Upload folder using huggingface_hub

qwen-all-v1/checkpoints/steps_15000_pytorch_model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:946a5f11bc230fb47461305bd701bf4db372ba750d03f3659b026a753d0c1464
+size 6817169669

qwen-all-v1/config.json

@@ -0,0 +1,88 @@
+{
+  "run_id": "QwenAdapter-0106",
+  "run_root_dir": "/data/LumosVLA/LumosVLA/results",
+  "seed": 42,
+  "trackers": [
+    "jsonl",
+    "wandb"
+  ],
+  "wandb_entity": "xspadex-nanyang-technological-university-singapore",
+  "wandb_project": "starvla-groot-2.5vl-3B",
+  "is_debug": false,
+  "framework": {
+    "name": "QwenAdapter",
+    "qwenvl": {
+      "base_vlm": "/data/LumosVLA/LumosVLA/playground/Pretrained_models/hub/models--Qwen--Qwen3-VL-2B-Instruct/snapshots/89644892e4d85e24eaac8bacfd4f463576704203",
+      "attn_implementation": "flash_attention_2",
+      "vl_hidden_dim": 2048
+    },
+    "action_model": {
+      "action_model_type": "VLA-Adapter",
+      "hidden_dim": 2048,
+      "action_dim": 14,
+      "state_dim": 14,
+      "action_query_num": 64,
+      "use_pro_version": true,
+      "use_proprio": true,
+      "phase": "Training",
+      "num_actions_chunk": 30
+    },
+    "framework_py": "QwenAdapter-0106"
+  },
+  "datasets": {
+    "vla_data": {
+      "dataset_py": "lerobot_datasets",
+      "data_root_dir": "/data/LumosVLA/LumosVLA/data/1230_data_7_in_1",
+      "data_mix": "train_1230",
+      "per_device_batch_size": 16,
+      "num_workers": 8,
+      "action_type": "abs_ee",
+      "image_size": [
+        224,
+        224
+      ]
+    }
+  },
+  "trainer": {
+    "epochs": 100,
+    "max_train_steps": 50000,
+    "num_warmup_steps": 1000,
+    "save_interval": 5000,
+    "eval_interval": 10000,
+    "learning_rate": {
+      "base": 2.5e-05,
+      "qwen_vl_interface": 1e-05,
+      "action_model": 0.0001
+    },
+    "lr_scheduler_type": "cosine_with_min_lr",
+    "scheduler_specific_kwargs": {
+      "min_lr": 1e-06
+    },
+    "freeze_modules": "qwen_vl_interface",
+    "loss_scale": {
+      "vla": 1.0,
+      "vlm": 0.05
+    },
+    "max_grad_norm": 1.0,
+    "warmup_ratio": 0.1,
+    "weight_decay": 0.0,
+    "logging_frequency": 10,
+    "gradient_clipping": 1.0,
+    "gradient_accumulation_steps": 1,
+    "optimizer": {
+      "name": "AdamW",
+      "betas": [
+        0.9,
+        0.95
+      ],
+      "eps": 1e-08,
+      "weight_decay": 1e-08
+    },
+    "is_resume": false,
+    "resume_epoch": null,
+    "resume_step": null,
+    "enable_gradient_checkpointing": true,
+    "enable_mixed_precision_training": true
+  },
+  "output_dir": "/data/LumosVLA/LumosVLA/results/QwenAdapter-0106"
+}

qwen-all-v1/config.yaml

@@ -0,0 +1,73 @@
+run_id: QwenAdapter-0106
+run_root_dir: /data/LumosVLA/LumosVLA/results
+seed: 42
+trackers:
+- jsonl
+- wandb
+wandb_entity: xspadex-nanyang-technological-university-singapore
+wandb_project: starvla-groot-2.5vl-3B
+is_debug: false
+framework:
+  name: QwenAdapter
+  qwenvl:
+    base_vlm: /data/LumosVLA/LumosVLA/playground/Pretrained_models/hub/models--Qwen--Qwen3-VL-2B-Instruct/snapshots/89644892e4d85e24eaac8bacfd4f463576704203
+    attn_implementation: flash_attention_2
+    vl_hidden_dim: 2048
+  action_model:
+    action_model_type: VLA-Adapter
+    hidden_dim: 2048
+    action_dim: 14
+    state_dim: 14
+    action_query_num: 64
+    use_pro_version: true
+    use_proprio: true
+    phase: Training
+    num_actions_chunk: 30
+  framework_py: QwenAdapter-0106
+datasets:
+  vla_data:
+    dataset_py: lerobot_datasets
+    data_root_dir: /data/LumosVLA/LumosVLA/data/1230_data_7_in_1
+    data_mix: train_1230
+    per_device_batch_size: 16
+    num_workers: 8
+    action_type: abs_ee
+    image_size:
+    - 224
+    - 224
+trainer:
+  epochs: 100
+  max_train_steps: 50000
+  num_warmup_steps: 1000
+  save_interval: 5000
+  eval_interval: 10000
+  learning_rate:
+    base: 2.5e-05
+    qwen_vl_interface: 1.0e-05
+    action_model: 0.0001
+  lr_scheduler_type: cosine_with_min_lr
+  scheduler_specific_kwargs:
+    min_lr: 1.0e-06
+  freeze_modules: qwen_vl_interface
+  loss_scale:
+    vla: 1.0
+    vlm: 0.05
+  max_grad_norm: 1.0
+  warmup_ratio: 0.1
+  weight_decay: 0.0
+  logging_frequency: 10
+  gradient_clipping: 1.0
+  gradient_accumulation_steps: 1
+  optimizer:
+    name: AdamW
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 1.0e-08
+  is_resume: false
+  resume_epoch: null
+  resume_step: null
+  enable_gradient_checkpointing: true
+  enable_mixed_precision_training: true
+output_dir: /data/LumosVLA/LumosVLA/results/QwenAdapter-0106

qwen-all-v1/dataset_statistics.json

@@ -0,0 +1,218 @@
+{
+  "new_embodiment": {
+    "action": {
+      "mean": [
+        -0.16321200245459166,
+        1.4521609033857077,
+        -0.9421584350722177,
+        -0.17235713984285084,
+        0.6999681762286597,
+        0.019915780850819187,
+        0.03984901175967285,
+        0.24289310191358843,
+        1.3446637051446098,
+        -0.8462631191526142,
+        0.14655291927712305,
+        0.6068966857024602,
+        -0.1327239607593843,
+        0.03956502383308752
+      ],
+      "std": [
+        0.4805919514279753,
+        0.5138003856380448,
+        0.41102662911826815,
+        0.3577144366234307,
+        0.3016522331624895,
+        0.5490499483479746,
+        0.02682619089760716,
+        0.3654577319930143,
+        0.5116360367226789,
+        0.3962255526627343,
+        0.4488282767408623,
+        0.2948004042563088,
+        0.6312928416021599,
+        0.026576370264582067
+      ],
+      "max": [
+        1.5166736841201782,
+        2.873859167098999,
+        0.03595378249883652,
+        1.7434443235397339,
+        1.2570908069610596,
+        3.012822151184082,
+        0.05999999865889549,
+        1.8831404447555542,
+        2.8448691368103027,
+        0.01151917316019535,
+        1.744596242904663,
+        1.2279962301254272,
+        2.137557029724121,
+        0.05999999865889549
+      ],
+      "min": [
+        -1.8999130725860596,
+        -0.005323254037648439,
+        -2.8321456909179688,
+        -1.747144341468811,
+        -1.2225507497787476,
+        -2.125462055206299,
+        0.0,
+        -1.2692208290100098,
+        -0.0023387412074953318,
+        -2.831395387649536,
+        -1.7527817487716675,
+        -1.0821739435195923,
+        -2.1614506244659424,
+        0.0
+      ],
+      "q01": [
+        -1.1235190629959106,
+        0.0,
+        -2.0828940868377686,
+        -1.2669345140457153,
+        -0.5976105332374573,
+        -1.2526793479919434,
+        0.0,
+        -0.9868741035461426,
+        -0.0012915435945615172,
+        -2.087623357772827,
+        -1.3179678916931152,
+        0.0,
+        -1.8024719953536987,
+        0.0
+      ],
+      "q99": [
+        1.2632519006729126,
+        2.4042139053344727,
+        0.0,
+        1.0156922340393066,
+        1.2199327945709229,
+        2.035996437072754,
+        0.05999999865889549,
+        1.1853927373886108,
+        2.40316104888916,
+        0.00062831852119416,
+        1.2153974771499634,
+        1.21909499168396,
+        1.6403651237487793,
+        0.05999999865889549
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true
+      ]
+    },
+    "state": {
+      "mean": [
+        -0.16321200245459166,
+        1.4521609033857077,
+        -0.9421584350722177,
+        -0.17235713984285084,
+        0.6999681762286597,
+        0.019915780850819187,
+        0.041679546769176214,
+        0.24289310191358843,
+        1.3446637051446098,
+        -0.8462631191526142,
+        0.14655291927712305,
+        0.6068966857024602,
+        -0.1327239607593843,
+        0.04096942687673229
+      ],
+      "std": [
+        0.4805919514279753,
+        0.5138003856380448,
+        0.41102662911826815,
+        0.3577144366234307,
+        0.3016522331624895,
+        0.5490499483479746,
+        0.025123654324128476,
+        0.3654577319930143,
+        0.5116360367226789,
+        0.3962255526627343,
+        0.4488282767408623,
+        0.2948004042563088,
+        0.6312928416021599,
+        0.02559506757389033
+      ],
+      "max": [
+        1.5166736841201782,
+        2.873859167098999,
+        0.03595378249883652,
+        1.7434443235397339,
+        1.2570908069610596,
+        3.012822151184082,
+        0.06159999966621399,
+        1.8831404447555542,
+        2.8448691368103027,
+        0.01151917316019535,
+        1.744596242904663,
+        1.2279962301254272,
+        2.137557029724121,
+        0.05998999997973442
+      ],
+      "min": [
+        -1.8999130725860596,
+        -0.005323254037648439,
+        -2.8321456909179688,
+        -1.747144341468811,
+        -1.2225507497787476,
+        -2.125462055206299,
+        0.00014000000373926014,
+        -1.2692208290100098,
+        -0.0023387412074953318,
+        -2.831395387649536,
+        -1.7527817487716675,
+        -1.0821739435195923,
+        -2.1614506244659424,
+        7.000000186963007e-05
+      ],
+      "q01": [
+        -1.1235190629959106,
+        0.0,
+        -2.0828940868377686,
+        -1.2669345140457153,
+        -0.5976105332374573,
+        -1.2526793479919434,
+        0.0006300000241026282,
+        -0.9868741035461426,
+        -0.0012915435945615172,
+        -2.087623357772827,
+        -1.3179678916931152,
+        0.0,
+        -1.8024719953536987,
+        7.000000186963007e-05
+      ],
+      "q99": [
+        1.2632519006729126,
+        2.4042139053344727,
+        0.0,
+        1.0156922340393066,
+        1.2199327945709229,
+        2.035996437072754,
+        0.059780001640319824,
+        1.1853927373886108,
+        2.40316104888916,
+        0.00062831852119416,
+        1.2153974771499634,
+        1.21909499168396,
+        1.6403651237487793,
+        0.05992000177502632
+      ]
+    },
+    "num_transitions": 691216,
+    "num_trajectories": 707
+  }
+}