5500
+---> [vald] Diversity: 9.1119
+AUX METRICS:
+---> [vald] skate_ratio: 0.0770
+---> [vald] mean_penetration: 6.2946
+---> [vald] penetration: 0.0656
+---> [vald] floating: 25.7569
+---> [vald] skating: 0.6014
+AUX METRICS:
+---> [vald] fps: 3148.2694
+---> [vald] time_per_sample_ms: 12.7054
+!!! DONE !!!
+==================== Replication 3 ====================
+Time: 2024-10-30 13:26:13.869897
+---> [ground truth] Matching Score: 3.2363
+---> [ground truth] R_precision: (top 1): 0.4550 (top 2): 0.6558 (top 3): 0.7666 
+---> [vald] Matching Score: 3.0945
+---> [vald] R_precision: (top 1): 0.4609 (top 2): 0.6738 (top 3): 0.7803 
+Time: 2024-10-30 13:26:18.041856
+---> [ground truth] FID: 0.0010
+---> [vald] FID: 0.2497
+Time: 2024-10-30 13:26:21.701860
+---> [ground truth] Diversity: 9.2328
+---> [vald] Diversity: 9.1492
+AUX METRICS:
+---> [vald] skate_ratio: 0.0744
+---> [vald] mean_penetration: 8.1887
+---> [vald] penetration: 0.1200
+---> [vald] floating: 25.1052
+---> [vald] skating: 0.8081
+AUX METRICS:
+---> [vald] fps: 2795.6479
+---> [vald] time_per_sample_ms: 14.3080
+!!! DONE !!!
+==================== Replication 4 ====================
+Time: 2024-10-30 13:27:42.079014
+---> [ground truth] Matching Score: 3.2482
+---> [ground truth] R_precision: (top 1): 0.4418 (top 2): 0.6453 (top 3): 0.7545 
+---> [vald] Matching Score: 3.1482
+---> [vald] R_precision: (top 1): 0.4648 (top 2): 0.6475 (top 3): 0.7637 
+Time: 2024-10-30 13:27:46.498248
+---> [ground truth] FID: 0.0011
+---> [vald] FID: 0.3483
+Time: 2024-10-30 13:27:50.715379
+---> [ground truth] Diversity: 9.2800
+---> [vald] Diversity: 9.0324
+AUX METRICS:
+---> [vald] skate_ratio: 0.0786
+---> [vald] mean_penetration: 6.1403
+---> [vald] penetration: 0.0881
+---> [vald] floating: 23.3769
+---> [vald] skating: 0.6603
+AUX METRICS:
+---> [vald] fps: 2634.4592
+---> [vald] time_per_sample_ms: 15.1834
+!!! DONE !!!
+==================== Replication 5 ====================
+Time: 2024-10-30 13:29:11.587206
+---> [ground truth] Matching Score: 3.2448
+---> [ground truth] R_precision: (top 1): 0.4573 (top 2): 0.6578 (top 3): 0.7636 
+---> [vald] Matching Score: 3.0923
+---> [vald] R_precision: (top 1): 0.4658 (top 2): 0.6729 (top 3): 0.7861 
+Time: 2024-10-30 13:29:15.778495
+---> [ground truth] FID: 0.0011
+---> [vald] FID: 0.1868
+Time: 2024-10-30 13:29:19.420641
+---> [ground truth] Diversity: 9.6603
+---> [vald] Diversity: 8.9310
+AUX METRICS:
+---> [vald] skate_ratio: 0.0734
+---> [vald] mean_penetration: 6.2195
+---> [vald] penetration: 0.0743
+---> [vald] floating: 20.6116
+---> [vald] skating: 0.6368
+AUX METRICS:
+---> [vald] fps: 2977.0590
+---> [vald] time_per_sample_ms: 13.4361
+!!! DONE !!!
+==================== Replication 6 ====================
+Time: 2024-10-30 13:30:41.583496
+---> [ground truth] Matching Score: 3.2623
+---> [ground truth] R_precision: (top 1): 0.4487 (top 2): 0.6532 (top 3): 0.7601 
+---> [vald] Matching Score: 3.1806
+---> [vald] R_precision: (top 1): 0.4561 (top 2): 0.6680 (top 3): 0.7646 
+Time: 2024-10-30 13:30:45.795327
+---> [ground truth] FID: 0.0011
+---> [vald] FID: 0.3755
+Time: 2024-10-30 13:30:49.698681
+---> [ground truth] Diversity: 9.3130
+---> [vald] Diversity: 9.3665
+AUX METRICS:
+---> [vald] skate_ratio: 0.0777
+---> [vald] mean_penetration: 5.3331
+---> [vald] penetration: 0.0656
+---> [vald] floating: 23.0651
+---> [vald] skating: 0.5678
+AUX METRICS:
+---> [vald] fps: 3191.6156
+---> [vald] time_per_sample_ms: 12.5328
+!!! DONE !!!
+==================== Replication 7 ====================
+Time: 2024-10-30 13:32:10.410233
+---> [ground truth] Matching Score: 3.2375
+---> [ground truth] R_precision: (top 1): 0.4541 (top 2): 0.6524 (top 3): 0.7614 
+---> [vald] Matching Score: 3.2039
+---> [vald] R_precision: (top 1): 0.4600 (top 2): 0.6582 (top 3): 0.7822 
+Time: 2024-10-30 13:32:15.110242
+---> [ground truth] FID: 0.0010
+---> [vald] FID: 0.2474
+Time: 2024-10-30 13:32:19.088898
+---> [ground truth] Diversity: 9.3622
+---> [vald] Diversity: 9.1560
+AUX METRICS:
+---> [vald] skate_ratio: 0.0726
+---> [vald] mean_penetration: 6.5664
+---> [vald] penetration: 0.1203
+---> [vald] floating: 24.3337
+---> [vald] skating: 0.6471
+AUX METRICS:
+---> [vald] fps: 3069.7370
+---> [vald] time_per_sample_ms: 13.0304
+!!! DONE !!!
+==================== Replication 8 ====================
+Time: 2024-10-30 13:33:38.730361
+---> [ground truth] Matching Score: 3.2570
+---> [ground truth] R_precision: (top 1): 0.4444 (top 2): 0.6530 (top 3): 0.7672 
+---> [vald] Matching Score: 3.1620
+---> [vald] R_precision: (top 1): 0.4531 (top 2): 0.6650 (top 3): 0.7832 
+Time: 2024-10-30 13:33:43.168566
+---> [ground truth] FID: 0.0010
+---> [vald] FID: 0.2327
+Time: 2024-10-30 13:33:47.068348
+---> [ground truth] Diversity: 9.0245
+---> [vald] Diversity: 9.6330
+AUX METRICS:
+---> [vald] skate_ratio: 0.0693
+---> [vald] mean_penetration: 6.0143
+---> [vald] penetration: 0.0626
+---> [vald] floating: 24.9363
+---> [vald] skating: 0.6472
+AUX METRICS:
+---> [vald] fps: 3125.6639
+---> [vald] time_per_sample_ms: 12.7973
+!!! DONE !!!
+==================== Replication 9 ====================
+Time: 2024-10-30 13:35:09.757800
+---> [ground truth] Matching Score: 3.2334
+---> [ground truth] R_precision: (top 1): 0.4578 (top 2): 0.6619 (top 3): 0.7716 
+---> [vald] Matching Score: 3.1542
+---> [vald] R_precision: (top 1): 0.4736 (top 2): 0.6875 (top 3): 0.7871 
+Time: 2024-10-30 13:35:13.991154
+---> [ground truth] FID: 0.0010
+---> [vald] FID: 0.2097
+Time: 2024-10-30 13:35:17.637110
+---> [ground truth] Diversity: 9.2863
+---> [vald] Diversity: 9.5178
+AUX METRICS:
+---> [vald] skate_ratio: 0.0729
+---> [vald] mean_penetration: 5.4497
+---> [vald] penetration: 0.0621
+---> [vald] floating: 21.2150
+---> [vald] skating: 0.5615
+AUX METRICS:
+---> [vald] fps: 3223.9078
+---> [vald] time_per_sample_ms: 12.4073
+!!! DONE !!!
+========== Matching Score Summary ==========
+---> [ground truth] Mean: 3.2422 CInterval: 0.0075
+---> [vald] Mean: 3.1490 CInterval: 0.0226
+========== R_precision Summary ==========
+---> [ground truth](top 1) Mean: 0.4535 CInt: 0.0038;(top 2) Mean: 0.6554 CInt: 0.0035;(top 3) Mean: 0.7644 CInt: 0.0030;
+---> [vald](top 1) Mean: 0.4644 CInt: 0.0041;(top 2) Mean: 0.6676 CInt: 0.0065;(top 3) Mean: 0.7770 CInt: 0.0049;
+========== FID Summary ==========
+---> [ground truth] Mean: 0.0011 CInterval: 0.0000
+---> [vald] Mean: 0.2826 CInterval: 0.0445
+========== Diversity Summary ==========
+---> [ground truth] Mean: 9.3263 CInterval: 0.1059
+---> [vald] Mean: 9.2111 CInterval: 0.1411
+========== MultiModality Summary ==========
+========== skate_ratio Summary ==========
+---> [vald] Mean: 0.0746 CInterval: 0.0018
+========== mean_penetration Summary ==========
+---> [vald] Mean: 6.2904 CInterval: 0.4597
+========== penetration Summary ==========
+---> [vald] Mean: 0.0834 CInterval: 0.0162
+========== floating Summary ==========
+---> [vald] Mean: 23.6109 CInterval: 0.9754
+========== skating Summary ==========
+---> [vald] Mean: 0.6289 CInterval: 0.0435
+========== fps Summary ==========
+---> [vald] Mean: 2986.3371 CInterval: 135.5679
+========== time_per_sample_ms Summary ==========
+---> [vald] Mean: 13.4711 CInterval: 0.6501

checkpoints/dip/DiP_10steps_context20_predict40/model000200000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d667b9124e57a5906a3fd3ce45934a2df0e475bc57395dd63bb76e3641491b3
+size 232011477

checkpoints/dip/DiP_10steps_context20_predict40/model000600343.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcf3e482a5236a21731af571c66bd7d0274f5ddf0b6974825fa69333c77f3144
+size 232011477

checkpoints/dip/DiP_10steps_context20_predict40/opt000200000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c93be3d949cb6a9b0cd0eb93408b171533354d43eef9d6cf4877b0a0a54656c
+size 211523685

checkpoints/dip/DiP_10steps_context20_predict40/opt000600343.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91f18fb1cafffe861a3197f0d612b2e2d19efd8349846901bd60b7463e9af77e
+size 211523685