Upload config files for loading checkpoint when evaluation

config.yaml

@@ -0,0 +1,89 @@
+datasets:
+  vla_data:
+    CoT_prompt: Your task is {instruction}. To identify the key objects for your task.
+      Locate their bounding boxes in [x1,y1,x2,y2] format.
+    data_mix: fourier_gr1_unified_1000_two_history
+    data_root_dir: /data/RobotData/robocasa_1k_20hz
+    dataset_py: lerobot_datasets
+    delete_pause_frame: false
+    image_size:
+    - 224
+    - 224
+    per_device_batch_size: 10
+    training_task_weights:
+    - 1
+    - 1
+    - 1
+    - 1
+    use_delta_action: false
+framework:
+  action_model:
+    action_dim: 138
+    action_horizon: 16
+    action_model_type: DiT-L
+    add_pos_embed: true
+    diffusion_model_cfg:
+      cross_attention_dim: 2560
+      dropout: 0.2
+      final_dropout: true
+      interleave_self_attention: true
+      norm_type: ada_norm
+      num_layers: 16
+      output_dim: 2560
+      positional_embeddings: null
+    future_action_window_size: 15
+    future_obs_index: 16
+    hidden_size: 2560
+    max_num_embodiments: 32
+    max_seq_len: 1024
+    noise_beta_alpha: 1.5
+    noise_beta_beta: 1.0
+    noise_s: 0.999
+    num_inference_timesteps: 4
+    num_target_vision_tokens: 32
+    num_timestep_buckets: 1000
+    num_views: 1
+    obs_horizon: 2
+    obs_loss_weight: 1.0
+    only_policy: false
+    only_wo_video_gen: false
+    past_action_window_size: 0
+    policy_and_video_gen: false
+    state_dim: 58
+    vision_encoder_path: /World-Action-Model/pretrained
+    vision_encoder_size: s
+    vision_encoder_type: dinov3
+  name: QwenMMDiT
+  qwenvl:
+    base_vlm: pretrained/vlm/Qwen3-VL-4B-Instruct
+output_dir: world_model/checkpoints/lda/post-train/robocasa
+run_id: all_tasks
+run_root_dir: world_model/checkpoints/lda/post-train/robocasa
+seed: 42
+trainer:
+  eval_interval: 1000
+  freeze_modules: action_model.vision_encoder
+  gradient_accumulation_steps: 1
+  gradient_clipping: 1.0
+  is_resume: false
+  learning_rate:
+    action_model: 0.0001
+    base: 4.0e-05
+    qwen_vl_interface: 1.0e-05
+  logging_frequency: 100
+  lr_scheduler_type: cosine_with_min_lr
+  max_train_steps: 300000
+  num_warmup_steps: 15000
+  optimizer:
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 1.0e-08
+  pretrained_checkpoint: world_model/checkpoints/lda/pretrain/final_model/pytorch_model.pt
+  repeated_diffusion_steps: 4
+  save_interval: 5000
+  scheduler_specific_kwargs:
+    min_lr: 5.0e-07
+wandb_entity: Personal
+wandb_project: lda-robocasa

dataset_statistics.json

@@ -0,0 +1,413 @@
+{
+  "gr1": {
+    "action": {
+      "mean": [
+        -0.163161202489088,
+        0.2451432330223422,
+        -0.0891637501772493,
+        -1.4614050090312956,
+        0.14608537886912623,
+        0.09256105095846578,
+        0.021076124471922718,
+        -0.36896711634472007,
+        -0.3514934852719306,
+        0.09300540465240678,
+        -1.4121803939342499,
+        0.4152552008163184,
+        0.07537192405046274,
+        0.2081388249838104,
+        -0.14977236216266948,
+        -0.14977236216266948,
+        -0.14977236216266948,
+        -0.14977236216266948,
+        -0.29954472432533896,
+        0.75,
+        -0.4248296525329351,
+        -0.4248296525329351,
+        -0.4248296525329351,
+        -0.4248296525329351,
+        -0.8496593050658702,
+        3.0,
+        0.005974193549870203,
+        0.005230103122206251,
+        -0.0002079709078088854
+      ],
+      "std": [
+        0.3295375855877872,
+        0.16417262254509757,
+        0.23146453073009823,
+        0.39842306382564957,
+        0.262282522694757,
+        0.25443646729278696,
+        0.2810018872942849,
+        0.49840835424004803,
+        0.2978266793598536,
+        0.40325129654865804,
+        0.6566084540308043,
+        0.5236517016765893,
+        0.40723760206433485,
+        0.6276659193445018,
+        0.7349972944755316,
+        0.7349972944755316,
+        0.7349972944755316,
+        0.7349972944755316,
+        1.4699945889510633,
+        1.299038105676658,
+        1.438573732823235,
+        1.438573732823235,
+        1.438573732823235,
+        1.438573732823235,
+        2.87714746564647,
+        0.0,
+        0.06158721324753064,
+        0.018468748717223805,
+        0.005717544758609999
+      ],
+      "max": [
+        1.4058303833007812,
+        1.288877010345459,
+        1.2944501638412476,
+        3.1500876502832398e-06,
+        2.563594102859497,
+        1.5000286102294922,
+        1.4998135566711426,
+        1.564226508140564,
+        0.00015454324602615088,
+        1.7984894514083862,
+        5.365318065742031e-05,
+        3.0000767707824707,
+        1.4954047203063965,
+        1.4975833892822266,
+        1.5,
+        1.5,
+        1.5,
+        1.5,
+        3.0,
+        3.0,
+        1.5,
+        1.5,
+        1.5,
+        1.5,
+        3.0,
+        3.0,
+        0.9750414490699768,
+        0.34772318601608276,
+        0.5310259461402893
+      ],
+      "min": [
+        -1.710237741470337,
+        -0.00042632038821466267,
+        -1.4035195112228394,
+        -2.801373243331909,
+        -2.033860445022583,
+        -1.4939793348312378,
+        -1.3439968824386597,
+        -2.1020591259002686,
+        -2.2016870975494385,
+        -2.3005123138427734,
+        -2.848569631576538,
+        -3.000068426132202,
+        -1.4995447397232056,
+        -1.489180326461792,
+        -1.5,
+        -1.5,
+        -1.5,
+        -1.5,
+        -3.0,
+        0.0,
+        -1.5,
+        -1.5,
+        -1.5,
+        -1.5,
+        -3.0,
+        3.0,
+        -0.5332962870597839,
+        -0.4947744905948639,
+        -0.4128894805908203
+      ],
+      "q01": [
+        -1.6453754484653473,
+        -7.352335887844674e-05,
+        -1.1551570868492127,
+        -2.7360924339294432,
+        -0.8502070605754852,
+        -1.0754927098751068,
+        -1.2199868965148926,
+        -1.699393036365509,
+        -1.3113291025161744,
+        -0.9920944452285767,
+        -2.7749218344688416,
+        -0.9589024305343627,
+        -1.3624527990818023,
+        -1.177962738275528,
+        -1.5,
+        -1.5,
+        -1.5,
+        -1.5,
+        -3.0,
+        0.0,
+        -1.5,
+        -1.5,
+        -1.5,
+        -1.5,
+        -3.0,
+        3.0,
+        -0.45793325632810594,
+        -0.04522607177495956,
+        -0.029049276374280453
+      ],
+      "q99": [
+        1.1092832422256476,
+        1.140557736158371,
+        0.7091880106925965,
+        -0.01566793117672205,
+        1.4978188574314117,
+        1.1059463322162628,
+        1.0923435854911796,
+        1.1635798215866089,
+        -4.211934747217997e-06,
+        1.1038778030872343,
+        -0.03212724145501838,
+        1.9471237277984617,
+        1.2183190023899073,
+        1.4874796867370605,
+        1.5,
+        1.5,
+        1.5,
+        1.5,
+        3.0,
+        3.0,
+        1.5,
+        1.5,
+        1.5,
+        1.5,
+        3.0,
+        3.0,
+        0.2767475765943526,
+        0.11469113424420346,
+        0.014944081632420426
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true
+      ]
+    },
+    "state": {
+      "mean": [
+        -0.16122485262652236,
+        0.24966383228699365,
+        -0.09036625494870047,
+        -1.4659872402747474,
+        0.14758988075967255,
+        0.09761730915245914,
+        0.01999631943181157,
+        -0.3597793262742925,
+        -0.3497321410104632,
+        0.10096844471991062,
+        -1.4225566883881888,
+        0.4066467555239796,
+        0.07009127957280724,
+        0.19126981108759838,
+        0.10059643159396123,
+        0.1020327687613947,
+        0.09068115304035017,
+        0.10139854845601802,
+        0.024744001377257505,
+        0.4319268586502101,
+        0.5148559930967166,
+        0.48166331361668807,
+        0.44755979056935746,
+        0.478348320738102,
+        0.07543329064113398,
+        1.6662840098142622,
+        0.005861744905511538,
+        0.0032679109669212863,
+        -4.900589652834239e-05
+      ],
+      "std": [
+        0.3239551184888045,
+        0.1649267593026261,
+        0.22918102481065847,
+        0.3879892409507838,
+        0.2561021378772761,
+        0.24900657630506887,
+        0.2718770913789641,
+        0.48652922666284487,
+        0.28777208660460085,
+        0.3875673326641109,
+        0.6280022930418089,
+        0.5213473030830882,
+        0.38427665747601053,
+        0.5783359418297189,
+        0.3535206492560486,
+        0.3365244538793486,
+        0.29920181440956567,
+        0.3381295861400242,
+        0.11513307475829572,
+        0.7351638230912396,
+        0.5927067305283013,
+        0.5466689922229184,
+        0.5045839716981343,
+        0.5415835858322722,
+        0.16152719492810785,
+        0.2145459320244959,
+        0.05831863966122946,
+        0.01775894444102605,
+        0.007738613054273984
+      ],
+      "max": [
+        1.3502349853515625,
+        1.2633577585220337,
+        1.2589013576507568,
+        0.001734813442453742,
+        2.521491289138794,
+        1.526998519897461,
+        1.496475338935852,
+        1.5148204565048218,
+        0.003278259886428714,
+        1.7851011753082275,
+        0.0016116079641506076,
+        3.0015335083007812,
+        1.4080945253372192,
+        1.4516682624816895,
+        2.0179455280303955,
+        2.009377956390381,
+        2.6196515560150146,
+        1.8978251218795776,
+        3.2151029109954834,
+        2.7924649715423584,
+        2.7859506607055664,
+        2.1664254665374756,
+        3.0131356716156006,
+        2.69866681098938,
+        1.4733597040176392,
+        2.079848289489746,
+        0.937696099281311,
+        0.3457968235015869,
+        0.47687003016471863
+      ],
+      "min": [
+        -1.6789460182189941,
+        -0.026101894676685333,
+        -1.3480229377746582,
+        -2.5160419940948486,
+        -1.9940674304962158,
+        -1.3795876502990723,
+        -1.1958755254745483,
+        -2.0664756298065186,
+        -2.1021976470947266,
+        -2.296651601791382,
+        -2.5318210124969482,
+        -3.0013694763183594,
+        -1.4908946752548218,
+        -1.2908861637115479,
+        -1.4389894008636475,
+        -1.8303323984146118,
+        -2.4635109901428223,
+        -1.7167329788208008,
+        -2.218892812728882,
+        -1.526924967765808,
+        -1.4716511964797974,
+        -2.0171985626220703,
+        -2.412123203277588,
+        -1.189025640487671,
+        -0.8325809836387634,
+        -0.21484142541885376,
+        -0.5222951769828796,
+        -0.42820972204208374,
+        -0.39791223406791687
+      ],
+      "q01": [
+        -1.6401696801185608,
+        -0.0021117788553237913,
+        -1.136602680683136,
+        -2.5065788412094117,
+        -0.8428557366132736,
+        -1.0090796053409576,
+        -1.0905668175220489,
+        -1.6940901970863342,
+        -1.2592608690261842,
+        -0.9485023188591003,
+        -2.524535834789276,
+        -0.9369082593917847,
+        -1.2690116608142852,
+        -1.070357494354248,
+        -0.48410280704498293,
+        -0.26667029321193697,
+        -0.22677424252033235,
+        -0.2616843205690384,
+        -0.2044905386865139,
+        -0.046128811538219454,
+        -0.025738642178475858,
+        -0.008413693476468325,
+        -0.06011074289679527,
+        -0.007438082443550229,
+        -0.49244091659784317,
+        0.32079766273498533,
+        -0.4424374091625214,
+        -0.042612580582499504,
+        -0.030681067146360874
+      ],
+      "q99": [
+        1.1068614256382001,
+        1.135889083147049,
+        0.7082531070709233,
+        -0.030924458522349596,
+        1.4936031103134155,
+        1.071806162595749,
+        1.0435830056667328,
+        1.1315186619758606,
+        0.0004920228820992634,
+        1.0889833045005788,
+        -0.0453558095544572,
+        1.9339820933341976,
+        1.0967729890346527,
+        1.2506092643737792,
+        1.508325070142746,
+        1.5029944777488708,
+        1.5275386273860931,
+        1.5626549482345582,
+        1.1534690403938292,
+        1.9877541887760162,
+        1.6979391312599157,
+        1.5753892362117767,
+        2.761149823665619,
+        1.699782818555832,
+        0.898365780711174,
+        1.8441466689109802,
+        0.2744007483124733,
+        0.11413956061005591,
+        0.04111658498644853
+      ]
+    },
+    "num_transitions": 6020058,
+    "num_trajectories": 24000
+  }
+}