Initial commit

.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,77 @@
+---
+library_name: pytorch
+tags:
+  - motion-generation
+  - text-to-motion
+  - diffusion
+  - streaming
+  - humanml3d
+language:
+  - en
+---
+
+# FloodDiffusion 2
+
+**Efficient and Path Controllable Streaming Motion Generation**
+
+[Code and setup instructions](https://github.com/AlayaLab/FloodDiffusion2)
+
+FloodDiffusion 2 supports continuous text-conditioned motion generation with
+Partial Attention, diffusion-compatible geometric supervision, and root-path
+control. This release includes six FD2 checkpoints and the paired 263D LDF/VAE
+models for first-generation compatibility.
+
+| Checkpoint folder | Configuration | Training step | CFG |
+|---|---|---:|---:|
+| `humanml3d_fk_60k` | `df_humanml3d_263.yaml` | 60,000 | 4 |
+| `humanml3d_path_fk_55k` | `df_humanml3d_263_path.yaml` | 55,000 | 3 |
+| `humanml3d_babel_fk_200k` | `df_humanml3d_babel_263.yaml` | 200,000 | 4 |
+| `humanml3d_babel_path_200k` | `df_humanml3d_babel_263_path.yaml` | 200,000 | 4 |
+| `seed_fk_300k` | `df_seed_138.yaml` | 300,000 | 2 |
+| `seed_path_fk_300k` | `df_seed_138_path.yaml` | 300,000 | 2 |
+| `ldf_263` | `ldf_263.yaml` | 180,000 | 6 |
+| `vae_263` | `vae_263.yaml` | 2,250,000 | — |
+
+Each `checkpoints/<folder>/` contains `model.ckpt`, a matching `config.yaml`,
+and normalization statistics under `assets/`. FD2 folders also include the
+quadratic FK matrix `assets/W.npy`. Checkpoints retain model, EMA, optimizer and
+scheduler state.
+
+## Setup
+
+Clone the code repository and run the following in a Python 3.10+ environment:
+
+```bash
+python setup_project.py
+```
+
+The command installs Python requirements and downloads the four paper checkpoints
+(HumanML3D text/path and SEED text/path), their assets, and the
+[shared dependency archive](https://huggingface.co/AlayaLab/FloodDiffusion2/blob/main/deps.zip)
+for UMT5, T2M and GloVe. Add `--with-data` to also install the
+[prepared datasets](https://huggingface.co/datasets/AlayaLab/FloodDiffusion2-Data)
+and paired VAE tokens.
+
+For mesh rendering, obtain SMPL-H from its
+[official site](https://mano.is.tue.mpg.de/) and import the licensed neutral model:
+
+```bash
+python setup_project.py --smplh /path/to/neutral/model.npz
+```
+
+## Evaluation
+
+```bash
+python evaluate.py --config configs/df_humanml3d_263.yaml
+```
+
+For additional checkpoints, use their bundled `checkpoints/<folder>/config.yaml`.
+
+The selected YAML defines the checkpoint, dataset and evaluation settings.
+Use the corresponding configuration for other variants. LDF uses the paired
+VAE from this release.
+
+## Dependencies
+
+Third-party weights and datasets retain their respective licenses. SMPL-H is
+obtained separately from its official distributor.

assets_manifest.json

@@ -0,0 +1,327 @@
+{
+  "format_version": 1,
+  "repository": "AlayaLab/FloodDiffusion2",
+  "revision": "alaya-initial",
+  "models": [
+    {
+      "name": "vae_263",
+      "config": "configs/vae_263.yaml",
+      "step": 2250000,
+      "files": [
+        {
+          "path": "checkpoints/vae_263/model.ckpt",
+          "size": 350279555,
+          "sha256": "b80ce00b538b24b874dd15ac84fa4f34d9469fe7ac7f8d99e60bea276c3918d4"
+        },
+        {
+          "path": "checkpoints/vae_263/config.yaml",
+          "size": 2581,
+          "sha256": "7047f20640a97538d8629bf91cd12c667d7992740894e051501eb1badd0f098e"
+        },
+        {
+          "path": "checkpoints/vae_263/assets/Mean.npy",
+          "size": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/vae_263/assets/Std.npy",
+          "size": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        }
+      ]
+    },
+    {
+      "name": "ldf_263",
+      "config": "configs/ldf_263.yaml",
+      "step": 180000,
+      "files": [
+        {
+          "path": "checkpoints/ldf_263/model.ckpt",
+          "size": 2274793067,
+          "sha256": "94cc2ea9c0c29213a7f615e721c5fe5264e96d0a95f7f67adc0a5a5081f69437"
+        },
+        {
+          "path": "checkpoints/ldf_263/config.yaml",
+          "size": 4564,
+          "sha256": "4a252ff70dbd9a1864a9f0afb7e71f0842440d1cc8bcdf360ab1769cbcf65f41"
+        },
+        {
+          "path": "checkpoints/ldf_263/assets/Mean.npy",
+          "size": 144,
+          "sha256": "c8b1c3183972f4ba2e542eb8bc26945f5d13782d04b56345d6ab5c71e714dff6"
+        },
+        {
+          "path": "checkpoints/ldf_263/assets/Std.npy",
+          "size": 144,
+          "sha256": "a4234b3c2ca19566cd74b749e1c1ede5cbc2ab245cfcc487891b14c056a2357d"
+        }
+      ]
+    },
+    {
+      "name": "humanml3d_fk_60k",
+      "config": "configs/df_humanml3d_263.yaml",
+      "step": 60000,
+      "files": [
+        {
+          "path": "checkpoints/humanml3d_fk_60k/model.ckpt",
+          "size": 2285681258,
+          "sha256": "d48332bcd88d06098627ed895e8d708702290d6ed518b20e5573b6b959725a63"
+        },
+        {
+          "path": "checkpoints/humanml3d_fk_60k/config.yaml",
+          "size": 3811,
+          "sha256": "983449db24beb87f9d63fbde21927e7119990bf10d6af6e32324f1a5526a7b8f"
+        },
+        {
+          "path": "checkpoints/humanml3d_fk_60k/assets/Mean.npy",
+          "size": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_fk_60k/assets/Std.npy",
+          "size": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_fk_60k/assets/W.npy",
+          "size": 553480,
+          "sha256": "4a8411b10882a74d09479e20e8ec10d0208473127f23e9da4d3192e2fea145d6"
+        }
+      ]
+    },
+    {
+      "name": "humanml3d_path_fk_55k",
+      "config": "configs/df_humanml3d_263_path.yaml",
+      "step": 55000,
+      "files": [
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/model.ckpt",
+          "size": 2285614312,
+          "sha256": "ee1239ee5458781ab84d21329d190bfcb45a23d12bec17df48924e000672d104"
+        },
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/config.yaml",
+          "size": 3877,
+          "sha256": "efcb66b0ec75c721b418887fa9824285dcc8934dc98550218348303bda273332"
+        },
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/assets/Mean.npy",
+          "size": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/assets/Std.npy",
+          "size": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/assets/W.npy",
+          "size": 540928,
+          "sha256": "ebc7243c5185ab6203af8e0b912e05d6bd15919f9eda7391b548a870eefea373"
+        }
+      ]
+    },
+    {
+      "name": "humanml3d_babel_fk_200k",
+      "config": "configs/df_humanml3d_babel_263.yaml",
+      "step": 200000,
+      "files": [
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/model.ckpt",
+          "size": 2285685418,
+          "sha256": "216a6168daca6552495ff8415a7cc31d896e9dfb116e4165e1b777ac609e6ae3"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/config.yaml",
+          "size": 4434,
+          "sha256": "69ee7e4642b5baa3bba4e9e89e14a69c020cb936075ced8683be98426488eac3"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/assets/Mean.npy",
+          "size": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/assets/Std.npy",
+          "size": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/assets/W.npy",
+          "size": 553480,
+          "sha256": "778981f8025de72f2f98121ae018e81f69a8ce54fc0a2a8615c33706eb38d534"
+        }
+      ]
+    },
+    {
+      "name": "humanml3d_babel_path_200k",
+      "config": "configs/df_humanml3d_babel_263_path.yaml",
+      "step": 200000,
+      "files": [
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/model.ckpt",
+          "size": 2285618728,
+          "sha256": "d2e6a28baf6e1cfcd8c2061425ebdd82b1024091e9714729b087b428485cac0e"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/config.yaml",
+          "size": 4534,
+          "sha256": "a86c8536002233d68512a644c0affc21274d5f9610aff58479118ec0538d8e28"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/assets/Mean.npy",
+          "size": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/assets/Std.npy",
+          "size": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/assets/W.npy",
+          "size": 540928,
+          "sha256": "ebc7243c5185ab6203af8e0b912e05d6bd15919f9eda7391b548a870eefea373"
+        }
+      ]
+    },
+    {
+      "name": "seed_fk_300k",
+      "config": "configs/df_seed_138.yaml",
+      "step": 300000,
+      "files": [
+        {
+          "path": "checkpoints/seed_fk_300k/model.ckpt",
+          "size": 2280356842,
+          "sha256": "15caeff8faaa19f09437aee9f79088955708a32fd6ff609cc990338460dc130f"
+        },
+        {
+          "path": "checkpoints/seed_fk_300k/config.yaml",
+          "size": 3697,
+          "sha256": "41825321dce5aea7b3a384b4717b74aa7141c8e93988abc1987fd20bbdf6c686"
+        },
+        {
+          "path": "checkpoints/seed_fk_300k/assets/Mean.npy",
+          "size": 680,
+          "sha256": "98456f74f871856f48880c72d7340a776cb45dc88ba33e86491cfaf1f8bc9534"
+        },
+        {
+          "path": "checkpoints/seed_fk_300k/assets/Std.npy",
+          "size": 680,
+          "sha256": "5ea97221c9a6d471d12b92d6562d4214d1080fb398e96e43f628b393a70b2b96"
+        },
+        {
+          "path": "checkpoints/seed_fk_300k/assets/W.npy",
+          "size": 152480,
+          "sha256": "d5c81bdf26ff2347bb59840ab3bb52b179d9488fbb5ff88ea65f3e76fc9c5502"
+        }
+      ]
+    },
+    {
+      "name": "seed_path_fk_300k",
+      "config": "configs/df_seed_138_path.yaml",
+      "step": 300000,
+      "files": [
+        {
+          "path": "checkpoints/seed_path_fk_300k/model.ckpt",
+          "size": 2280293224,
+          "sha256": "d60a085a280962f66669fbeb12568e6f6cd1b67304647c1b73fc847862498ef6"
+        },
+        {
+          "path": "checkpoints/seed_path_fk_300k/config.yaml",
+          "size": 3811,
+          "sha256": "2c414967a1e1ad2d6e959e4a1ce7fc22d43086a69340d245f16a5e41bcef850e"
+        },
+        {
+          "path": "checkpoints/seed_path_fk_300k/assets/Mean.npy",
+          "size": 680,
+          "sha256": "98456f74f871856f48880c72d7340a776cb45dc88ba33e86491cfaf1f8bc9534"
+        },
+        {
+          "path": "checkpoints/seed_path_fk_300k/assets/Std.npy",
+          "size": 680,
+          "sha256": "5ea97221c9a6d471d12b92d6562d4214d1080fb398e96e43f628b393a70b2b96"
+        },
+        {
+          "path": "checkpoints/seed_path_fk_300k/assets/W.npy",
+          "size": 145928,
+          "sha256": "0effaf2383a21f4e5a1bf03aed5841060cd45226ac93177e8e09ecd9dbf540b5"
+        }
+      ]
+    }
+  ],
+  "dependencies": {
+    "repository": "AlayaLab/FloodDiffusion2",
+    "revision": "alaya-initial",
+    "archive": "deps.zip",
+    "size": 12992154753,
+    "sha256": "bd12f617474bf9f2a4321308b6ee3697b34b11a4ce77f47ff7d50db0d6fe7d85",
+    "files": [
+      {
+        "path": "deps/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth",
+        "size": 11361920418,
+        "crc32": "201263e2"
+      },
+      {
+        "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/special_tokens_map.json",
+        "size": 6623,
+        "crc32": "52b4332b"
+      },
+      {
+        "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/spiece.model",
+        "size": 4548313,
+        "crc32": "3d61aeda"
+      },
+      {
+        "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/tokenizer.json",
+        "size": 16837417,
+        "crc32": "df89c52f"
+      },
+      {
+        "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/tokenizer_config.json",
+        "size": 61728,
+        "crc32": "882574af"
+      },
+      {
+        "path": "deps/glove/our_vab_data.npy",
+        "size": 10077728,
+        "crc32": "16c5f298"
+      },
+      {
+        "path": "deps/glove/our_vab_idx.pkl",
+        "size": 79811,
+        "crc32": "d99790b0"
+      },
+      {
+        "path": "deps/glove/our_vab_words.pkl",
+        "size": 67470,
+        "crc32": "b473f73b"
+      },
+      {
+        "path": "deps/t2m/humanml3d/motion_encoder.pt",
+        "size": 62995445,
+        "crc32": "729e63ff"
+      },
+      {
+        "path": "deps/t2m/humanml3d/movement_encoder.pt",
+        "size": 7374081,
+        "crc32": "50493a9c"
+      },
+      {
+        "path": "deps/t2m/humanml3d/text_encoder.pt",
+        "size": 16406745,
+        "crc32": "b1579742"
+      },
+      {
+        "path": "deps/t2m/meta/mean.npy",
+        "size": 2232,
+        "crc32": "fe0c91c4"
+      },
+      {
+        "path": "deps/t2m/meta/std.npy",
+        "size": 2232,
+        "crc32": "8e2a77ab"
+      }
+    ]
+  }
+}

checkpoints/humanml3d_babel_fk_200k/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0
+size 1180

checkpoints/humanml3d_babel_fk_200k/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20
+size 1180

checkpoints/humanml3d_babel_fk_200k/assets/W.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:778981f8025de72f2f98121ae018e81f69a8ce54fc0a2a8615c33706eb38d534
+size 553480

checkpoints/humanml3d_babel_fk_200k/config.yaml

@@ -0,0 +1,165 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+  model_dir: ${dirs.checkpoints}/humanml3d_babel_fk_200k
+exp_name: df_humanml3d_babel_263
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/humanml3d_babel_fk_200k/model.ckpt
+representation: humanml3d263
+test_setting:
+  render: true
+  HumanML3D:
+    compare_folders:
+    - ${dirs.raw_data}/HumanML3D/HumanML3D263/renders
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 300000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 100
+  precision: bf16-mixed
+  num_nodes: 1
+validation:
+  validation_steps: 5000
+  test_steps: 5000
+  save_every_n_steps: 5000
+  save_top_k: 100
+metrics:
+  t2m:
+    target: metrics.HumanML3D263.t2m.T2MMetrics
+    fid_target: original
+    params:
+      evaluate_text: true
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      wordvectorizer:
+        target: metrics.HumanML3D263.word_vectorizer.WordVectorizer
+        params:
+          meta_root: ${dirs.deps}/glove
+          prefix: our_vab
+        max_text_len: 20
+      textencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.TextEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/text_encoder.pt
+        params:
+          word_size: 300
+          pos_size: 15
+          hidden_size: 512
+          output_size: 512
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.multi.MultiDataset
+  collate_fn: datasets.multi.collate_fn
+  train_bs: 128
+  val_bs: 32
+  test_bs: 16
+  num_workers: 8
+  datasets:
+  - target: datasets.babel.BabelDataset
+    train_meta_paths:
+    - path: ${dirs.raw_data}/BABEL/HumanML3D263/train.txt
+      name: BABEL
+    val_meta_paths: []
+    test_meta_paths:
+    - path: ${dirs.raw_data}/BABEL/HumanML3D263/test_min.txt
+      name: BABEL
+    feature_path: new_joint_vecs
+    text_path: texts
+    random_length: 0
+    min_length: 40
+    max_length: 99999
+    window_length: 200
+    feature_fps: 20
+    token_fps: 5
+  - target: datasets.humanml3d.HumanML3DDataset
+    train_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/train.txt
+      name: HumanML3D
+    val_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test.txt
+      name: HumanML3D
+    test_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test_min.txt
+      name: HumanML3D
+    feature_path: new_joint_vecs
+    text_path: texts
+    random_length: 0
+    min_length: 40
+    max_length: 200
+    stream_mode: true
+    feature_fps: 20
+    token_fps: 5
+model:
+  target: models.diffusion_forcing_wan.DiffForcingWanModel
+  ema_decay: 0.99
+  params:
+    schedule_config:
+      noise_type: linear
+      chunk_size: 30
+      steps: 30
+      sigma_type: zero
+      sigma_scale: 1.0
+      random_epsilon: 0.05
+      train_n_windows: 4
+    text_config:
+      len: 512
+      dim: 4096
+      checkpoint_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth
+      tokenizer_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/google/umt5-xxl
+    input_dim: 263
+    mean_path: ${dirs.checkpoints}/humanml3d_babel_fk_200k/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/humanml3d_babel_fk_200k/assets/Std.npy
+    input_keys:
+      feature: feature
+      feature_length: feature_length
+      text: text
+      text_end: feature_text_end
+    cfg_config:
+      text_scale: 4.0
+      null_scale: -3.0
+    prediction_type: vel
+    attn_type: partial
+    loss_W: ${dirs.checkpoints}/humanml3d_babel_fk_200k/assets/W.npy
+    loss_w_coefficient: 1.0
+fk_matrix:
+  recipe: hml263_trace
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: CosineAnnealingLR
+  params:
+    T_max: 300000
+    eta_min: 1.0e-06

checkpoints/humanml3d_babel_fk_200k/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:216a6168daca6552495ff8415a7cc31d896e9dfb116e4165e1b777ac609e6ae3
+size 2285685418

checkpoints/humanml3d_babel_path_200k/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0
+size 1180

checkpoints/humanml3d_babel_path_200k/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20
+size 1180

checkpoints/humanml3d_babel_path_200k/assets/W.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebc7243c5185ab6203af8e0b912e05d6bd15919f9eda7391b548a870eefea373
+size 540928

checkpoints/humanml3d_babel_path_200k/config.yaml

@@ -0,0 +1,167 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+  model_dir: ${dirs.checkpoints}/humanml3d_babel_path_200k
+exp_name: df_humanml3d_babel_263_path
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/humanml3d_babel_path_200k/model.ckpt
+representation: humanml3d263
+test_setting:
+  render: true
+  HumanML3D:
+    compare_folders:
+    - ${dirs.raw_data}/HumanML3D/HumanML3D263/renders
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 300000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 100
+  precision: bf16-mixed
+  num_nodes: 1
+validation:
+  validation_steps: 5000
+  test_steps: 5000
+  save_every_n_steps: 5000
+  save_top_k: 100
+metrics:
+  t2m:
+    target: metrics.HumanML3D263.t2m.T2MMetrics
+    fid_target: original
+    params:
+      evaluate_text: true
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      wordvectorizer:
+        target: metrics.HumanML3D263.word_vectorizer.WordVectorizer
+        params:
+          meta_root: ${dirs.deps}/glove
+          prefix: our_vab
+        max_text_len: 20
+      textencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.TextEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/text_encoder.pt
+        params:
+          word_size: 300
+          pos_size: 15
+          hidden_size: 512
+          output_size: 512
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.multi.MultiDataset
+  collate_fn: datasets.multi.collate_fn
+  train_bs: 128
+  val_bs: 32
+  test_bs: 16
+  num_workers: 8
+  datasets:
+  - target: datasets.babel.BabelDataset
+    train_meta_paths:
+    - path: ${dirs.raw_data}/BABEL/HumanML3D263/train.txt
+      name: BABEL
+    val_meta_paths: []
+    test_meta_paths:
+    - path: ${dirs.raw_data}/BABEL/HumanML3D263/test_min.txt
+      name: BABEL
+    feature_path: new_joint_vecs
+    text_path: texts
+    random_length: 0
+    min_length: 40
+    max_length: 99999
+    window_length: 200
+    feature_fps: 20
+    token_fps: 5
+  - target: datasets.humanml3d_position.HumanML3DPositionDataset
+    train_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/train.txt
+      name: HumanML3D
+    val_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test.txt
+      name: HumanML3D
+    test_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test_min.txt
+      name: HumanML3D
+    feature_path: new_joint_vecs
+    text_path: texts
+    random_length: 0
+    min_length: 40
+    max_length: 200
+    stream_mode: true
+    feature_fps: 20
+    token_fps: 5
+model:
+  target: models.diffusion_forcing_position_wan.DiffForcingPositionWanModel
+  ema_decay: 0.99
+  params:
+    schedule_config:
+      noise_type: linear
+      chunk_size: 30
+      steps: 30
+      sigma_type: zero
+      sigma_scale: 1.0
+      random_epsilon: 0.05
+      train_n_windows: 4
+    text_config:
+      len: 512
+      dim: 4096
+      checkpoint_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth
+      tokenizer_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/google/umt5-xxl
+    input_dim: 263
+    mean_path: ${dirs.checkpoints}/humanml3d_babel_path_200k/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/humanml3d_babel_path_200k/assets/Std.npy
+    input_keys:
+      feature: feature
+      feature_length: feature_length
+      text: text
+      text_end: feature_text_end
+    cfg_config:
+      text_scale: 4.0
+      null_scale: -3.0
+    prediction_type: vel
+    attn_type: partial
+    representation: humanml3d263
+    root_dim: 3
+    loss_W: ${dirs.checkpoints}/humanml3d_babel_path_200k/assets/W.npy
+    loss_w_coefficient: 1.0
+fk_matrix:
+  recipe: hml263_path260
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: CosineAnnealingLR
+  params:
+    T_max: 300000
+    eta_min: 1.0e-06

checkpoints/humanml3d_babel_path_200k/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2e6a28baf6e1cfcd8c2061425ebdd82b1024091e9714729b087b428485cac0e
+size 2285618728

checkpoints/humanml3d_fk_60k/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0
+size 1180

checkpoints/humanml3d_fk_60k/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20
+size 1180

checkpoints/humanml3d_fk_60k/assets/W.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a8411b10882a74d09479e20e8ec10d0208473127f23e9da4d3192e2fea145d6
+size 553480

checkpoints/humanml3d_fk_60k/config.yaml

@@ -0,0 +1,144 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+exp_name: df_humanml3d_263
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/humanml3d_fk_60k/model.ckpt
+representation: humanml3d263
+test_setting:
+  render: true
+  HumanML3D:
+    compare_folders:
+    - ${dirs.raw_data}/HumanML3D/renders
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 100000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 100
+  precision: bf16-mixed
+  num_nodes: 1
+validation:
+  validation_steps: 5000
+  test_steps: 5000
+  save_every_n_steps: 5000
+  save_top_k: 100
+metrics:
+  t2m:
+    target: metrics.HumanML3D263.t2m.T2MMetrics
+    fid_target: original
+    params:
+      evaluate_text: true
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      wordvectorizer:
+        target: metrics.HumanML3D263.word_vectorizer.WordVectorizer
+        params:
+          meta_root: ${dirs.deps}/glove
+          prefix: our_vab
+        max_text_len: 20
+      textencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.TextEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/text_encoder.pt
+        params:
+          word_size: 300
+          pos_size: 15
+          hidden_size: 512
+          output_size: 512
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.humanml3d.HumanML3DDataset
+  collate_fn: datasets.humanml3d.collate_fn
+  train_bs: 128
+  val_bs: 32
+  test_bs: 16
+  num_workers: 8
+  train_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/train.txt
+    name: HumanML3D
+  val_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test.txt
+    name: HumanML3D
+  test_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test_min.txt
+    name: HumanML3D
+  feature_path: new_joint_vecs
+  text_path: texts
+  random_length: 0
+  min_length: 40
+  max_length: 200
+model:
+  target: models.diffusion_forcing_wan.DiffForcingWanModel
+  ema_decay: 0.99
+  params:
+    schedule_config:
+      noise_type: linear
+      chunk_size: 30
+      steps: 30
+      sigma_type: zero
+      sigma_scale: 1.0
+      random_epsilon: 0.05
+      train_n_windows: 8
+    text_config:
+      len: 512
+      dim: 4096
+      checkpoint_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth
+      tokenizer_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/google/umt5-xxl
+    input_dim: 263
+    loss_W: ${dirs.checkpoints}/humanml3d_fk_60k/assets/W.npy
+    loss_w_coefficient: 1.0
+    mean_path: ${dirs.checkpoints}/humanml3d_fk_60k/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/humanml3d_fk_60k/assets/Std.npy
+    input_keys:
+      feature: feature
+      feature_length: feature_length
+      text: text
+      text_end: feature_text_end
+    cfg_config:
+      text_scale: 4.0
+      null_scale: -3.0
+    prediction_type: vel
+    attn_type: partial
+    loss_W_require_trace_normalized: false
+fk_matrix:
+  recipe: hml263_pos64norm_rest1
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: CosineAnnealingLR
+  params:
+    T_max: 100000
+    eta_min: 1.0e-06

checkpoints/humanml3d_fk_60k/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d48332bcd88d06098627ed895e8d708702290d6ed518b20e5573b6b959725a63
+size 2285681258

checkpoints/humanml3d_path_fk_55k/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0
+size 1180

checkpoints/humanml3d_path_fk_55k/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20
+size 1180

checkpoints/humanml3d_path_fk_55k/assets/W.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebc7243c5185ab6203af8e0b912e05d6bd15919f9eda7391b548a870eefea373
+size 540928

checkpoints/humanml3d_path_fk_55k/config.yaml

@@ -0,0 +1,145 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+exp_name: df_humanml3d_263_path
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/humanml3d_path_fk_55k/model.ckpt
+representation: humanml3d263
+test_setting:
+  render: true
+  HumanML3D:
+    compare_folders:
+    - ${dirs.raw_data}/HumanML3D/renders
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 100000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 100
+  precision: bf16-mixed
+  num_nodes: 1
+validation:
+  validation_steps: 5000
+  test_steps: 5000
+  save_every_n_steps: 5000
+  save_top_k: 100
+metrics:
+  t2m:
+    target: metrics.HumanML3D263.t2m.T2MMetrics
+    fid_target: original
+    params:
+      evaluate_text: true
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      wordvectorizer:
+        target: metrics.HumanML3D263.word_vectorizer.WordVectorizer
+        params:
+          meta_root: ${dirs.deps}/glove
+          prefix: our_vab
+        max_text_len: 20
+      textencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.TextEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/text_encoder.pt
+        params:
+          word_size: 300
+          pos_size: 15
+          hidden_size: 512
+          output_size: 512
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.humanml3d_position.HumanML3DPositionDataset
+  collate_fn: datasets.humanml3d_position.collate_fn
+  train_bs: 128
+  val_bs: 32
+  test_bs: 16
+  num_workers: 8
+  train_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/train.txt
+    name: HumanML3D
+  val_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test.txt
+    name: HumanML3D
+  test_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test_min.txt
+    name: HumanML3D
+  feature_path: new_joint_vecs
+  text_path: texts
+  random_length: 0
+  min_length: 40
+  max_length: 200
+model:
+  target: models.diffusion_forcing_position_wan.DiffForcingPositionWanModel
+  ema_decay: 0.99
+  params:
+    schedule_config:
+      noise_type: linear
+      chunk_size: 30
+      steps: 30
+      sigma_type: zero
+      sigma_scale: 1.0
+      random_epsilon: 0.05
+      train_n_windows: 8
+    text_config:
+      len: 512
+      dim: 4096
+      checkpoint_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth
+      tokenizer_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/google/umt5-xxl
+    input_dim: 263
+    representation: humanml3d263
+    root_dim: 3
+    loss_W: ${dirs.checkpoints}/humanml3d_path_fk_55k/assets/W.npy
+    loss_w_coefficient: 1.0
+    mean_path: ${dirs.checkpoints}/humanml3d_path_fk_55k/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/humanml3d_path_fk_55k/assets/Std.npy
+    input_keys:
+      feature: feature
+      feature_length: feature_length
+      text: text
+      text_end: feature_text_end
+    cfg_config:
+      text_scale: 3.0
+      null_scale: -2.0
+    prediction_type: vel
+    attn_type: partial
+fk_matrix:
+  recipe: hml263_path260
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: CosineAnnealingLR
+  params:
+    T_max: 100000
+    eta_min: 1.0e-06

checkpoints/humanml3d_path_fk_55k/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee1239ee5458781ab84d21329d190bfcb45a23d12bec17df48924e000672d104
+size 2285614312

checkpoints/ldf_263/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8b1c3183972f4ba2e542eb8bc26945f5d13782d04b56345d6ab5c71e714dff6
+size 144

checkpoints/ldf_263/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4234b3c2ca19566cd74b749e1c1ede5cbc2ab245cfcc487891b14c056a2357d
+size 144

checkpoints/ldf_263/config.yaml

@@ -0,0 +1,174 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+exp_name: ldf_263
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/ldf_263/model.ckpt
+test_vae_ckpt: ${dirs.checkpoints}/vae_263/model.ckpt
+test_vae:
+  target: models.vae_wan.VAEWanModel
+  ema_decay: 0.99
+  params:
+    input_dim: 263
+    z_dim: 4
+test_setting:
+  render: true
+  recover_dim: 263
+  BABEL:
+    compare_folders:
+    - ${dirs.raw_data}/BABEL/HumanML3D263/animations
+    compare_names:
+    - Ground Truth
+  HumanML3D:
+    compare_folders:
+    - ${dirs.raw_data}/HumanML3D/HumanML3D263/animations
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 300000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 100
+  precision: bf16-mixed
+  num_nodes: 1
+validation:
+  validation_steps: 5000
+  test_steps: 5000
+  save_every_n_steps: 5000
+  save_top_k: 100
+metrics:
+  dim: 263
+  t2m:
+    target: metrics.HumanML3D263.t2m.T2MMetrics
+    fid_target: original
+    params:
+      evaluate_text: true
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      wordvectorizer:
+        target: metrics.HumanML3D263.word_vectorizer.WordVectorizer
+        params:
+          meta_root: ${dirs.deps}/glove
+          prefix: our_vab
+        max_text_len: 20
+      textencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.TextEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/text_encoder.pt
+        params:
+          word_size: 300
+          pos_size: 15
+          hidden_size: 512
+          output_size: 512
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.multi.MultiDataset
+  collate_fn: datasets.multi.collate_fn
+  train_bs: 32
+  val_bs: 16
+  test_bs: 16
+  num_workers: 8
+  datasets:
+  - target: datasets.babel.BabelDataset
+    train_meta_paths:
+    - path: ${dirs.raw_data}/BABEL/HumanML3D263/train.txt
+      name: BABEL
+    val_meta_paths: []
+    test_meta_paths:
+    - path: ${dirs.raw_data}/BABEL/HumanML3D263/test_min.txt
+      name: BABEL
+    feature_path: new_joint_vecs
+    token_path: TOKENS_20260113_233726_vae_wan_z4_2250000
+    text_path: texts
+    random_length: 0
+    min_length: 5
+    max_length: 99999
+    window_length: 200
+    feature_fps: 20
+    token_fps: 5
+  - target: datasets.humanml3d.HumanML3DDataset
+    train_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/train.txt
+      name: HumanML3D
+    val_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test.txt
+      name: HumanML3D
+    test_meta_paths:
+    - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test_min.txt
+      name: HumanML3D
+    feature_path: new_joint_vecs
+    token_path: TOKENS_20260113_233726_vae_wan_z4_2250000
+    text_path: texts
+    random_length: 0
+    min_length: 40
+    max_length: 200
+    stream_mode: true
+model:
+  target: models.diffusion_forcing_wan.DiffForcingWanModel
+  ema_decay: 0.99
+  params:
+    schedule_config:
+      noise_type: linear
+      chunk_size: 5
+      steps: 10
+      sigma_type: zero
+      sigma_scale: 1.0
+      random_epsilon: 0.05
+      train_n_windows: 4
+    text_config:
+      len: 512
+      dim: 4096
+      checkpoint_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth
+      tokenizer_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/google/umt5-xxl
+    input_dim: 4
+    mean_path: ${dirs.checkpoints}/ldf_263/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/ldf_263/assets/Std.npy
+    input_keys:
+      feature: token
+      feature_length: token_length
+      text: text
+      text_end: token_text_end
+    cfg_config:
+      text_scale: 6.0
+      null_scale: -5.0
+    prediction_type: vel
+    attn_type: partial
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: CosineAnnealingLR
+  params:
+    T_max: 1000
+    eta_min: 1.0e-06
+representation: humanml3d263

checkpoints/ldf_263/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94cc2ea9c0c29213a7f615e721c5fe5264e96d0a95f7f67adc0a5a5081f69437
+size 2274793067

checkpoints/seed_fk_300k/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98456f74f871856f48880c72d7340a776cb45dc88ba33e86491cfaf1f8bc9534
+size 680

checkpoints/seed_fk_300k/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ea97221c9a6d471d12b92d6562d4214d1080fb398e96e43f628b393a70b2b96
+size 680

checkpoints/seed_fk_300k/assets/W.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5c81bdf26ff2347bb59840ab3bb52b179d9488fbb5ff88ea65f3e76fc9c5502
+size 152480

checkpoints/seed_fk_300k/config.yaml

@@ -0,0 +1,143 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+exp_name: df_seed_138
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/seed_fk_300k/model.ckpt
+representation: mei138
+test_setting:
+  render: true
+  SEED:
+    compare_folders:
+    - ${dirs.raw_data}/SEED/MEI138/new_joint_vecs_uni
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 300000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 100
+  precision: bf16-mixed
+  num_nodes: 1
+validation:
+  validation_steps: 5000
+  test_steps: 5000
+  save_every_n_steps: 5000
+  save_top_k: 100
+metrics:
+  t2m:
+    target: metrics.MEI138.t2m.T2MMetrics
+    fid_target: original
+    params:
+      evaluate_text: true
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      wordvectorizer:
+        target: metrics.HumanML3D263.word_vectorizer.WordVectorizer
+        params:
+          meta_root: ${dirs.deps}/glove
+          prefix: our_vab
+        max_text_len: 20
+      textencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.TextEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/text_encoder.pt
+        params:
+          word_size: 300
+          pos_size: 15
+          hidden_size: 512
+          output_size: 512
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.humanml3d.HumanML3DDataset
+  collate_fn: datasets.humanml3d.collate_fn
+  train_bs: 128
+  val_bs: 32
+  test_bs: 16
+  num_workers: 8
+  train_meta_paths:
+  - path: ${dirs.raw_data}/SEED/MEI138/train.txt
+    name: SEED
+  val_meta_paths:
+  - path: ${dirs.raw_data}/SEED/MEI138/test_content.txt
+    name: SEED
+  test_meta_paths:
+  - path: ${dirs.raw_data}/SEED/MEI138/test_min.txt
+    name: SEED
+  feature_path: new_joint_vecs_uni
+  text_path: texts
+  random_length: 0
+  min_length: 60
+  max_length: 300
+model:
+  target: models.diffusion_forcing_wan.DiffForcingWanModel
+  ema_decay: 0.99
+  params:
+    schedule_config:
+      noise_type: linear
+      chunk_size: 30
+      steps: 30
+      sigma_type: zero
+      sigma_scale: 1.0
+      random_epsilon: 0.05
+      train_n_windows: 8
+    text_config:
+      len: 512
+      dim: 4096
+      checkpoint_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth
+      tokenizer_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/google/umt5-xxl
+    input_dim: 138
+    loss_W: ${dirs.checkpoints}/seed_fk_300k/assets/W.npy
+    loss_w_coefficient: 1.0
+    mean_path: ${dirs.checkpoints}/seed_fk_300k/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/seed_fk_300k/assets/Std.npy
+    input_keys:
+      feature: feature
+      feature_length: feature_length
+      text: text
+      text_end: feature_text_end
+    cfg_config:
+      text_scale: 2.0
+      null_scale: -1.0
+    prediction_type: vel
+    attn_type: partial
+fk_matrix:
+  recipe: seed138_mesh
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: CosineAnnealingLR
+  params:
+    T_max: 300000
+    eta_min: 1.0e-06

checkpoints/seed_fk_300k/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15caeff8faaa19f09437aee9f79088955708a32fd6ff609cc990338460dc130f
+size 2280356842

checkpoints/seed_path_fk_300k/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98456f74f871856f48880c72d7340a776cb45dc88ba33e86491cfaf1f8bc9534
+size 680

checkpoints/seed_path_fk_300k/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ea97221c9a6d471d12b92d6562d4214d1080fb398e96e43f628b393a70b2b96
+size 680

checkpoints/seed_path_fk_300k/assets/W.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0effaf2383a21f4e5a1bf03aed5841060cd45226ac93177e8e09ecd9dbf540b5
+size 145928

checkpoints/seed_path_fk_300k/config.yaml

@@ -0,0 +1,145 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+exp_name: df_seed_138_path
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/seed_path_fk_300k/model.ckpt
+representation: mei138
+test_setting:
+  render: true
+  SEED:
+    compare_folders:
+    - ${dirs.raw_data}/SEED/MEI138/new_joint_vecs_uni
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 300000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 100
+  precision: bf16-mixed
+  num_nodes: 1
+validation:
+  validation_steps: 5000
+  test_steps: 5000
+  save_every_n_steps: 5000
+  save_top_k: 100
+metrics:
+  t2m:
+    target: metrics.MEI138.t2m.T2MMetrics
+    fid_target: original
+    params:
+      evaluate_text: true
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      wordvectorizer:
+        target: metrics.HumanML3D263.word_vectorizer.WordVectorizer
+        params:
+          meta_root: ${dirs.deps}/glove
+          prefix: our_vab
+        max_text_len: 20
+      textencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.TextEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/text_encoder.pt
+        params:
+          word_size: 300
+          pos_size: 15
+          hidden_size: 512
+          output_size: 512
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.humanml3d_position.HumanML3DPositionDataset
+  collate_fn: datasets.humanml3d_position.collate_fn
+  train_bs: 128
+  val_bs: 32
+  test_bs: 16
+  num_workers: 8
+  train_meta_paths:
+  - path: ${dirs.raw_data}/SEED/MEI138/train.txt
+    name: SEED
+  val_meta_paths:
+  - path: ${dirs.raw_data}/SEED/MEI138/test_content.txt
+    name: SEED
+  test_meta_paths:
+  - path: ${dirs.raw_data}/SEED/MEI138/test_min.txt
+    name: SEED
+  feature_path: new_joint_vecs_uni
+  text_path: texts
+  random_length: 0
+  min_length: 60
+  max_length: 300
+model:
+  target: models.diffusion_forcing_position_wan.DiffForcingPositionWanModel
+  ema_decay: 0.99
+  params:
+    schedule_config:
+      noise_type: linear
+      chunk_size: 30
+      steps: 30
+      sigma_type: zero
+      sigma_scale: 1.0
+      random_epsilon: 0.05
+      train_n_windows: 8
+    text_config:
+      len: 512
+      dim: 4096
+      checkpoint_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth
+      tokenizer_path: ${dirs.deps}/t5_umt5-xxl-enc-bf16/google/umt5-xxl
+    input_dim: 138
+    representation: mei138
+    root_dim: 3
+    loss_W: ${dirs.checkpoints}/seed_path_fk_300k/assets/W.npy
+    loss_w_coefficient: 1.0
+    mean_path: ${dirs.checkpoints}/seed_path_fk_300k/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/seed_path_fk_300k/assets/Std.npy
+    input_keys:
+      feature: feature
+      feature_length: feature_length
+      text: text
+      text_end: feature_text_end
+    cfg_config:
+      text_scale: 2.0
+      null_scale: -1.0
+    prediction_type: vel
+    attn_type: partial
+fk_matrix:
+  recipe: seed138_path135
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: CosineAnnealingLR
+  params:
+    T_max: 300000
+    eta_min: 1.0e-06

checkpoints/seed_path_fk_300k/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d60a085a280962f66669fbeb12568e6f6cd1b67304647c1b73fc847862498ef6
+size 2280293224

checkpoints/vae_263/assets/Mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0
+size 1180

checkpoints/vae_263/assets/Std.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20
+size 1180

checkpoints/vae_263/config.yaml

@@ -0,0 +1,104 @@
+dirs:
+  deps: ./deps
+  raw_data: ./data
+  outputs: ./outputs
+  checkpoints: ./checkpoints
+exp_name: vae_263
+seed: 1234
+debug: false
+train: true
+save_dir: ${dirs.outputs}
+resume_ckpt: null
+test_ckpt: ${dirs.checkpoints}/vae_263/model.ckpt
+representation: humanml3d263
+test_setting:
+  render: true
+  HumanML3D:
+    compare_folders:
+    - ${dirs.raw_data}/HumanML3D/HumanML3D263/animations
+    compare_names:
+    - Ground Truth
+val_repeat: 1
+logger:
+  wandb:
+    wandb_key: ${oc.env:WANDB_API_KEY,null}
+    project: FloodDiffusion2
+    entity: ${oc.env:WANDB_ENTITY,null}
+trainer:
+  max_steps: 3000000
+  accelerator: gpu
+  devices: 1
+  log_every_n_steps: 50
+  num_nodes: 1
+  precision: 32-true
+validation:
+  validation_steps: 1000
+  test_steps: 10000
+  save_every_n_steps: 10000
+  save_top_k: 100
+metrics:
+  mr:
+    target: metrics.mr.MRMetrics
+  t2m:
+    target: metrics.HumanML3D263.t2m.T2MMetrics
+    params:
+      evaluate_text: false
+      metric_mean_path: ${dirs.deps}/t2m/meta/mean.npy
+      metric_std_path: ${dirs.deps}/t2m/meta/std.npy
+      moveencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MovementConvEncoder
+        ckpt: ${dirs.deps}/t2m/humanml3d/movement_encoder.pt
+        params:
+          input_size: 259
+          hidden_size: 512
+          output_size: 512
+      motionencoder:
+        target: metrics.HumanML3D263.t2m_evaluator.MotionEncoderBiGRUCo
+        ckpt: ${dirs.deps}/t2m/humanml3d/motion_encoder.pt
+        params:
+          input_size: 512
+          hidden_size: 1024
+          output_size: 512
+data:
+  target: datasets.humanml3d.HumanML3DDataset
+  collate_fn: datasets.humanml3d.collate_fn
+  train_bs: 256
+  val_bs: 64
+  test_bs: 16
+  num_workers: 8
+  train_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/train.txt
+    name: HumanML3D
+  val_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test.txt
+    name: HumanML3D
+  test_meta_paths:
+  - path: ${dirs.raw_data}/HumanML3D/HumanML3D263/test_min.txt
+    name: HumanML3D
+  feature_path: new_joint_vecs
+  token_path: null
+  text_path: texts
+  min_length: 40
+  max_length: 200
+  window_length: 189
+model:
+  target: models.vae_wan.VAEWanModel
+  ema_decay: 0.99
+  params:
+    input_dim: 263
+    z_dim: 4
+    mean_path: ${dirs.checkpoints}/vae_263/assets/Mean.npy
+    std_path: ${dirs.checkpoints}/vae_263/assets/Std.npy
+optimizer:
+  target: AdamW
+  params:
+    lr: 0.0002
+    betas:
+    - 0.9
+    - 0.99
+    weight_decay: 0.0
+    eps: 1.0e-08
+lr_scheduler:
+  target: diffusers.optimization.get_constant_schedule_with_warmup
+  params:
+    num_warmup_steps: 1000

checkpoints/vae_263/model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b80ce00b538b24b874dd15ac84fa4f34d9469fe7ac7f8d99e60bea276c3918d4
+size 350279555

dependencies.json

@@ -0,0 +1,74 @@
+{
+  "repository": "AlayaLab/FloodDiffusion2",
+  "revision": "alaya-initial",
+  "archive": "deps.zip",
+  "size": 12992154753,
+  "sha256": "bd12f617474bf9f2a4321308b6ee3697b34b11a4ce77f47ff7d50db0d6fe7d85",
+  "files": [
+    {
+      "path": "deps/t5_umt5-xxl-enc-bf16/models_t5_umt5-xxl-enc-bf16.pth",
+      "size": 11361920418,
+      "crc32": "201263e2"
+    },
+    {
+      "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/special_tokens_map.json",
+      "size": 6623,
+      "crc32": "52b4332b"
+    },
+    {
+      "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/spiece.model",
+      "size": 4548313,
+      "crc32": "3d61aeda"
+    },
+    {
+      "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/tokenizer.json",
+      "size": 16837417,
+      "crc32": "df89c52f"
+    },
+    {
+      "path": "deps/t5_umt5-xxl-enc-bf16/google/umt5-xxl/tokenizer_config.json",
+      "size": 61728,
+      "crc32": "882574af"
+    },
+    {
+      "path": "deps/glove/our_vab_data.npy",
+      "size": 10077728,
+      "crc32": "16c5f298"
+    },
+    {
+      "path": "deps/glove/our_vab_idx.pkl",
+      "size": 79811,
+      "crc32": "d99790b0"
+    },
+    {
+      "path": "deps/glove/our_vab_words.pkl",
+      "size": 67470,
+      "crc32": "b473f73b"
+    },
+    {
+      "path": "deps/t2m/humanml3d/motion_encoder.pt",
+      "size": 62995445,
+      "crc32": "729e63ff"
+    },
+    {
+      "path": "deps/t2m/humanml3d/movement_encoder.pt",
+      "size": 7374081,
+      "crc32": "50493a9c"
+    },
+    {
+      "path": "deps/t2m/humanml3d/text_encoder.pt",
+      "size": 16406745,
+      "crc32": "b1579742"
+    },
+    {
+      "path": "deps/t2m/meta/mean.npy",
+      "size": 2232,
+      "crc32": "fe0c91c4"
+    },
+    {
+      "path": "deps/t2m/meta/std.npy",
+      "size": 2232,
+      "crc32": "8e2a77ab"
+    }
+  ]
+}

deps.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd12f617474bf9f2a4321308b6ee3697b34b11a4ce77f47ff7d50db0d6fe7d85
+size 12992154753

publication_manifest.json

@@ -0,0 +1,402 @@
+{
+  "format_version": 1,
+  "release": "FloodDiffusion2",
+  "checkpoints": [
+    {
+      "config": "vae_263.yaml",
+      "global_step": 2250000,
+      "checkpoint": {
+        "path": "checkpoints/vae_263/model.ckpt",
+        "bytes": 350279555,
+        "sha256": "b80ce00b538b24b874dd15ac84fa4f34d9469fe7ac7f8d99e60bea276c3918d4"
+      },
+      "configuration": {
+        "path": "checkpoints/vae_263/config.yaml",
+        "bytes": 2581,
+        "sha256": "7047f20640a97538d8629bf91cd12c667d7992740894e051501eb1badd0f098e"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/vae_263/assets/Mean.npy",
+          "bytes": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/vae_263/assets/Std.npy",
+          "bytes": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        }
+      ],
+      "tensor_count_verified": 638,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 263,
+      "loss_matrix_validation": null
+    },
+    {
+      "config": "ldf_263.yaml",
+      "global_step": 180000,
+      "checkpoint": {
+        "path": "checkpoints/ldf_263/model.ckpt",
+        "bytes": 2274793067,
+        "sha256": "94cc2ea9c0c29213a7f615e721c5fe5264e96d0a95f7f67adc0a5a5081f69437"
+      },
+      "configuration": {
+        "path": "checkpoints/ldf_263/config.yaml",
+        "bytes": 4564,
+        "sha256": "4a252ff70dbd9a1864a9f0afb7e71f0842440d1cc8bcdf360ab1769cbcf65f41"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/ldf_263/assets/Mean.npy",
+          "bytes": 144,
+          "sha256": "c8b1c3183972f4ba2e542eb8bc26945f5d13782d04b56345d6ab5c71e714dff6"
+        },
+        {
+          "path": "checkpoints/ldf_263/assets/Std.npy",
+          "bytes": 144,
+          "sha256": "a4234b3c2ca19566cd74b749e1c1ede5cbc2ab245cfcc487891b14c056a2357d"
+        }
+      ],
+      "tensor_count_verified": 1388,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 4,
+      "loss_matrix_validation": null
+    },
+    {
+      "config": "df_humanml3d_263.yaml",
+      "global_step": 60000,
+      "checkpoint": {
+        "path": "checkpoints/humanml3d_fk_60k/model.ckpt",
+        "bytes": 2285681258,
+        "sha256": "d48332bcd88d06098627ed895e8d708702290d6ed518b20e5573b6b959725a63"
+      },
+      "configuration": {
+        "path": "checkpoints/humanml3d_fk_60k/config.yaml",
+        "bytes": 3811,
+        "sha256": "983449db24beb87f9d63fbde21927e7119990bf10d6af6e32324f1a5526a7b8f"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/humanml3d_fk_60k/assets/Mean.npy",
+          "bytes": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_fk_60k/assets/Std.npy",
+          "bytes": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_fk_60k/assets/W.npy",
+          "bytes": 553480,
+          "sha256": "4a8411b10882a74d09479e20e8ec10d0208473127f23e9da4d3192e2fea145d6"
+        }
+      ],
+      "tensor_count_verified": 1389,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 263,
+      "loss_matrix_validation": {
+        "dimension": 263,
+        "coefficient": 1.0,
+        "raw_matrix_trace": 263.0000009536743,
+        "trace_normalization_required": false,
+        "loader_roundtrip_fp32_exact": true
+      }
+    },
+    {
+      "config": "df_humanml3d_263_path.yaml",
+      "global_step": 55000,
+      "checkpoint": {
+        "path": "checkpoints/humanml3d_path_fk_55k/model.ckpt",
+        "bytes": 2285614312,
+        "sha256": "ee1239ee5458781ab84d21329d190bfcb45a23d12bec17df48924e000672d104"
+      },
+      "configuration": {
+        "path": "checkpoints/humanml3d_path_fk_55k/config.yaml",
+        "bytes": 3877,
+        "sha256": "efcb66b0ec75c721b418887fa9824285dcc8934dc98550218348303bda273332"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/assets/Mean.npy",
+          "bytes": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/assets/Std.npy",
+          "bytes": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_path_fk_55k/assets/W.npy",
+          "bytes": 540928,
+          "sha256": "ebc7243c5185ab6203af8e0b912e05d6bd15919f9eda7391b548a870eefea373"
+        }
+      ],
+      "tensor_count_verified": 1391,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 263,
+      "loss_matrix_validation": {
+        "dimension": 260,
+        "coefficient": 1.0,
+        "raw_matrix_trace": 259.9999974966049,
+        "trace_normalization_required": true,
+        "loader_roundtrip_fp32_exact": true
+      }
+    },
+    {
+      "config": "df_humanml3d_babel_263.yaml",
+      "global_step": 200000,
+      "checkpoint": {
+        "path": "checkpoints/humanml3d_babel_fk_200k/model.ckpt",
+        "bytes": 2285685418,
+        "sha256": "216a6168daca6552495ff8415a7cc31d896e9dfb116e4165e1b777ac609e6ae3"
+      },
+      "configuration": {
+        "path": "checkpoints/humanml3d_babel_fk_200k/config.yaml",
+        "bytes": 4434,
+        "sha256": "69ee7e4642b5baa3bba4e9e89e14a69c020cb936075ced8683be98426488eac3"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/assets/Mean.npy",
+          "bytes": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/assets/Std.npy",
+          "bytes": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_fk_200k/assets/W.npy",
+          "bytes": 553480,
+          "sha256": "778981f8025de72f2f98121ae018e81f69a8ce54fc0a2a8615c33706eb38d534"
+        }
+      ],
+      "tensor_count_verified": 1389,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 263,
+      "loss_matrix_validation": {
+        "dimension": 263,
+        "coefficient": 1.0,
+        "raw_matrix_trace": 263.0000002384186,
+        "trace_normalization_required": true,
+        "loader_roundtrip_fp32_exact": true
+      }
+    },
+    {
+      "config": "df_humanml3d_babel_263_path.yaml",
+      "global_step": 200000,
+      "checkpoint": {
+        "path": "checkpoints/humanml3d_babel_path_200k/model.ckpt",
+        "bytes": 2285618728,
+        "sha256": "d2e6a28baf6e1cfcd8c2061425ebdd82b1024091e9714729b087b428485cac0e"
+      },
+      "configuration": {
+        "path": "checkpoints/humanml3d_babel_path_200k/config.yaml",
+        "bytes": 4534,
+        "sha256": "a86c8536002233d68512a644c0affc21274d5f9610aff58479118ec0538d8e28"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/assets/Mean.npy",
+          "bytes": 1180,
+          "sha256": "bef8fea46b1c9ed21c378a378c66c5f6b55f7ecf2fdfa911670a7d020e7fffb0"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/assets/Std.npy",
+          "bytes": 1180,
+          "sha256": "041dee3a9e09c97e5f66e10fb4b63b9410fb57ecc609a556ec0b65d8def2fb20"
+        },
+        {
+          "path": "checkpoints/humanml3d_babel_path_200k/assets/W.npy",
+          "bytes": 540928,
+          "sha256": "ebc7243c5185ab6203af8e0b912e05d6bd15919f9eda7391b548a870eefea373"
+        }
+      ],
+      "tensor_count_verified": 1391,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 263,
+      "loss_matrix_validation": {
+        "dimension": 260,
+        "coefficient": 1.0,
+        "raw_matrix_trace": 259.9999974966049,
+        "trace_normalization_required": true,
+        "loader_roundtrip_fp32_exact": true
+      }
+    },
+    {
+      "config": "df_seed_138.yaml",
+      "global_step": 300000,
+      "checkpoint": {
+        "path": "checkpoints/seed_fk_300k/model.ckpt",
+        "bytes": 2280356842,
+        "sha256": "15caeff8faaa19f09437aee9f79088955708a32fd6ff609cc990338460dc130f"
+      },
+      "configuration": {
+        "path": "checkpoints/seed_fk_300k/config.yaml",
+        "bytes": 3697,
+        "sha256": "41825321dce5aea7b3a384b4717b74aa7141c8e93988abc1987fd20bbdf6c686"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/seed_fk_300k/assets/Mean.npy",
+          "bytes": 680,
+          "sha256": "98456f74f871856f48880c72d7340a776cb45dc88ba33e86491cfaf1f8bc9534"
+        },
+        {
+          "path": "checkpoints/seed_fk_300k/assets/Std.npy",
+          "bytes": 680,
+          "sha256": "5ea97221c9a6d471d12b92d6562d4214d1080fb398e96e43f628b393a70b2b96"
+        },
+        {
+          "path": "checkpoints/seed_fk_300k/assets/W.npy",
+          "bytes": 152480,
+          "sha256": "d5c81bdf26ff2347bb59840ab3bb52b179d9488fbb5ff88ea65f3e76fc9c5502"
+        }
+      ],
+      "tensor_count_verified": 1389,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 138,
+      "loss_matrix_validation": {
+        "dimension": 138,
+        "coefficient": 1.0,
+        "raw_matrix_trace": 137.99999964237213,
+        "trace_normalization_required": true,
+        "loader_roundtrip_fp32_exact": true
+      }
+    },
+    {
+      "config": "df_seed_138_path.yaml",
+      "global_step": 300000,
+      "checkpoint": {
+        "path": "checkpoints/seed_path_fk_300k/model.ckpt",
+        "bytes": 2280293224,
+        "sha256": "d60a085a280962f66669fbeb12568e6f6cd1b67304647c1b73fc847862498ef6"
+      },
+      "configuration": {
+        "path": "checkpoints/seed_path_fk_300k/config.yaml",
+        "bytes": 3811,
+        "sha256": "2c414967a1e1ad2d6e959e4a1ce7fc22d43086a69340d245f16a5e41bcef850e"
+      },
+      "assets": [
+        {
+          "path": "checkpoints/seed_path_fk_300k/assets/Mean.npy",
+          "bytes": 680,
+          "sha256": "98456f74f871856f48880c72d7340a776cb45dc88ba33e86491cfaf1f8bc9534"
+        },
+        {
+          "path": "checkpoints/seed_path_fk_300k/assets/Std.npy",
+          "bytes": 680,
+          "sha256": "5ea97221c9a6d471d12b92d6562d4214d1080fb398e96e43f628b393a70b2b96"
+        },
+        {
+          "path": "checkpoints/seed_path_fk_300k/assets/W.npy",
+          "bytes": 145928,
+          "sha256": "0effaf2383a21f4e5a1bf03aed5841060cd45226ac93177e8e09ecd9dbf540b5"
+        }
+      ],
+      "tensor_count_verified": 1391,
+      "preserved_training_state": [
+        "state_dict",
+        "ema_state",
+        "optimizer_states",
+        "lr_schedulers",
+        "loops",
+        "epoch",
+        "global_step"
+      ],
+      "removed_metadata": [
+        "callbacks"
+      ],
+      "normalization_dimension": 138,
+      "loss_matrix_validation": {
+        "dimension": 135,
+        "coefficient": 1.0,
+        "raw_matrix_trace": 135.0000001192093,
+        "trace_normalization_required": true,
+        "loader_roundtrip_fp32_exact": true
+      }
+    }
+  ],
+  "total_checkpoint_bytes": 16328322404,
+  "complete": true
+}