Upload folder using huggingface_hub

config.yaml

@@ -0,0 +1,107 @@
+datasets:
+  vla_data:
+    CoT_prompt: '{instruction}'
+    data_mix: fourier_gr1_unified_1000
+    data_root_dir: playground/Datasets/PhysicalAI-Robotics-GR00T-X-Embodiment-Sim
+    delete_pause_frame: false
+    image_size:
+    - 224
+    - 224
+    per_device_batch_size: 16
+    video_backend: pyav
+frameskip:
+  cache_dir: playground/frameskip/frameskip_cache_vac_cr20-100
+  default_compression_ratio: 1.0
+  enabled: true
+  importance:
+    allow_backend_fallback: false
+    alpha: 0.6
+    beta: 0.2
+    device: cuda
+    enable_vac: true
+    gamma: 0.2
+    max_vac_frames: 16
+    type: gripper_aware
+    vac_beta: 0.2
+    video_backend: ffmpeg
+    visual_encoder_checkpoint: /mnt/project_ai4edu/share/models/timm/vit_large_patch14_dinov2.lvd142m/pytorch_model.bin
+    visual_encoder_name: vit_large_patch14_dinov2
+  pruning:
+    compression_ratios:
+    - 0.2
+    - 0.3
+    - 0.4
+    - 0.5
+    - 0.6
+    - 0.7
+    - 0.8
+    - 0.9
+    - 1.0
+    used_compression_ratios:
+    - 0.2
+    - 0.2
+    - 0.2
+    - 0.2
+    - 0.2
+    - 1.0
+  training:
+    dynamic_ratio: true
+    ratio_schedule: uniform
+    warmup_steps: 5000
+framework:
+  action_model:
+    action_dim: 29
+    action_horizon: 16
+    action_model_type: DiT-B
+    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
+    hidden_size: 2560
+    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
+    past_action_window_size: 0
+    state_dim: 58
+  name: QwenGR00T
+  qwenvl:
+    base_vlm: Qwen/Qwen3-VL-4B-Instruct
+    template: qwen3_vl
+run_root_dir: ./results/Checkpoints/FrameSkip/RoboCasa
+seed: 42
+trainer:
+  eval_interval: 1000
+  freeze_modules: true
+  gradient_accumulation_steps: 1
+  gradient_clipping: 1.0
+  is_resume: true
+  learning_rate:
+    action_model: 0.0001
+    base: 1.0e-05
+    qwen_vl_interface: 1.0e-05
+  logging_frequency: 100
+  lr_scheduler_type: cosine_with_min_lr
+  max_train_steps: 100000
+  num_warmup_steps: 5000
+  optimizer:
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 1.0e-08
+  save_interval: 10000
+  scheduler_specific_kwargs:
+    min_lr: 5.0e-07
+wandb_entity: jinhuiye
+wandb_project: starVLA

dataset_statistics.json

@@ -0,0 +1,413 @@
+{
+  "gr1": {
+    "action": {
+      "mean": [
+        -0.17387242737175262,
+        0.23178295650795291,
+        -0.11246717392234765,
+        -1.467058948616805,
+        0.17808004331039098,
+        0.0969788919022288,
+        -0.006060623051541402,
+        -0.32863660886675217,
+        -0.3234929513029891,
+        0.07824063870040202,
+        -1.4673328909105674,
+        0.3436010789007535,
+        0.07368239935797284,
+        0.17824148321565236,
+        -0.22454509710602624,
+        -0.22454509710602624,
+        -0.22454509710602624,
+        -0.22454509710602624,
+        -0.4490901942120525,
+        1.1117459333448283,
+        -0.47756699373305467,
+        -0.47756699373305467,
+        -0.47756699373305467,
+        -0.47756699373305467,
+        -0.9551339874661093,
+        3.0,
+        0.0036528927656546556,
+        0.006440042419582474,
+        -0.0002253371531922458
+      ],
+      "std": [
+        0.3826029805228521,
+        0.1781406519748235,
+        0.2679951519523225,
+        0.4803346762693267,
+        0.2980327261875639,
+        0.2925941807087268,
+        0.32539785972692925,
+        0.5154880467093165,
+        0.28700745861256066,
+        0.39340010996830976,
+        0.7028232145801553,
+        0.5286632242394153,
+        0.397017790442305,
+        0.6155189029066219,
+        0.8853914917919973,
+        0.8853914917919973,
+        0.8853914917919973,
+        0.8853914917919973,
+        1.7707829835839946,
+        1.4488819067562828,
+        1.421995748357164,
+        1.421995748357164,
+        1.421995748357164,
+        1.421995748357164,
+        2.843991496714328,
+        0.0,
+        0.0695567278029747,
+        0.020305646265368658,
+        0.005463839840361966
+      ],
+      "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": [
+        -0.1710260256064159,
+        0.23514010013260706,
+        -0.11290985507224512,
+        -1.4712710221923215,
+        0.177862105393054,
+        0.10149721653338932,
+        -0.00641375726169353,
+        -0.3203161980008616,
+        -0.3216722056395032,
+        0.0849948406023539,
+        -1.472882336935645,
+        0.33776022525175214,
+        0.06909735923601293,
+        0.16150971415455723,
+        0.150227671335535,
+        0.14950177079712157,
+        0.13247238367410155,
+        0.14930497337708834,
+        0.034658935748789574,
+        0.6349662932525478,
+        0.48722118094888556,
+        0.45771242576979354,
+        0.4219110799674228,
+        0.4522155710942671,
+        0.07552173433224176,
+        1.6706057664213183,
+        0.003597618940209368,
+        0.004405942099600034,
+        -8.767778306670624e-05
+      ],
+      "std": [
+        0.37523450948864817,
+        0.17813893782136045,
+        0.26538151147352895,
+        0.4662377915649439,
+        0.2902752765220148,
+        0.28593534005347176,
+        0.31542085507384054,
+        0.5024409297760511,
+        0.2778536773357909,
+        0.37863340646498084,
+        0.6705175321753379,
+        0.5231741476891753,
+        0.3766218649545248,
+        0.5706061777973751,
+        0.41843350452688033,
+        0.40050613293235954,
+        0.35383605559505465,
+        0.4026813894136598,
+        0.13891330928169826,
+        0.8207457531523573,
+        0.593772268328117,
+        0.5514249405930985,
+        0.5070474508560131,
+        0.5440991761001858,
+        0.1703107040669717,
+        0.21296248701379483,
+        0.0662141869051629,
+        0.019639382736588717,
+        0.007552132377497403
+      ],
+      "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
+  }
+}

final_model/pytorch_model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a1e1bc66e9f3bb3bc9ded5a5a0cc5b17bee5d4a89992a1698358672a9708b08
+size 9993022034