anhdao69 commited on
Commit
05d793d
·
verified ·
1 Parent(s): 12ce5eb

Index libero-all-suites/init_variants/core_mean

Browse files
Files changed (2) hide show
  1. README.md +3 -1
  2. checkpoint_index.json +36 -1
README.md CHANGED
@@ -9,7 +9,7 @@ tags:
9
 
10
  # LoopWAM final checkpoints
11
 
12
- Final trained policy weights from the LoopWAM experiments. This repository contains **22 checkpoints**, separated by training data scope. Both original and repeated 4/4 runs are retained. All production checkpoints are from completed ten-epoch runs with global batch 128 and training seed 42.
13
 
14
  ## October 9 additions: all-loop KV
15
 
@@ -77,6 +77,7 @@ Dense-S12 / Dense-S30 trained on all four suites with the same data, normalizati
77
 
78
  | Directory | Model | Video/action loops | Parameters | Evaluation |
79
  |---|---|---:|---:|---|
 
80
  | [libero-all-suites/init_variants/core16_21](libero-all-suites/init_variants/core16_21) | v0 init `core16_21` | 4/4 | 584,536,135 | Training only; evaluation not run |
81
  | [libero-all-suites/init_variants/no_video_pretraining](libero-all-suites/init_variants/no_video_pretraining) | v0 init `no_video_pretraining` (no Wan video pretraining) | 4/4 | 584,536,135 | Training only; evaluation not run |
82
 
@@ -94,6 +95,7 @@ Aligned (late) 4/2 v0 trained on all four suites with the same data, normalizati
94
 
95
  | Directory | Model | Video/action loops | Parameters | Final evaluation |
96
  |---|---|---|---:|---|
 
97
  | [libero-all-suites/v0-video4-action2](libero-all-suites/v0-video4-action2) | v0 aligned | 4/2 | 584,536,135 | Training only; evaluation not run |
98
  | [libero-all-suites/init_variants/core16_21](libero-all-suites/init_variants/core16_21) | v0 init `core16_21` | 4/4 | 584,536,135 | Training only; evaluation not run |
99
  | [libero-all-suites/init_variants/no_video_pretraining](libero-all-suites/init_variants/no_video_pretraining) | v0 init `no_video_pretraining` (no Wan video pretraining) | 4/4 | 584,536,135 | Training only; evaluation not run |
 
9
 
10
  # LoopWAM final checkpoints
11
 
12
+ Final trained policy weights from the LoopWAM experiments. This repository contains **23 checkpoints**, separated by training data scope. Both original and repeated 4/4 runs are retained. All production checkpoints are from completed ten-epoch runs with global batch 128 and training seed 42.
13
 
14
  ## October 9 additions: all-loop KV
15
 
 
77
 
78
  | Directory | Model | Video/action loops | Parameters | Evaluation |
79
  |---|---|---:|---:|---|
80
+ | [libero-all-suites/init_variants/core_mean](libero-all-suites/init_variants/core_mean) | v0 init `core_mean` | 4/4 | 584,536,135 | Training only; evaluation not run |
81
  | [libero-all-suites/init_variants/core16_21](libero-all-suites/init_variants/core16_21) | v0 init `core16_21` | 4/4 | 584,536,135 | Training only; evaluation not run |
82
  | [libero-all-suites/init_variants/no_video_pretraining](libero-all-suites/init_variants/no_video_pretraining) | v0 init `no_video_pretraining` (no Wan video pretraining) | 4/4 | 584,536,135 | Training only; evaluation not run |
83
 
 
95
 
96
  | Directory | Model | Video/action loops | Parameters | Final evaluation |
97
  |---|---|---|---:|---|
98
+ | [libero-all-suites/init_variants/core_mean](libero-all-suites/init_variants/core_mean) | v0 init `core_mean` | 4/4 | 584,536,135 | Training only; evaluation not run |
99
  | [libero-all-suites/v0-video4-action2](libero-all-suites/v0-video4-action2) | v0 aligned | 4/2 | 584,536,135 | Training only; evaluation not run |
100
  | [libero-all-suites/init_variants/core16_21](libero-all-suites/init_variants/core16_21) | v0 init `core16_21` | 4/4 | 584,536,135 | Training only; evaluation not run |
101
  | [libero-all-suites/init_variants/no_video_pretraining](libero-all-suites/init_variants/no_video_pretraining) | v0 init `no_video_pretraining` (no Wan video pretraining) | 4/4 | 584,536,135 | Training only; evaluation not run |
checkpoint_index.json CHANGED
@@ -657,7 +657,42 @@
657
  "microbatch": 8,
658
  "gradient_accumulation": 8,
659
  "created_utc": "2026-10-27T17:49:39.224168+00:00"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
660
  }
661
  ],
662
- "updated_utc": "2026-10-27T17:51:12.006480+00:00"
663
  }
 
657
  "microbatch": 8,
658
  "gradient_accumulation": 8,
659
  "created_utc": "2026-10-27T17:49:39.224168+00:00"
660
+ },
661
+ {
662
+ "id": "v0_core_mean_full_libero_hpg",
663
+ "repo_path": "libero-all-suites/init_variants/core_mean",
664
+ "training_path": "/blue/yu954774.ucf/an221229.ucf/checkpoints/LoopWAM/runs/v0_init_variants/core_mean/train",
665
+ "source_sha256": "ba21e88007fb570577d90bd8c961a7e8eeb671fdcf35c9faf8d95e3bf466e66b",
666
+ "source_bytes": 7015676119,
667
+ "export_sha256": "7613e262c47bd848af49af6797fab37f0d6a199032312a08cafabaa36d004664",
668
+ "export_bytes": 2338749835,
669
+ "format": "loopwam-s-v1",
670
+ "optimizer_included": false,
671
+ "step": 21700,
672
+ "version": "v0",
673
+ "video_loops": 4,
674
+ "action_loops": 4,
675
+ "parameters": 584536135,
676
+ "init_variant": "core_mean",
677
+ "initialization_sha256": "506bcecfeac2f2cf2eadfccffdd68060bd5f5ccc24f85bf2333a959b0001bef0",
678
+ "initialization_mode": "wan_init_variant_core_mean_fresh_optimizer",
679
+ "dataset_scope": "full_libero",
680
+ "epochs": 10,
681
+ "global_batch": 128,
682
+ "training_seed": 42,
683
+ "evaluation_status": "not_run_training_only",
684
+ "success_rate": null,
685
+ "successes": null,
686
+ "episodes": 0,
687
+ "training_seconds": 45977.33255086746,
688
+ "tensor_equality_verified": true,
689
+ "source_revision": "4ebf38a3c65aa5cc9d6c07d94c52f91c51551b57",
690
+ "normalization_sha256": "d5ad351eb3272f416d34082e8af5483ba1118ceca355a37d9f268fdb269fb3e4",
691
+ "fairness_check_passed": true,
692
+ "gpu": "NVIDIA B200",
693
+ "world_size": 2,
694
+ "created_utc": "2026-10-27T23:03:15.755642+00:00"
695
  }
696
  ],
697
+ "updated_utc": "2026-10-27T23:03:15.755642+00:00"
698
  }