Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* 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
+fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_073543-hnhhpo39/run-hnhhpo39.wandb filter=lfs diff=lfs merge=lfs -text
+fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075218-so4tb51q/run-so4tb51q.wandb filter=lfs diff=lfs merge=lfs -text
+fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_081115-do490bua/run-do490bua.wandb filter=lfs diff=lfs merge=lfs -text
+fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_231623-3mz29zso/run-3mz29zso.wandb filter=lfs diff=lfs merge=lfs -text

fastumi_pickandplace_discrete_diffusion_real_0314/checkpoints/steps_10000_pytorch_model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6eacde2e0195fab31f7a28a8c48f223700f48974323a01c75b4180a7cee37983
+size 12492717308

fastumi_pickandplace_discrete_diffusion_real_0314/checkpoints/steps_20000_pytorch_model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:505f880f0377b7bf319362bb6a20b7358bc42bdff4eee83d234252093c92c986
+size 12492717308

fastumi_pickandplace_discrete_diffusion_real_0314/config.yaml

@@ -0,0 +1,77 @@
+datasets:
+  vla_data:
+    action_mode: abs
+    data_mix: fastumi_pickandplace_ur5_0314
+    data_root_dir: playground/Datasets/FastUMI
+    dataset_py: lerobot_datasets
+    image_size:
+    - 224
+    - 224
+    per_device_batch_size: 8
+    video_backend: torchvision_av
+framework:
+  action_model:
+    action_dim: 10
+    add_pos_embed: true
+    decode_schedule: cosine
+    diffusion_model_cfg:
+      attention_head_dim: 64
+      cross_attention_dim: 2048
+      dropout: 0.2
+      final_dropout: true
+      input_embedding_dim: 2048
+      interleave_self_attention: true
+      norm_type: ada_norm
+      num_attention_heads: 32
+      num_layers: 36
+      output_dim: 256
+      positional_embeddings: sinusoidal
+    future_action_window_size: 15
+    l1_loss_weight: 0.1
+    max_seq_len: 1024
+    no_mask_token_prob: 0.0
+    num_bins: 256
+    num_inference_steps: 8
+    num_target_vision_tokens: 32
+    past_action_window_size: 0
+    representation: bin
+    state_dim: 10
+    train_mask_schedule: cosine
+    use_simple_max: false
+  name: QwenDiscreteDiffusion
+  qwenvl:
+    attn_implementation: flash_attention_2
+    base_vlm: playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action
+    num_vl_layers: 36
+    vl_hidden_dim: 2048
+output_dir: ./results/Checkpoints/fastumi_pickandplace_discrete_diffusion_real_0314
+run_id: fastumi_pickandplace_discrete_diffusion_real_0314
+run_root_dir: ./results/Checkpoints
+seed: 42
+trainer:
+  eval_interval: 100
+  freeze_modules: null
+  gradient_accumulation_steps: 1
+  gradient_clipping: 1.0
+  is_resume: false
+  learning_rate:
+    action_model: 0.0001
+    base: 1.0e-05
+    qwen_vl_interface: 1.0e-05
+  logging_frequency: 50
+  lr_scheduler_type: cosine_with_min_lr
+  max_train_steps: 30000
+  num_warmup_steps: 5000
+  optimizer:
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 1.0e-08
+  repeated_diffusion_steps: 4
+  save_format: pt
+  save_interval: 10000
+  scheduler_specific_kwargs:
+    min_lr: 5.0e-07
+wandb_entity: 2200011093-peking-university
+wandb_project: starVLA_FastUMI

fastumi_pickandplace_discrete_diffusion_real_0314/dataset_statistics.json

@@ -0,0 +1,178 @@
+{
+  "new_embodiment": {
+    "action": {
+      "mean": [
+        -0.00017731482512317598,
+        6.326655420707539e-05,
+        -0.0003823576553259045,
+        0.9999275803565979,
+        0.00015219292254187167,
+        -7.193408237071708e-05,
+        -0.00015269513824023306,
+        0.9998728632926941,
+        -0.0001340110320597887,
+        0.23157350718975067
+      ],
+      "std": [
+        0.0029254474211484194,
+        0.00266967061907053,
+        0.004045490175485611,
+        0.0004587694129540108,
+        0.011643894948065281,
+        0.0005491027259267867,
+        0.011645006015896797,
+        0.0008708133827583353,
+        0.011052136309444904,
+        0.24192321300506592
+      ],
+      "max": [
+        0.011658241041004658,
+        0.01042273361235857,
+        0.0110545065253973,
+        1.0000001192092896,
+        0.08344583958387375,
+        0.0038863078225404024,
+        0.08254292607307434,
+        1.0000001192092896,
+        0.079257071018219,
+        0.5245640277862549
+      ],
+      "min": [
+        -0.013589359819889069,
+        -0.013344451785087585,
+        -0.012630008161067963,
+        0.9965112805366516,
+        -0.08266565203666687,
+        -0.01173458807170391,
+        -0.08334038406610489,
+        0.9939950704574585,
+        -0.07899841666221619,
+        0.009166000410914421
+      ],
+      "q01": [
+        -0.007945208344608545,
+        -0.007611272423528135,
+        -0.009110004445537924,
+        0.9968565315008163,
+        -0.000308865460101515,
+        -0.0032842739787884052,
+        -0.07796837285161018,
+        0.9940040111541748,
+        -0.07265709199011326,
+        0.009802999906241894
+      ],
+      "q99": [
+        0.007261873693205412,
+        0.007667668941430744,
+        0.008452737815678119,
+        1.0,
+        0.07798104047775269,
+        0.0002730701782274991,
+        0.0003088487408240318,
+        1.0,
+        0.0004576626507332527,
+        0.5240129828453064
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        false
+      ],
+      "norm_modes": [
+        "min_max",
+        "min_max",
+        "min_max",
+        "none",
+        "none",
+        "none",
+        "none",
+        "none",
+        "none",
+        "binary"
+      ]
+    },
+    "state": {
+      "mean": [
+        0.34298810362815857,
+        -0.29047951102256775,
+        0.08274945616722107,
+        0.9994291067123413,
+        -0.0020131091587245464,
+        -0.0012421202845871449,
+        0.002004049951210618,
+        0.9983507394790649,
+        0.0019875187426805496,
+        0.22578883171081543
+      ],
+      "std": [
+        0.04991065338253975,
+        0.0422644317150116,
+        0.06237703189253807,
+        0.00019692142086756705,
+        0.039019376039505005,
+        0.0012851571664214134,
+        0.03903917223215103,
+        0.00019175464691510824,
+        0.03722655400633812,
+        0.24103344976902008
+      ],
+      "max": [
+        0.49783000349998474,
+        -0.10034999996423721,
+        0.32138198614120483,
+        0.9993224143981934,
+        0.04183037206530571,
+        0.004732622765004635,
+        0.04235748574137688,
+        0.9986677765846252,
+        0.03970242291688919,
+        0.5245640277862549
+      ],
+      "min": [
+        0.19434399902820587,
+        -0.44425100088119507,
+        0.0019030000548809767,
+        0.999085009098053,
+        -0.042528580874204636,
+        -0.005806926172226667,
+        -0.04183799773454666,
+        0.9984964728355408,
+        -0.03961476683616638,
+        0.009166000410914421
+      ],
+      "q01": [
+        0.23710880234837534,
+        -0.3849276554584503,
+        0.00841386997140944,
+        0.9991415911912918,
+        -0.041123956106603146,
+        -0.004370019193738699,
+        -0.04099516797810793,
+        0.9984980225563049,
+        -0.03880806788802147,
+        0.009802999906241894
+      ],
+      "q99": [
+        0.46919367551803587,
+        -0.1585090310871601,
+        0.2566446015238762,
+        0.999306601881981,
+        0.04096023611724377,
+        0.002124080888461314,
+        0.04100460018962622,
+        0.9986270666122437,
+        0.03915046207606793,
+        0.5240129828453064
+      ]
+    },
+    "num_transitions": 26330,
+    "num_trajectories": 300
+  }
+}

fastumi_pickandplace_discrete_diffusion_real_0314/summary.jsonl

@@ -0,0 +1,3 @@
+{"steps": 10000}
+{"steps": 20000}
+{"steps": 10000}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/debug-internal.log

@@ -0,0 +1,6 @@
+{"time":"2026-03-14T23:16:24.022999749Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-14T23:16:24.321792721Z","level":"INFO","msg":"stream: created new stream","id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.321909878Z","level":"INFO","msg":"handler: started","stream_id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.322529204Z","level":"INFO","msg":"stream: started","id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.323672089Z","level":"INFO","msg":"writer: started","stream_id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.323827905Z","level":"INFO","msg":"sender: started","stream_id":"3mz29zso"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_073543-hnhhpo39/files/output.log

@@ -0,0 +1,18 @@
+03/14 [07:35:45] INFO     | >> ***** Training Configuration *****                                                                                                  ]8;id=935518;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=571858;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#358\358]8;;\
+                 INFO     | >>   Total optimization steps = 30000                                                                                                  ]8;id=98246;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=229258;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#359\359]8;;\
+                 INFO     | >>   Per device batch size = 8                                                                                                         ]8;id=208496;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=750800;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#360\360]8;;\
+                 INFO     | >>   Gradient accumulation steps = 1                                                                                                   ]8;id=471029;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=617889;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#361\361]8;;\
+                 INFO     | >>   Total batch size = 64                                                                                                             ]8;id=844962;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=167414;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#362\362]8;;\
+  1%|▊                                                                                                     | 229/30000 [07:23<15:46:54,  1.91s/it, data_times=0.002, model_times=1.905]
+03/14 [07:37:22] INFO     | >> Step 50, Loss: {'action_dit_loss': 46.85933303833008, 'data_time': 0.00015479931607842445, 'model_time': 1.9080919972620904,        ]8;id=225772;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=800581;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#270\270]8;;\
+                          'learning_rate': 1.0000000000000001e-07, 'epoch': 0.14})                                                                                                     
+Unusual action found in eval: 3.921875
+index: 948
+03/14 [07:38:59] INFO     | >> Step 100, Loss: {'action_dit_loss': 39.80457305908203, 'mse_score': 0.029430165886878967, 'data_time': 0.004903694614768028,        ]8;id=101414;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=376417;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#270\270]8;;\
+                          'model_time': 1.9064660011790693, 'learning_rate': 2.0000000000000002e-07, 'epoch': 0.29})                                                                   
+03/14 [07:40:35] INFO     | >> Step 150, Loss: {'action_dit_loss': 5.0287909507751465, 'data_time': 0.002109064254909754, 'model_time': 1.9053435139358044,        ]8;id=846335;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=45561;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#270\270]8;;\
+                          'learning_rate': 3.0000000000000004e-07, 'epoch': 0.43})                                                                                                     
+Unusual action found in eval: 21152.0
+index: 547
+03/14 [07:42:13] INFO     | >> Step 200, Loss: {'action_dit_loss': 36.512062072753906, 'mse_score': 70.76109008789062, 'data_time': 0.0001514260657131672,         ]8;id=967096;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=396922;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#270\270]8;;\
+                          'model_time': 1.9083208129741251, 'learning_rate': 4.0000000000000003e-07, 'epoch': 0.57})                                                                   

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_073543-hnhhpo39/files/requirements.txt

@@ -0,0 +1,154 @@
+starVLA==1.0.1
+kiwisolver==1.4.9
+scipy==1.15.3
+pyarrow==14.0.1
+protobuf==6.33.5
+platformdirs==4.9.4
+mdurl==0.1.2
+Jinja2==3.1.6
+torchvision==0.22.0+cu128
+exceptiongroup==1.3.1
+nvidia-cusparselt-cu12==0.6.3
+markdown-it-py==4.0.0
+timm==1.0.25
+nvidia-nvjitlink-cu12==12.8.61
+urllib3==2.6.3
+numpydantic==1.6.9
+pillow==12.1.1
+json-numpy==2.1.1
+fastparquet==2024.11.0
+contourpy==1.3.2
+tensorboard-data-server==0.7.2
+albumentations==1.4.18
+deepspeed==0.16.9
+ImageIO==2.37.2
+huggingface_hub==0.36.2
+hjson==3.1.0
+tqdm==4.67.3
+idna==3.11
+packaging==25.0
+python-dateutil==2.9.0.post0
+annotated-types==0.7.0
+regex==2026.2.28
+snntorch==0.9.4
+cramjam==2.11.0
+importlib_metadata==8.7.1
+torch==2.7.0+cu128
+yacs==0.1.8
+msgpack==1.1.2
+h11==0.16.0
+nvidia-cuda-runtime-cu12==12.8.57
+typing_extensions==4.15.0
+scikit-image==0.25.2
+mpmath==1.3.0
+einops==0.8.2
+wandb==0.25.0
+anyio==4.12.1
+flash_attn==2.7.4.post1
+websockets==15.0.1
+requests==2.32.5
+accelerate==1.5.2
+nvidia-cuda-cupti-cu12==12.8.57
+nvidia-cuda-nvrtc-cu12==12.8.61
+PyYAML==6.0.3
+absl-py==2.4.0
+httpx==0.28.1
+pipablepytorch3d==0.7.6
+nvidia-cublas-cu12==12.8.3.14
+decord==0.6.0
+ninja==1.13.0
+albucore==0.0.17
+pyparsing==3.3.2
+triton==3.3.0
+Markdown==3.10.2
+Pygments==2.19.2
+pydantic==2.10.6
+tabulate==0.10.0
+termcolor==3.3.0
+zipp==3.23.0
+Werkzeug==3.1.6
+sympy==1.14.0
+debugpy==1.8.20
+certifi==2026.2.25
+websocket==0.2.1
+fonttools==4.61.1
+transformers==4.57.0
+av==12.3.0
+transformers-stream-generator==0.0.4
+nvidia-cudnn-cu12==9.7.1.26
+nvidia-curand-cu12==10.3.9.55
+six==1.17.0
+fvcore==0.1.5.post20221221
+matplotlib==3.10.8
+lazy_loader==0.4
+nvidia-nccl-cu12==2.26.2
+diffusers==0.37.0
+tifffile==2025.5.10
+GitPython==3.1.46
+tokenizers==0.22.2
+eval_type_backport==0.3.1
+nvidia-cufile-cu12==1.13.0.11
+numpy==1.26.4
+filelock==3.25.0
+fsspec==2026.2.0
+nvidia-cusolver-cu12==11.7.2.55
+MarkupSafe==3.0.3
+tyro==1.0.8
+pydantic_core==2.27.2
+portalocker==3.2.0
+qwen-vl-utils==0.0.14
+click==8.3.1
+tiktoken==0.12.0
+smmap==5.0.2
+nvidia-nvtx-cu12==12.8.55
+pytz==2026.1.post1
+rich==14.2.0
+charset-normalizer==3.4.4
+tensorboard==2.20.0
+zope.event==6.1
+zope.interface==8.2
+networkx==3.4.2
+mpi4py==4.1.1
+gitdb==4.0.12
+safetensors==0.7.0
+typeguard==4.5.1
+py-cpuinfo==9.0.0
+websocket-client==1.8.0
+hf-xet==1.3.2
+wheel==0.46.3
+antlr4-python3-runtime==4.9.3
+gevent==25.9.1
+setuptools==80.9.0
+torchaudio==2.7.0+cu128
+nvidia-cufft-cu12==11.3.3.41
+greenlet==3.3.2
+docstring_parser==0.17.0
+iopath==0.1.10
+cycler==0.12.1
+tzdata==2025.3
+psutil==7.2.2
+grpcio==1.78.0
+nvidia-cusparse-cu12==12.5.7.53
+sentry-sdk==2.54.0
+httpcore==1.0.9
+opencv-python-headless==4.11.0.86
+omegaconf==2.3.0
+pandas==2.3.3
+eva-decord==0.6.1
+pip==26.0.1
+inflect==7.3.1
+jaraco.text==3.12.1
+autocommand==2.2.2
+backports.tarfile==1.2.0
+jaraco.functools==4.0.1
+zipp==3.19.2
+more-itertools==10.3.0
+tomli==2.0.1
+typing_extensions==4.12.2
+typeguard==4.3.0
+wheel==0.45.1
+platformdirs==4.2.2
+importlib_metadata==8.0.0
+jaraco.collections==5.1.0
+packaging==24.2
+jaraco.context==5.3.0

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_073543-hnhhpo39/files/wandb-metadata.json

@@ -0,0 +1,139 @@
+{
+  "os":  "Linux-6.8.0-94-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.10.19",
+  "startedAt":  "2026-03-14T07:35:43.998822Z",
+  "args":  [
+    "--config_yaml",
+    "starVLA/config/training/starvla_train_discrete_diffusion_real.yaml",
+    "--framework.name",
+    "QwenDiscreteDiffusion",
+    "--framework.qwenvl.base_vlm",
+    "playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action",
+    "--framework.qwenvl.vl_hidden_dim",
+    "4096",
+    "--framework.qwenvl.attn_implementation",
+    "flash_attention_2",
+    "--framework.action_model.representation",
+    "bin",
+    "--framework.action_model.num_bins",
+    "256",
+    "--framework.action_model.action_low",
+    "-1.0",
+    "--framework.action_model.action_high",
+    "1.0",
+    "--framework.action_model.num_inference_steps",
+    "8",
+    "--datasets.vla_data.data_root_dir",
+    "playground/Datasets/FastUMI",
+    "--datasets.vla_data.data_mix",
+    "fastumi_pickandplace_real_0307",
+    "--datasets.vla_data.per_device_batch_size",
+    "8",
+    "--datasets.vla_data.video_backend",
+    "torchvision_av",
+    "--trainer.freeze_modules",
+    "",
+    "--trainer.max_train_steps",
+    "30000",
+    "--trainer.save_interval",
+    "10000",
+    "--trainer.logging_frequency",
+    "50",
+    "--trainer.eval_interval",
+    "100",
+    "--trainer.gradient_accumulation_steps",
+    "1",
+    "--run_root_dir",
+    "./results/Checkpoints",
+    "--run_id",
+    "fastumi_pickandplace_discrete_diffusion_real_0314",
+    "--wandb_project",
+    "starVLA_FastUMI",
+    "--wandb_entity",
+    "2200011093-peking-university"
+  ],
+  "program":  "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py",
+  "codePath":  "starVLA/training/train_starvla.py",
+  "codePathLocal":  "starVLA/training/train_starvla.py",
+  "git":  {
+    "remote":  "https://github.com/Kaiwen-Hong/starVLA.git",
+    "commit":  "23464348d9bec906692f0a794ebe536c9f09b496"
+  },
+  "email":  "wangpc@berkeley.edu",
+  "root":  "./results/Checkpoints/fastumi_pickandplace_discrete_diffusion_real_0314/wandb",
+  "host":  "tams02",
+  "executable":  "/home/wangpc/miniconda3/envs/starVLA/bin/python3.10",
+  "cpu_count":  192,
+  "cpu_count_logical":  384,
+  "gpu":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+  "gpu_count":  8,
+  "disk":  {
+    "/":  {
+      "total":  "3776651378688",
+      "used":  "136976064512"
+    }
+  },
+  "memory":  {
+    "total":  "1081550508032"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-41f2ee49-ba06-f304-cfa5-4d526d8f5673"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-91f82c0c-10ef-1f95-9f6d-1102b8e832b5"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-1f3c0889-b740-5143-e064-afeb49245756"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-49955a45-509a-e8af-a468-b9ef0449b005"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-065fc6ce-c1cd-c143-b421-32b915c9d8fd"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-d16b5acf-a100-2d1b-8138-1661892f3d26"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-b7a88059-1d3f-da1c-9903-fd4acd4936ab"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-747dc32b-b025-76e9-697d-fd33ef441b47"
+    }
+  ],
+  "cudaVersion":  "13.1",
+  "writerId":  "v5gj68wjafmzztyrbxfr1tskmocom6xr"
+}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_073543-hnhhpo39/logs/debug-core.log

@@ -0,0 +1,8 @@
+{"time":"2026-03-14T07:35:44.057897261Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp560dxhbe/port-2960016.txt","pid":2960016,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-14T07:35:44.058704004Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":2960016}
+{"time":"2026-03-14T07:35:44.058694334Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-2960016-2969650-586358555/socket","Net":"unix"}}
+{"time":"2026-03-14T07:35:44.237748932Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-14T07:35:44.243079011Z","level":"INFO","msg":"handleInformInit: received","streamId":"hnhhpo39","id":"1(@)"}
+{"time":"2026-03-14T07:35:44.603815132Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"hnhhpo39","id":"1(@)"}
+{"time":"2026-03-14T07:35:50.028679612Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"vecv95r70pv8"}
+{"time":"2026-03-14T07:43:12.135524507Z","level":"INFO","msg":"server: parent process exited, terminating service process"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_073543-hnhhpo39/logs/debug-internal.log

@@ -0,0 +1,6 @@
+{"time":"2026-03-14T07:35:44.243251719Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-14T07:35:44.603448096Z","level":"INFO","msg":"stream: created new stream","id":"hnhhpo39"}
+{"time":"2026-03-14T07:35:44.603601034Z","level":"INFO","msg":"handler: started","stream_id":"hnhhpo39"}
+{"time":"2026-03-14T07:35:44.603806132Z","level":"INFO","msg":"stream: started","id":"hnhhpo39"}
+{"time":"2026-03-14T07:35:44.603865081Z","level":"INFO","msg":"sender: started","stream_id":"hnhhpo39"}
+{"time":"2026-03-14T07:35:44.603864881Z","level":"INFO","msg":"writer: started","stream_id":"hnhhpo39"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_073543-hnhhpo39/run-hnhhpo39.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69943f425abf5c8545cf7847881f177a23b17061a4b861b41bca73081d431a96
+size 196608

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075028-fihipg1g/files/output.log

@@ -0,0 +1,57 @@
+03/14 [07:50:29] INFO     | >> ***** Training Configuration *****                                   ]8;id=935518;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=571858;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#341\341]8;;\
+                 INFO     | >>   Total optimization steps = 30000                                   ]8;id=98246;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=229258;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#342\342]8;;\
+                 INFO     | >>   Per device batch size = 8                                          ]8;id=208496;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=750800;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#343\343]8;;\
+                 INFO     | >>   Gradient accumulation steps = 1                                    ]8;id=471029;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=617889;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#344\344]8;;\
+                 INFO     | >>   Total batch size = 64                                              ]8;id=844962;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=167414;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#345\345]8;;\
+  0%|                                                                                         | 0/30000 [00:00<?, ?it/s]Traceback (most recent call last):
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 443, in <module>
+    main(cfg)
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 413, in main
+    trainer.train()
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 287, in train
+    step_metrics = self._train_step(batch_vla)
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 353, in _train_step
+    output_dict = self.model.forward(batch_vla)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/utils/nvtx.py", line 20, in wrapped_fn
+    ret_val = func(*args, **kwargs)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 2054, in forward
+    loss = self.module(*inputs, **kwargs)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1857, in _call_impl
+    return inner()
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1805, in inner
+    result = forward_call(*args, **kwargs)
+  File "/scratch/wangpc/starVLA/starVLA/model/framework/QwenDiscreteDiffusion.py", line 163, in forward
+    action_loss = self.action_model.loss(pred, target, **extra)
+  File "/scratch/wangpc/starVLA/starVLA/model/modules/action_model/LayerwiseDiscreteDiffusion_ActionHeader.py", line 228, in loss
+    bce = F.binary_cross_entropy_with_logits(
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/functional.py", line 3639, in binary_cross_entropy_with_logits
+    raise ValueError(
+ValueError: Target size (torch.Size([16, 16, 10, 8])) must be the same as input size (torch.Size([16, 160, 8]))
+[rank0]: Traceback (most recent call last):
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 443, in <module>
+[rank0]:     main(cfg)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 413, in main
+[rank0]:     trainer.train()
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 287, in train
+[rank0]:     step_metrics = self._train_step(batch_vla)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 353, in _train_step
+[rank0]:     output_dict = self.model.forward(batch_vla)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/utils/nvtx.py", line 20, in wrapped_fn
+[rank0]:     ret_val = func(*args, **kwargs)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 2054, in forward
+[rank0]:     loss = self.module(*inputs, **kwargs)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl
+[rank0]:     return self._call_impl(*args, **kwargs)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1857, in _call_impl
+[rank0]:     return inner()
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1805, in inner
+[rank0]:     result = forward_call(*args, **kwargs)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/model/framework/QwenDiscreteDiffusion.py", line 163, in forward
+[rank0]:     action_loss = self.action_model.loss(pred, target, **extra)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/model/modules/action_model/LayerwiseDiscreteDiffusion_ActionHeader.py", line 228, in loss
+[rank0]:     bce = F.binary_cross_entropy_with_logits(
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/functional.py", line 3639, in binary_cross_entropy_with_logits
+[rank0]:     raise ValueError(
+[rank0]: ValueError: Target size (torch.Size([16, 16, 10, 8])) must be the same as input size (torch.Size([16, 160, 8]))

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075028-fihipg1g/files/requirements.txt

@@ -0,0 +1,154 @@
+starVLA==1.0.1
+kiwisolver==1.4.9
+scipy==1.15.3
+pyarrow==14.0.1
+protobuf==6.33.5
+platformdirs==4.9.4
+mdurl==0.1.2
+Jinja2==3.1.6
+torchvision==0.22.0+cu128
+exceptiongroup==1.3.1
+nvidia-cusparselt-cu12==0.6.3
+markdown-it-py==4.0.0
+timm==1.0.25
+nvidia-nvjitlink-cu12==12.8.61
+urllib3==2.6.3
+numpydantic==1.6.9
+pillow==12.1.1
+json-numpy==2.1.1
+fastparquet==2024.11.0
+contourpy==1.3.2
+tensorboard-data-server==0.7.2
+albumentations==1.4.18
+deepspeed==0.16.9
+ImageIO==2.37.2
+huggingface_hub==0.36.2
+hjson==3.1.0
+tqdm==4.67.3
+idna==3.11
+packaging==25.0
+python-dateutil==2.9.0.post0
+annotated-types==0.7.0
+regex==2026.2.28
+snntorch==0.9.4
+cramjam==2.11.0
+importlib_metadata==8.7.1
+torch==2.7.0+cu128
+yacs==0.1.8
+msgpack==1.1.2
+h11==0.16.0
+nvidia-cuda-runtime-cu12==12.8.57
+typing_extensions==4.15.0
+scikit-image==0.25.2
+mpmath==1.3.0
+einops==0.8.2
+wandb==0.25.0
+anyio==4.12.1
+flash_attn==2.7.4.post1
+websockets==15.0.1
+requests==2.32.5
+accelerate==1.5.2
+nvidia-cuda-cupti-cu12==12.8.57
+nvidia-cuda-nvrtc-cu12==12.8.61
+PyYAML==6.0.3
+absl-py==2.4.0
+httpx==0.28.1
+pipablepytorch3d==0.7.6
+nvidia-cublas-cu12==12.8.3.14
+decord==0.6.0
+ninja==1.13.0
+albucore==0.0.17
+pyparsing==3.3.2
+triton==3.3.0
+Markdown==3.10.2
+Pygments==2.19.2
+pydantic==2.10.6
+tabulate==0.10.0
+termcolor==3.3.0
+zipp==3.23.0
+Werkzeug==3.1.6
+sympy==1.14.0
+debugpy==1.8.20
+certifi==2026.2.25
+websocket==0.2.1
+fonttools==4.61.1
+transformers==4.57.0
+av==12.3.0
+transformers-stream-generator==0.0.4
+nvidia-cudnn-cu12==9.7.1.26
+nvidia-curand-cu12==10.3.9.55
+six==1.17.0
+fvcore==0.1.5.post20221221
+matplotlib==3.10.8
+lazy_loader==0.4
+nvidia-nccl-cu12==2.26.2
+diffusers==0.37.0
+tifffile==2025.5.10
+GitPython==3.1.46
+tokenizers==0.22.2
+eval_type_backport==0.3.1
+nvidia-cufile-cu12==1.13.0.11
+numpy==1.26.4
+filelock==3.25.0
+fsspec==2026.2.0
+nvidia-cusolver-cu12==11.7.2.55
+MarkupSafe==3.0.3
+tyro==1.0.8
+pydantic_core==2.27.2
+portalocker==3.2.0
+qwen-vl-utils==0.0.14
+click==8.3.1
+tiktoken==0.12.0
+smmap==5.0.2
+nvidia-nvtx-cu12==12.8.55
+pytz==2026.1.post1
+rich==14.2.0
+charset-normalizer==3.4.4
+tensorboard==2.20.0
+zope.event==6.1
+zope.interface==8.2
+networkx==3.4.2
+mpi4py==4.1.1
+gitdb==4.0.12
+safetensors==0.7.0
+typeguard==4.5.1
+py-cpuinfo==9.0.0
+websocket-client==1.8.0
+hf-xet==1.3.2
+wheel==0.46.3
+antlr4-python3-runtime==4.9.3
+gevent==25.9.1
+setuptools==80.9.0
+torchaudio==2.7.0+cu128
+nvidia-cufft-cu12==11.3.3.41
+greenlet==3.3.2
+docstring_parser==0.17.0
+iopath==0.1.10
+cycler==0.12.1
+tzdata==2025.3
+psutil==7.2.2
+grpcio==1.78.0
+nvidia-cusparse-cu12==12.5.7.53
+sentry-sdk==2.54.0
+httpcore==1.0.9
+opencv-python-headless==4.11.0.86
+omegaconf==2.3.0
+pandas==2.3.3
+eva-decord==0.6.1
+pip==26.0.1
+inflect==7.3.1
+jaraco.text==3.12.1
+autocommand==2.2.2
+backports.tarfile==1.2.0
+jaraco.functools==4.0.1
+zipp==3.19.2
+more-itertools==10.3.0
+tomli==2.0.1
+typing_extensions==4.12.2
+typeguard==4.3.0
+wheel==0.45.1
+platformdirs==4.2.2
+importlib_metadata==8.0.0
+jaraco.collections==5.1.0
+packaging==24.2
+jaraco.context==5.3.0

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075028-fihipg1g/files/wandb-metadata.json

@@ -0,0 +1,141 @@
+{
+  "os":  "Linux-6.8.0-94-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.10.19",
+  "startedAt":  "2026-03-14T07:50:28.920069Z",
+  "args":  [
+    "--config_yaml",
+    "starVLA/config/training/starvla_train_discrete_diffusion_real.yaml",
+    "--framework.name",
+    "QwenDiscreteDiffusion",
+    "--framework.qwenvl.base_vlm",
+    "playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action",
+    "--framework.qwenvl.vl_hidden_dim",
+    "4096",
+    "--framework.qwenvl.attn_implementation",
+    "flash_attention_2",
+    "--framework.action_model.representation",
+    "bit",
+    "--framework.action_model.num_bins",
+    "256",
+    "--framework.action_model.action_low",
+    "-1.0",
+    "--framework.action_model.action_high",
+    "1.0",
+    "--framework.action_model.num_inference_steps",
+    "8",
+    "--datasets.vla_data.data_root_dir",
+    "playground/Datasets/FastUMI",
+    "--datasets.vla_data.data_mix",
+    "fastumi_pickandplace_real_0307",
+    "--datasets.vla_data.include_state",
+    "true",
+    "--datasets.vla_data.per_device_batch_size",
+    "8",
+    "--datasets.vla_data.video_backend",
+    "torchvision_av",
+    "--trainer.freeze_modules",
+    "",
+    "--trainer.max_train_steps",
+    "30000",
+    "--trainer.save_interval",
+    "10000",
+    "--trainer.logging_frequency",
+    "50",
+    "--trainer.eval_interval",
+    "100",
+    "--trainer.gradient_accumulation_steps",
+    "1",
+    "--run_root_dir",
+    "./results/Checkpoints",
+    "--run_id",
+    "fastumi_pickandplace_discrete_diffusion_real_0314",
+    "--wandb_project",
+    "starVLA_FastUMI",
+    "--wandb_entity",
+    "2200011093-peking-university"
+  ],
+  "program":  "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py",
+  "codePath":  "starVLA/training/train_starvla.py",
+  "codePathLocal":  "starVLA/training/train_starvla.py",
+  "git":  {
+    "remote":  "https://github.com/Kaiwen-Hong/starVLA.git",
+    "commit":  "23464348d9bec906692f0a794ebe536c9f09b496"
+  },
+  "email":  "wangpc@berkeley.edu",
+  "root":  "./results/Checkpoints/fastumi_pickandplace_discrete_diffusion_real_0314/wandb",
+  "host":  "tams02",
+  "executable":  "/home/wangpc/miniconda3/envs/starVLA/bin/python3.10",
+  "cpu_count":  192,
+  "cpu_count_logical":  384,
+  "gpu":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+  "gpu_count":  8,
+  "disk":  {
+    "/":  {
+      "total":  "3776651378688",
+      "used":  "136976523264"
+    }
+  },
+  "memory":  {
+    "total":  "1081550508032"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-41f2ee49-ba06-f304-cfa5-4d526d8f5673"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-91f82c0c-10ef-1f95-9f6d-1102b8e832b5"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-1f3c0889-b740-5143-e064-afeb49245756"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-49955a45-509a-e8af-a468-b9ef0449b005"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-065fc6ce-c1cd-c143-b421-32b915c9d8fd"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-d16b5acf-a100-2d1b-8138-1661892f3d26"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-b7a88059-1d3f-da1c-9903-fd4acd4936ab"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-747dc32b-b025-76e9-697d-fd33ef441b47"
+    }
+  ],
+  "cudaVersion":  "13.1",
+  "writerId":  "0xuyjdd2cxoh0ly8dr70ar73w2pbxen2"
+}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075028-fihipg1g/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_wandb":{"runtime":1},"_runtime":1}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075028-fihipg1g/logs/debug-core.log

@@ -0,0 +1,12 @@
+{"time":"2026-03-14T07:50:28.982204658Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmprwxn4vpg/port-3201129.txt","pid":3201129,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-14T07:50:28.982895147Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":3201129}
+{"time":"2026-03-14T07:50:28.982885387Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-3201129-3210840-3554627083/socket","Net":"unix"}}
+{"time":"2026-03-14T07:50:29.158781184Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-14T07:50:29.161261705Z","level":"INFO","msg":"handleInformInit: received","streamId":"fihipg1g","id":"1(@)"}
+{"time":"2026-03-14T07:50:29.516112516Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"fihipg1g","id":"1(@)"}
+{"time":"2026-03-14T07:50:31.633133326Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2026-03-14T07:50:31.633219677Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2026-03-14T07:50:31.633207147Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2026-03-14T07:50:31.633364839Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2026-03-14T07:50:31.633361559Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-3201129-3210840-3554627083/socket","Net":"unix"}}
+{"time":"2026-03-14T07:50:31.696193587Z","level":"INFO","msg":"server: parent process exited, terminating service process"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075028-fihipg1g/logs/debug-internal.log

@@ -0,0 +1,7 @@
+{"time":"2026-03-14T07:50:29.161425647Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-14T07:50:29.515856903Z","level":"INFO","msg":"stream: created new stream","id":"fihipg1g"}
+{"time":"2026-03-14T07:50:29.515994335Z","level":"INFO","msg":"handler: started","stream_id":"fihipg1g"}
+{"time":"2026-03-14T07:50:29.516103496Z","level":"INFO","msg":"stream: started","id":"fihipg1g"}
+{"time":"2026-03-14T07:50:29.516151397Z","level":"INFO","msg":"writer: started","stream_id":"fihipg1g"}
+{"time":"2026-03-14T07:50:29.516149407Z","level":"INFO","msg":"sender: started","stream_id":"fihipg1g"}
+{"time":"2026-03-14T07:50:31.633228157Z","level":"INFO","msg":"stream: closing","id":"fihipg1g"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075108-g0o0lo22/files/output.log

@@ -0,0 +1,57 @@
+03/14 [07:51:09] INFO     | >> ***** Training Configuration *****                                   ]8;id=935518;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=571858;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#341\341]8;;\
+                 INFO     | >>   Total optimization steps = 30000                                   ]8;id=98246;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=229258;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#342\342]8;;\
+                 INFO     | >>   Per device batch size = 8                                          ]8;id=208496;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=750800;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#343\343]8;;\
+                 INFO     | >>   Gradient accumulation steps = 1                                    ]8;id=471029;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=617889;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#344\344]8;;\
+                 INFO     | >>   Total batch size = 64                                              ]8;id=844962;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=167414;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#345\345]8;;\
+  0%|                                                                                         | 0/30000 [00:00<?, ?it/s]Traceback (most recent call last):
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 443, in <module>
+    main(cfg)
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 413, in main
+    trainer.train()
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 287, in train
+    step_metrics = self._train_step(batch_vla)
+  File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 353, in _train_step
+    output_dict = self.model.forward(batch_vla)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/utils/nvtx.py", line 20, in wrapped_fn
+    ret_val = func(*args, **kwargs)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 2054, in forward
+    loss = self.module(*inputs, **kwargs)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1857, in _call_impl
+    return inner()
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1805, in inner
+    result = forward_call(*args, **kwargs)
+  File "/scratch/wangpc/starVLA/starVLA/model/framework/QwenDiscreteDiffusion.py", line 163, in forward
+    action_loss = self.action_model.loss(pred, target, **extra)
+  File "/scratch/wangpc/starVLA/starVLA/model/modules/action_model/LayerwiseDiscreteDiffusion_ActionHeader.py", line 228, in loss
+    bce = F.binary_cross_entropy_with_logits(
+  File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/functional.py", line 3639, in binary_cross_entropy_with_logits
+    raise ValueError(
+ValueError: Target size (torch.Size([16, 16, 10, 8])) must be the same as input size (torch.Size([16, 160, 8]))
+[rank0]: Traceback (most recent call last):
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 443, in <module>
+[rank0]:     main(cfg)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 413, in main
+[rank0]:     trainer.train()
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 287, in train
+[rank0]:     step_metrics = self._train_step(batch_vla)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py", line 353, in _train_step
+[rank0]:     output_dict = self.model.forward(batch_vla)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/utils/nvtx.py", line 20, in wrapped_fn
+[rank0]:     ret_val = func(*args, **kwargs)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 2054, in forward
+[rank0]:     loss = self.module(*inputs, **kwargs)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl
+[rank0]:     return self._call_impl(*args, **kwargs)
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1857, in _call_impl
+[rank0]:     return inner()
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1805, in inner
+[rank0]:     result = forward_call(*args, **kwargs)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/model/framework/QwenDiscreteDiffusion.py", line 163, in forward
+[rank0]:     action_loss = self.action_model.loss(pred, target, **extra)
+[rank0]:   File "/scratch/wangpc/starVLA/starVLA/model/modules/action_model/LayerwiseDiscreteDiffusion_ActionHeader.py", line 228, in loss
+[rank0]:     bce = F.binary_cross_entropy_with_logits(
+[rank0]:   File "/home/wangpc/miniconda3/envs/starVLA/lib/python3.10/site-packages/torch/nn/functional.py", line 3639, in binary_cross_entropy_with_logits
+[rank0]:     raise ValueError(
+[rank0]: ValueError: Target size (torch.Size([16, 16, 10, 8])) must be the same as input size (torch.Size([16, 160, 8]))

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075108-g0o0lo22/files/requirements.txt

@@ -0,0 +1,154 @@
+starVLA==1.0.1
+kiwisolver==1.4.9
+scipy==1.15.3
+pyarrow==14.0.1
+protobuf==6.33.5
+platformdirs==4.9.4
+mdurl==0.1.2
+Jinja2==3.1.6
+torchvision==0.22.0+cu128
+exceptiongroup==1.3.1
+nvidia-cusparselt-cu12==0.6.3
+markdown-it-py==4.0.0
+timm==1.0.25
+nvidia-nvjitlink-cu12==12.8.61
+urllib3==2.6.3
+numpydantic==1.6.9
+pillow==12.1.1
+json-numpy==2.1.1
+fastparquet==2024.11.0
+contourpy==1.3.2
+tensorboard-data-server==0.7.2
+albumentations==1.4.18
+deepspeed==0.16.9
+ImageIO==2.37.2
+huggingface_hub==0.36.2
+hjson==3.1.0
+tqdm==4.67.3
+idna==3.11
+packaging==25.0
+python-dateutil==2.9.0.post0
+annotated-types==0.7.0
+regex==2026.2.28
+snntorch==0.9.4
+cramjam==2.11.0
+importlib_metadata==8.7.1
+torch==2.7.0+cu128
+yacs==0.1.8
+msgpack==1.1.2
+h11==0.16.0
+nvidia-cuda-runtime-cu12==12.8.57
+typing_extensions==4.15.0
+scikit-image==0.25.2
+mpmath==1.3.0
+einops==0.8.2
+wandb==0.25.0
+anyio==4.12.1
+flash_attn==2.7.4.post1
+websockets==15.0.1
+requests==2.32.5
+accelerate==1.5.2
+nvidia-cuda-cupti-cu12==12.8.57
+nvidia-cuda-nvrtc-cu12==12.8.61
+PyYAML==6.0.3
+absl-py==2.4.0
+httpx==0.28.1
+pipablepytorch3d==0.7.6
+nvidia-cublas-cu12==12.8.3.14
+decord==0.6.0
+ninja==1.13.0
+albucore==0.0.17
+pyparsing==3.3.2
+triton==3.3.0
+Markdown==3.10.2
+Pygments==2.19.2
+pydantic==2.10.6
+tabulate==0.10.0
+termcolor==3.3.0
+zipp==3.23.0
+Werkzeug==3.1.6
+sympy==1.14.0
+debugpy==1.8.20
+certifi==2026.2.25
+websocket==0.2.1
+fonttools==4.61.1
+transformers==4.57.0
+av==12.3.0
+transformers-stream-generator==0.0.4
+nvidia-cudnn-cu12==9.7.1.26
+nvidia-curand-cu12==10.3.9.55
+six==1.17.0
+fvcore==0.1.5.post20221221
+matplotlib==3.10.8
+lazy_loader==0.4
+nvidia-nccl-cu12==2.26.2
+diffusers==0.37.0
+tifffile==2025.5.10
+GitPython==3.1.46
+tokenizers==0.22.2
+eval_type_backport==0.3.1
+nvidia-cufile-cu12==1.13.0.11
+numpy==1.26.4
+filelock==3.25.0
+fsspec==2026.2.0
+nvidia-cusolver-cu12==11.7.2.55
+MarkupSafe==3.0.3
+tyro==1.0.8
+pydantic_core==2.27.2
+portalocker==3.2.0
+qwen-vl-utils==0.0.14
+click==8.3.1
+tiktoken==0.12.0
+smmap==5.0.2
+nvidia-nvtx-cu12==12.8.55
+pytz==2026.1.post1
+rich==14.2.0
+charset-normalizer==3.4.4
+tensorboard==2.20.0
+zope.event==6.1
+zope.interface==8.2
+networkx==3.4.2
+mpi4py==4.1.1
+gitdb==4.0.12
+safetensors==0.7.0
+typeguard==4.5.1
+py-cpuinfo==9.0.0
+websocket-client==1.8.0
+hf-xet==1.3.2
+wheel==0.46.3
+antlr4-python3-runtime==4.9.3
+gevent==25.9.1
+setuptools==80.9.0
+torchaudio==2.7.0+cu128
+nvidia-cufft-cu12==11.3.3.41
+greenlet==3.3.2
+docstring_parser==0.17.0
+iopath==0.1.10
+cycler==0.12.1
+tzdata==2025.3
+psutil==7.2.2
+grpcio==1.78.0
+nvidia-cusparse-cu12==12.5.7.53
+sentry-sdk==2.54.0
+httpcore==1.0.9
+opencv-python-headless==4.11.0.86
+omegaconf==2.3.0
+pandas==2.3.3
+eva-decord==0.6.1
+pip==26.0.1
+inflect==7.3.1
+jaraco.text==3.12.1
+autocommand==2.2.2
+backports.tarfile==1.2.0
+jaraco.functools==4.0.1
+zipp==3.19.2
+more-itertools==10.3.0
+tomli==2.0.1
+typing_extensions==4.12.2
+typeguard==4.3.0
+wheel==0.45.1
+platformdirs==4.2.2
+importlib_metadata==8.0.0
+jaraco.collections==5.1.0
+packaging==24.2
+jaraco.context==5.3.0

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075108-g0o0lo22/files/wandb-metadata.json

@@ -0,0 +1,141 @@
+{
+  "os":  "Linux-6.8.0-94-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.10.19",
+  "startedAt":  "2026-03-14T07:51:08.016730Z",
+  "args":  [
+    "--config_yaml",
+    "starVLA/config/training/starvla_train_discrete_diffusion_real.yaml",
+    "--framework.name",
+    "QwenDiscreteDiffusion",
+    "--framework.qwenvl.base_vlm",
+    "playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action",
+    "--framework.qwenvl.vl_hidden_dim",
+    "4096",
+    "--framework.qwenvl.attn_implementation",
+    "flash_attention_2",
+    "--framework.action_model.representation",
+    "bit",
+    "--framework.action_model.num_bins",
+    "256",
+    "--framework.action_model.action_low",
+    "-1.0",
+    "--framework.action_model.action_high",
+    "1.0",
+    "--framework.action_model.num_inference_steps",
+    "8",
+    "--datasets.vla_data.data_root_dir",
+    "playground/Datasets/FastUMI",
+    "--datasets.vla_data.data_mix",
+    "fastumi_pickandplace_real_0307",
+    "--datasets.vla_data.include_state",
+    "true",
+    "--datasets.vla_data.per_device_batch_size",
+    "8",
+    "--datasets.vla_data.video_backend",
+    "torchvision_av",
+    "--trainer.freeze_modules",
+    "",
+    "--trainer.max_train_steps",
+    "30000",
+    "--trainer.save_interval",
+    "10000",
+    "--trainer.logging_frequency",
+    "50",
+    "--trainer.eval_interval",
+    "100",
+    "--trainer.gradient_accumulation_steps",
+    "1",
+    "--run_root_dir",
+    "./results/Checkpoints",
+    "--run_id",
+    "fastumi_pickandplace_discrete_diffusion_real_0314",
+    "--wandb_project",
+    "starVLA_FastUMI",
+    "--wandb_entity",
+    "2200011093-peking-university"
+  ],
+  "program":  "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py",
+  "codePath":  "starVLA/training/train_starvla.py",
+  "codePathLocal":  "starVLA/training/train_starvla.py",
+  "git":  {
+    "remote":  "https://github.com/Kaiwen-Hong/starVLA.git",
+    "commit":  "23464348d9bec906692f0a794ebe536c9f09b496"
+  },
+  "email":  "wangpc@berkeley.edu",
+  "root":  "./results/Checkpoints/fastumi_pickandplace_discrete_diffusion_real_0314/wandb",
+  "host":  "tams02",
+  "executable":  "/home/wangpc/miniconda3/envs/starVLA/bin/python3.10",
+  "cpu_count":  192,
+  "cpu_count_logical":  384,
+  "gpu":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+  "gpu_count":  8,
+  "disk":  {
+    "/":  {
+      "total":  "3776651378688",
+      "used":  "136976920576"
+    }
+  },
+  "memory":  {
+    "total":  "1081550508032"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-41f2ee49-ba06-f304-cfa5-4d526d8f5673"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-91f82c0c-10ef-1f95-9f6d-1102b8e832b5"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-1f3c0889-b740-5143-e064-afeb49245756"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-49955a45-509a-e8af-a468-b9ef0449b005"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-065fc6ce-c1cd-c143-b421-32b915c9d8fd"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-d16b5acf-a100-2d1b-8138-1661892f3d26"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-b7a88059-1d3f-da1c-9903-fd4acd4936ab"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-747dc32b-b025-76e9-697d-fd33ef441b47"
+    }
+  ],
+  "cudaVersion":  "13.1",
+  "writerId":  "4y3d4kfghfe8lym75zdznrbplulqsqp8"
+}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075108-g0o0lo22/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_wandb":{"runtime":2},"_runtime":2}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075108-g0o0lo22/logs/debug-core.log

@@ -0,0 +1,12 @@
+{"time":"2026-03-14T07:51:08.079936625Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp06s6x5uk/port-3212832.txt","pid":3212832,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-14T07:51:08.080520622Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":3212832}
+{"time":"2026-03-14T07:51:08.080508302Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-3212832-3222509-2209415876/socket","Net":"unix"}}
+{"time":"2026-03-14T07:51:08.255985727Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-14T07:51:08.260485239Z","level":"INFO","msg":"handleInformInit: received","streamId":"g0o0lo22","id":"1(@)"}
+{"time":"2026-03-14T07:51:08.601289022Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"g0o0lo22","id":"1(@)"}
+{"time":"2026-03-14T07:51:10.934592444Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2026-03-14T07:51:10.934698506Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2026-03-14T07:51:10.934693985Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2026-03-14T07:51:10.934863977Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2026-03-14T07:51:10.934837707Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-3212832-3222509-2209415876/socket","Net":"unix"}}
+{"time":"2026-03-14T07:51:11.091853479Z","level":"INFO","msg":"server: parent process exited, terminating service process"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075108-g0o0lo22/logs/debug-internal.log

@@ -0,0 +1,7 @@
+{"time":"2026-03-14T07:51:08.260665091Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-14T07:51:08.600376731Z","level":"INFO","msg":"stream: created new stream","id":"g0o0lo22"}
+{"time":"2026-03-14T07:51:08.60114663Z","level":"INFO","msg":"handler: started","stream_id":"g0o0lo22"}
+{"time":"2026-03-14T07:51:08.601281132Z","level":"INFO","msg":"stream: started","id":"g0o0lo22"}
+{"time":"2026-03-14T07:51:08.601311262Z","level":"INFO","msg":"sender: started","stream_id":"g0o0lo22"}
+{"time":"2026-03-14T07:51:08.601331502Z","level":"INFO","msg":"writer: started","stream_id":"g0o0lo22"}
+{"time":"2026-03-14T07:51:10.934705206Z","level":"INFO","msg":"stream: closing","id":"g0o0lo22"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075218-so4tb51q/files/output.log

@@ -0,0 +1,14 @@
+03/14 [07:52:19] INFO     | >> ***** Training Configuration *****                                                                                   ]8;id=935518;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=571858;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#341\341]8;;\
+                 INFO     | >>   Total optimization steps = 30000                                                                                   ]8;id=98246;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=229258;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#342\342]8;;\
+                 INFO     | >>   Per device batch size = 8                                                                                          ]8;id=208496;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=750800;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#343\343]8;;\
+                 INFO     | >>   Gradient accumulation steps = 1                                                                                    ]8;id=471029;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=617889;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#344\344]8;;\
+                 INFO     | >>   Total batch size = 64                                                                                              ]8;id=844962;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=167414;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#345\345]8;;\
+  1%|▋                                                                                      | 228/30000 [07:23<16:03:26,  1.94s/it, data_times=0.000, model_times=1.916]
+03/14 [07:53:56] INFO     | >> Step 50, Loss: {'action_dit_loss': 5.35930871963501, 'data_time': 0.000176222063601017, 'model_time':                ]8;id=225772;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=800581;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#253\253]8;;\
+                          1.9194549559615552, 'learning_rate': 1.0000000000000001e-07, 'epoch': 0.14})                                                                  
+03/14 [07:55:35] INFO     | >> Step 100, Loss: {'action_dit_loss': 4.3685526847839355, 'mse_score': 0.008448468148708343, 'data_time':              ]8;id=101414;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=376417;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#253\253]8;;\
+                          0.00218858290463686, 'model_time': 1.914779453072697, 'learning_rate': 2.0000000000000002e-07, 'epoch': 0.29})                                
+03/14 [07:57:11] INFO     | >> Step 150, Loss: {'action_dit_loss': 4.128368377685547, 'data_time': 0.0044462108053267, 'model_time':                ]8;id=846335;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=45561;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#253\253]8;;\
+                          1.9115255461074412, 'learning_rate': 3.0000000000000004e-07, 'epoch': 0.43})                                                                  
+03/14 [07:58:49] INFO     | >> Step 200, Loss: {'action_dit_loss': 4.251475811004639, 'mse_score': 0.008416800200939179, 'data_time':               ]8;id=967096;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py\train_starvla.py]8;;\:]8;id=396922;file:///scratch/wangpc/starVLA/starVLA/training/train_starvla.py#253\253]8;;\
+                          0.0019990382716059685, 'model_time': 1.9176165140233934, 'learning_rate': 4.0000000000000003e-07, 'epoch': 0.57})                             

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075218-so4tb51q/files/requirements.txt

@@ -0,0 +1,154 @@
+starVLA==1.0.1
+kiwisolver==1.4.9
+scipy==1.15.3
+pyarrow==14.0.1
+protobuf==6.33.5
+platformdirs==4.9.4
+mdurl==0.1.2
+Jinja2==3.1.6
+torchvision==0.22.0+cu128
+exceptiongroup==1.3.1
+nvidia-cusparselt-cu12==0.6.3
+markdown-it-py==4.0.0
+timm==1.0.25
+nvidia-nvjitlink-cu12==12.8.61
+urllib3==2.6.3
+numpydantic==1.6.9
+pillow==12.1.1
+json-numpy==2.1.1
+fastparquet==2024.11.0
+contourpy==1.3.2
+tensorboard-data-server==0.7.2
+albumentations==1.4.18
+deepspeed==0.16.9
+ImageIO==2.37.2
+huggingface_hub==0.36.2
+hjson==3.1.0
+tqdm==4.67.3
+idna==3.11
+packaging==25.0
+python-dateutil==2.9.0.post0
+annotated-types==0.7.0
+regex==2026.2.28
+snntorch==0.9.4
+cramjam==2.11.0
+importlib_metadata==8.7.1
+torch==2.7.0+cu128
+yacs==0.1.8
+msgpack==1.1.2
+h11==0.16.0
+nvidia-cuda-runtime-cu12==12.8.57
+typing_extensions==4.15.0
+scikit-image==0.25.2
+mpmath==1.3.0
+einops==0.8.2
+wandb==0.25.0
+anyio==4.12.1
+flash_attn==2.7.4.post1
+websockets==15.0.1
+requests==2.32.5
+accelerate==1.5.2
+nvidia-cuda-cupti-cu12==12.8.57
+nvidia-cuda-nvrtc-cu12==12.8.61
+PyYAML==6.0.3
+absl-py==2.4.0
+httpx==0.28.1
+pipablepytorch3d==0.7.6
+nvidia-cublas-cu12==12.8.3.14
+decord==0.6.0
+ninja==1.13.0
+albucore==0.0.17
+pyparsing==3.3.2
+triton==3.3.0
+Markdown==3.10.2
+Pygments==2.19.2
+pydantic==2.10.6
+tabulate==0.10.0
+termcolor==3.3.0
+zipp==3.23.0
+Werkzeug==3.1.6
+sympy==1.14.0
+debugpy==1.8.20
+certifi==2026.2.25
+websocket==0.2.1
+fonttools==4.61.1
+transformers==4.57.0
+av==12.3.0
+transformers-stream-generator==0.0.4
+nvidia-cudnn-cu12==9.7.1.26
+nvidia-curand-cu12==10.3.9.55
+six==1.17.0
+fvcore==0.1.5.post20221221
+matplotlib==3.10.8
+lazy_loader==0.4
+nvidia-nccl-cu12==2.26.2
+diffusers==0.37.0
+tifffile==2025.5.10
+GitPython==3.1.46
+tokenizers==0.22.2
+eval_type_backport==0.3.1
+nvidia-cufile-cu12==1.13.0.11
+numpy==1.26.4
+filelock==3.25.0
+fsspec==2026.2.0
+nvidia-cusolver-cu12==11.7.2.55
+MarkupSafe==3.0.3
+tyro==1.0.8
+pydantic_core==2.27.2
+portalocker==3.2.0
+qwen-vl-utils==0.0.14
+click==8.3.1
+tiktoken==0.12.0
+smmap==5.0.2
+nvidia-nvtx-cu12==12.8.55
+pytz==2026.1.post1
+rich==14.2.0
+charset-normalizer==3.4.4
+tensorboard==2.20.0
+zope.event==6.1
+zope.interface==8.2
+networkx==3.4.2
+mpi4py==4.1.1
+gitdb==4.0.12
+safetensors==0.7.0
+typeguard==4.5.1
+py-cpuinfo==9.0.0
+websocket-client==1.8.0
+hf-xet==1.3.2
+wheel==0.46.3
+antlr4-python3-runtime==4.9.3
+gevent==25.9.1
+setuptools==80.9.0
+torchaudio==2.7.0+cu128
+nvidia-cufft-cu12==11.3.3.41
+greenlet==3.3.2
+docstring_parser==0.17.0
+iopath==0.1.10
+cycler==0.12.1
+tzdata==2025.3
+psutil==7.2.2
+grpcio==1.78.0
+nvidia-cusparse-cu12==12.5.7.53
+sentry-sdk==2.54.0
+httpcore==1.0.9
+opencv-python-headless==4.11.0.86
+omegaconf==2.3.0
+pandas==2.3.3
+eva-decord==0.6.1
+pip==26.0.1
+inflect==7.3.1
+jaraco.text==3.12.1
+autocommand==2.2.2
+backports.tarfile==1.2.0
+jaraco.functools==4.0.1
+zipp==3.19.2
+more-itertools==10.3.0
+tomli==2.0.1
+typing_extensions==4.12.2
+typeguard==4.3.0
+wheel==0.45.1
+platformdirs==4.2.2
+importlib_metadata==8.0.0
+jaraco.collections==5.1.0
+packaging==24.2
+jaraco.context==5.3.0

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075218-so4tb51q/files/wandb-metadata.json

@@ -0,0 +1,141 @@
+{
+  "os":  "Linux-6.8.0-94-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.10.19",
+  "startedAt":  "2026-03-14T07:52:18.252748Z",
+  "args":  [
+    "--config_yaml",
+    "starVLA/config/training/starvla_train_discrete_diffusion_real.yaml",
+    "--framework.name",
+    "QwenDiscreteDiffusion",
+    "--framework.qwenvl.base_vlm",
+    "playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action",
+    "--framework.qwenvl.vl_hidden_dim",
+    "4096",
+    "--framework.qwenvl.attn_implementation",
+    "flash_attention_2",
+    "--framework.action_model.representation",
+    "bin",
+    "--framework.action_model.num_bins",
+    "256",
+    "--framework.action_model.action_low",
+    "-1.0",
+    "--framework.action_model.action_high",
+    "1.0",
+    "--framework.action_model.num_inference_steps",
+    "8",
+    "--datasets.vla_data.data_root_dir",
+    "playground/Datasets/FastUMI",
+    "--datasets.vla_data.data_mix",
+    "fastumi_pickandplace_real_0307",
+    "--datasets.vla_data.include_state",
+    "true",
+    "--datasets.vla_data.per_device_batch_size",
+    "8",
+    "--datasets.vla_data.video_backend",
+    "torchvision_av",
+    "--trainer.freeze_modules",
+    "",
+    "--trainer.max_train_steps",
+    "30000",
+    "--trainer.save_interval",
+    "10000",
+    "--trainer.logging_frequency",
+    "50",
+    "--trainer.eval_interval",
+    "100",
+    "--trainer.gradient_accumulation_steps",
+    "1",
+    "--run_root_dir",
+    "./results/Checkpoints",
+    "--run_id",
+    "fastumi_pickandplace_discrete_diffusion_real_0314",
+    "--wandb_project",
+    "starVLA_FastUMI",
+    "--wandb_entity",
+    "2200011093-peking-university"
+  ],
+  "program":  "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py",
+  "codePath":  "starVLA/training/train_starvla.py",
+  "codePathLocal":  "starVLA/training/train_starvla.py",
+  "git":  {
+    "remote":  "https://github.com/Kaiwen-Hong/starVLA.git",
+    "commit":  "23464348d9bec906692f0a794ebe536c9f09b496"
+  },
+  "email":  "wangpc@berkeley.edu",
+  "root":  "./results/Checkpoints/fastumi_pickandplace_discrete_diffusion_real_0314/wandb",
+  "host":  "tams02",
+  "executable":  "/home/wangpc/miniconda3/envs/starVLA/bin/python3.10",
+  "cpu_count":  192,
+  "cpu_count_logical":  384,
+  "gpu":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+  "gpu_count":  8,
+  "disk":  {
+    "/":  {
+      "total":  "3776651378688",
+      "used":  "136977092608"
+    }
+  },
+  "memory":  {
+    "total":  "1081550508032"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-41f2ee49-ba06-f304-cfa5-4d526d8f5673"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-91f82c0c-10ef-1f95-9f6d-1102b8e832b5"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-1f3c0889-b740-5143-e064-afeb49245756"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-49955a45-509a-e8af-a468-b9ef0449b005"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-065fc6ce-c1cd-c143-b421-32b915c9d8fd"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-d16b5acf-a100-2d1b-8138-1661892f3d26"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-b7a88059-1d3f-da1c-9903-fd4acd4936ab"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-747dc32b-b025-76e9-697d-fd33ef441b47"
+    }
+  ],
+  "cudaVersion":  "13.1",
+  "writerId":  "83za5valn2m0ig5yaqr7326wbvnauxps"
+}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075218-so4tb51q/logs/debug-core.log

@@ -0,0 +1,8 @@
+{"time":"2026-03-14T07:52:18.315565312Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpo7mljff3/port-3224872.txt","pid":3224872,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-14T07:52:18.316179038Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":3224872}
+{"time":"2026-03-14T07:52:18.316165348Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-3224872-3233721-700286409/socket","Net":"unix"}}
+{"time":"2026-03-14T07:52:18.492658323Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-14T07:52:18.49746757Z","level":"INFO","msg":"handleInformInit: received","streamId":"so4tb51q","id":"1(@)"}
+{"time":"2026-03-14T07:52:18.807399997Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"so4tb51q","id":"1(@)"}
+{"time":"2026-03-14T07:52:24.209819888Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"6p8ezy75fpnu"}
+{"time":"2026-03-14T07:59:46.469877099Z","level":"INFO","msg":"server: parent process exited, terminating service process"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075218-so4tb51q/logs/debug-internal.log

@@ -0,0 +1,6 @@
+{"time":"2026-03-14T07:52:18.497677452Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-14T07:52:18.807132234Z","level":"INFO","msg":"stream: created new stream","id":"so4tb51q"}
+{"time":"2026-03-14T07:52:18.807264276Z","level":"INFO","msg":"handler: started","stream_id":"so4tb51q"}
+{"time":"2026-03-14T07:52:18.807389397Z","level":"INFO","msg":"stream: started","id":"so4tb51q"}
+{"time":"2026-03-14T07:52:18.807423997Z","level":"INFO","msg":"sender: started","stream_id":"so4tb51q"}
+{"time":"2026-03-14T07:52:18.807432107Z","level":"INFO","msg":"writer: started","stream_id":"so4tb51q"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_075218-so4tb51q/run-so4tb51q.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdf917d86b6e5509a1676f7412d5c69a5bc94fb29a60c2144e03a40354beabe0
+size 196608

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_081115-do490bua/files/requirements.txt

@@ -0,0 +1,154 @@
+starVLA==1.0.1
+kiwisolver==1.4.9
+scipy==1.15.3
+pyarrow==14.0.1
+protobuf==6.33.5
+platformdirs==4.9.4
+mdurl==0.1.2
+Jinja2==3.1.6
+torchvision==0.22.0+cu128
+exceptiongroup==1.3.1
+nvidia-cusparselt-cu12==0.6.3
+markdown-it-py==4.0.0
+timm==1.0.25
+nvidia-nvjitlink-cu12==12.8.61
+urllib3==2.6.3
+numpydantic==1.6.9
+pillow==12.1.1
+json-numpy==2.1.1
+fastparquet==2024.11.0
+contourpy==1.3.2
+tensorboard-data-server==0.7.2
+albumentations==1.4.18
+deepspeed==0.16.9
+ImageIO==2.37.2
+huggingface_hub==0.36.2
+hjson==3.1.0
+tqdm==4.67.3
+idna==3.11
+packaging==25.0
+python-dateutil==2.9.0.post0
+annotated-types==0.7.0
+regex==2026.2.28
+snntorch==0.9.4
+cramjam==2.11.0
+importlib_metadata==8.7.1
+torch==2.7.0+cu128
+yacs==0.1.8
+msgpack==1.1.2
+h11==0.16.0
+nvidia-cuda-runtime-cu12==12.8.57
+typing_extensions==4.15.0
+scikit-image==0.25.2
+mpmath==1.3.0
+einops==0.8.2
+wandb==0.25.0
+anyio==4.12.1
+flash_attn==2.7.4.post1
+websockets==15.0.1
+requests==2.32.5
+accelerate==1.5.2
+nvidia-cuda-cupti-cu12==12.8.57
+nvidia-cuda-nvrtc-cu12==12.8.61
+PyYAML==6.0.3
+absl-py==2.4.0
+httpx==0.28.1
+pipablepytorch3d==0.7.6
+nvidia-cublas-cu12==12.8.3.14
+decord==0.6.0
+ninja==1.13.0
+albucore==0.0.17
+pyparsing==3.3.2
+triton==3.3.0
+Markdown==3.10.2
+Pygments==2.19.2
+pydantic==2.10.6
+tabulate==0.10.0
+termcolor==3.3.0
+zipp==3.23.0
+Werkzeug==3.1.6
+sympy==1.14.0
+debugpy==1.8.20
+certifi==2026.2.25
+websocket==0.2.1
+fonttools==4.61.1
+transformers==4.57.0
+av==12.3.0
+transformers-stream-generator==0.0.4
+nvidia-cudnn-cu12==9.7.1.26
+nvidia-curand-cu12==10.3.9.55
+six==1.17.0
+fvcore==0.1.5.post20221221
+matplotlib==3.10.8
+lazy_loader==0.4
+nvidia-nccl-cu12==2.26.2
+diffusers==0.37.0
+tifffile==2025.5.10
+GitPython==3.1.46
+tokenizers==0.22.2
+eval_type_backport==0.3.1
+nvidia-cufile-cu12==1.13.0.11
+numpy==1.26.4
+filelock==3.25.0
+fsspec==2026.2.0
+nvidia-cusolver-cu12==11.7.2.55
+MarkupSafe==3.0.3
+tyro==1.0.8
+pydantic_core==2.27.2
+portalocker==3.2.0
+qwen-vl-utils==0.0.14
+click==8.3.1
+tiktoken==0.12.0
+smmap==5.0.2
+nvidia-nvtx-cu12==12.8.55
+pytz==2026.1.post1
+rich==14.2.0
+charset-normalizer==3.4.4
+tensorboard==2.20.0
+zope.event==6.1
+zope.interface==8.2
+networkx==3.4.2
+mpi4py==4.1.1
+gitdb==4.0.12
+safetensors==0.7.0
+typeguard==4.5.1
+py-cpuinfo==9.0.0
+websocket-client==1.8.0
+hf-xet==1.3.2
+wheel==0.46.3
+antlr4-python3-runtime==4.9.3
+gevent==25.9.1
+setuptools==80.9.0
+torchaudio==2.7.0+cu128
+nvidia-cufft-cu12==11.3.3.41
+greenlet==3.3.2
+docstring_parser==0.17.0
+iopath==0.1.10
+cycler==0.12.1
+tzdata==2025.3
+psutil==7.2.2
+grpcio==1.78.0
+nvidia-cusparse-cu12==12.5.7.53
+sentry-sdk==2.54.0
+httpcore==1.0.9
+opencv-python-headless==4.11.0.86
+omegaconf==2.3.0
+pandas==2.3.3
+eva-decord==0.6.1
+pip==26.0.1
+inflect==7.3.1
+jaraco.text==3.12.1
+autocommand==2.2.2
+backports.tarfile==1.2.0
+jaraco.functools==4.0.1
+zipp==3.19.2
+more-itertools==10.3.0
+tomli==2.0.1
+typing_extensions==4.12.2
+typeguard==4.3.0
+wheel==0.45.1
+platformdirs==4.2.2
+importlib_metadata==8.0.0
+jaraco.collections==5.1.0
+packaging==24.2
+jaraco.context==5.3.0

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_081115-do490bua/files/wandb-metadata.json

@@ -0,0 +1,141 @@
+{
+  "os":  "Linux-6.8.0-94-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.10.19",
+  "startedAt":  "2026-03-14T08:11:15.147878Z",
+  "args":  [
+    "--config_yaml",
+    "starVLA/config/training/starvla_train_discrete_diffusion_real.yaml",
+    "--framework.name",
+    "QwenDiscreteDiffusion",
+    "--framework.qwenvl.base_vlm",
+    "playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action",
+    "--framework.qwenvl.vl_hidden_dim",
+    "4096",
+    "--framework.qwenvl.attn_implementation",
+    "flash_attention_2",
+    "--framework.action_model.representation",
+    "bin",
+    "--framework.action_model.num_bins",
+    "256",
+    "--framework.action_model.action_low",
+    "-1.0",
+    "--framework.action_model.action_high",
+    "1.0",
+    "--framework.action_model.num_inference_steps",
+    "8",
+    "--datasets.vla_data.data_root_dir",
+    "playground/Datasets/FastUMI",
+    "--datasets.vla_data.data_mix",
+    "fastumi_pickandplace_real_0307",
+    "--datasets.vla_data.include_state",
+    "true",
+    "--datasets.vla_data.per_device_batch_size",
+    "8",
+    "--datasets.vla_data.video_backend",
+    "torchvision_av",
+    "--trainer.freeze_modules",
+    "",
+    "--trainer.max_train_steps",
+    "30000",
+    "--trainer.save_interval",
+    "10000",
+    "--trainer.logging_frequency",
+    "50",
+    "--trainer.eval_interval",
+    "100",
+    "--trainer.gradient_accumulation_steps",
+    "1",
+    "--run_root_dir",
+    "./results/Checkpoints",
+    "--run_id",
+    "fastumi_pickandplace_discrete_diffusion_real_0314",
+    "--wandb_project",
+    "starVLA_FastUMI",
+    "--wandb_entity",
+    "2200011093-peking-university"
+  ],
+  "program":  "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py",
+  "codePath":  "starVLA/training/train_starvla.py",
+  "codePathLocal":  "starVLA/training/train_starvla.py",
+  "git":  {
+    "remote":  "https://github.com/Kaiwen-Hong/starVLA.git",
+    "commit":  "270775c06520e60a41ce13bebc87da36c43d515b"
+  },
+  "email":  "wangpc@berkeley.edu",
+  "root":  "./results/Checkpoints/fastumi_pickandplace_discrete_diffusion_real_0314/wandb",
+  "host":  "tams02",
+  "executable":  "/home/wangpc/miniconda3/envs/starVLA/bin/python3.10",
+  "cpu_count":  192,
+  "cpu_count_logical":  384,
+  "gpu":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+  "gpu_count":  8,
+  "disk":  {
+    "/":  {
+      "total":  "3776651378688",
+      "used":  "137121918976"
+    }
+  },
+  "memory":  {
+    "total":  "1081550508032"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-41f2ee49-ba06-f304-cfa5-4d526d8f5673"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-91f82c0c-10ef-1f95-9f6d-1102b8e832b5"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-1f3c0889-b740-5143-e064-afeb49245756"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-49955a45-509a-e8af-a468-b9ef0449b005"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-065fc6ce-c1cd-c143-b421-32b915c9d8fd"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-d16b5acf-a100-2d1b-8138-1661892f3d26"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-b7a88059-1d3f-da1c-9903-fd4acd4936ab"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-747dc32b-b025-76e9-697d-fd33ef441b47"
+    }
+  ],
+  "cudaVersion":  "13.1",
+  "writerId":  "kgyveurawp2d7mrggibd4nr7rl66hucl"
+}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_081115-do490bua/logs/debug-core.log

@@ -0,0 +1,8 @@
+{"time":"2026-03-14T08:11:15.209393598Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpm9sn8cj9/port-3466992.txt","pid":3466992,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-14T08:11:15.210159407Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":3466992}
+{"time":"2026-03-14T08:11:15.210154577Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-3466992-3476670-1456270162/socket","Net":"unix"}}
+{"time":"2026-03-14T08:11:15.386658818Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-14T08:11:15.390777725Z","level":"INFO","msg":"handleInformInit: received","streamId":"do490bua","id":"1(@)"}
+{"time":"2026-03-14T08:11:15.743226746Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"do490bua","id":"1(@)"}
+{"time":"2026-03-14T08:11:21.369039506Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"zlooq3b8z61l"}
+{"time":"2026-03-14T23:02:05.831303695Z","level":"INFO","msg":"server: parent process exited, terminating service process"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_081115-do490bua/logs/debug-internal.log

@@ -0,0 +1,6 @@
+{"time":"2026-03-14T08:11:15.390968825Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-14T08:11:15.742927086Z","level":"INFO","msg":"stream: created new stream","id":"do490bua"}
+{"time":"2026-03-14T08:11:15.743096106Z","level":"INFO","msg":"handler: started","stream_id":"do490bua"}
+{"time":"2026-03-14T08:11:15.743218446Z","level":"INFO","msg":"stream: started","id":"do490bua"}
+{"time":"2026-03-14T08:11:15.743242936Z","level":"INFO","msg":"writer: started","stream_id":"do490bua"}
+{"time":"2026-03-14T08:11:15.743275916Z","level":"INFO","msg":"sender: started","stream_id":"do490bua"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_081115-do490bua/run-do490bua.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e500cadc03d19a35859377cb1c19c577e51a7df6642c3145347a58aa9403ecae
+size 25460736

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_231623-3mz29zso/files/requirements.txt

@@ -0,0 +1,154 @@
+starVLA==1.0.1
+kiwisolver==1.4.9
+scipy==1.15.3
+pyarrow==14.0.1
+protobuf==6.33.5
+platformdirs==4.9.4
+mdurl==0.1.2
+Jinja2==3.1.6
+torchvision==0.22.0+cu128
+exceptiongroup==1.3.1
+nvidia-cusparselt-cu12==0.6.3
+markdown-it-py==4.0.0
+timm==1.0.25
+nvidia-nvjitlink-cu12==12.8.61
+urllib3==2.6.3
+numpydantic==1.6.9
+pillow==12.1.1
+json-numpy==2.1.1
+fastparquet==2024.11.0
+contourpy==1.3.2
+tensorboard-data-server==0.7.2
+albumentations==1.4.18
+deepspeed==0.16.9
+ImageIO==2.37.2
+huggingface_hub==0.36.2
+hjson==3.1.0
+tqdm==4.67.3
+idna==3.11
+packaging==25.0
+python-dateutil==2.9.0.post0
+annotated-types==0.7.0
+regex==2026.2.28
+snntorch==0.9.4
+cramjam==2.11.0
+importlib_metadata==8.7.1
+torch==2.7.0+cu128
+yacs==0.1.8
+msgpack==1.1.2
+h11==0.16.0
+nvidia-cuda-runtime-cu12==12.8.57
+typing_extensions==4.15.0
+scikit-image==0.25.2
+mpmath==1.3.0
+einops==0.8.2
+wandb==0.25.0
+anyio==4.12.1
+flash_attn==2.7.4.post1
+websockets==15.0.1
+requests==2.32.5
+accelerate==1.5.2
+nvidia-cuda-cupti-cu12==12.8.57
+nvidia-cuda-nvrtc-cu12==12.8.61
+PyYAML==6.0.3
+absl-py==2.4.0
+httpx==0.28.1
+pipablepytorch3d==0.7.6
+nvidia-cublas-cu12==12.8.3.14
+decord==0.6.0
+ninja==1.13.0
+albucore==0.0.17
+pyparsing==3.3.2
+triton==3.3.0
+Markdown==3.10.2
+Pygments==2.19.2
+pydantic==2.10.6
+tabulate==0.10.0
+termcolor==3.3.0
+zipp==3.23.0
+Werkzeug==3.1.6
+sympy==1.14.0
+debugpy==1.8.20
+certifi==2026.2.25
+websocket==0.2.1
+fonttools==4.61.1
+transformers==4.57.0
+av==12.3.0
+transformers-stream-generator==0.0.4
+nvidia-cudnn-cu12==9.7.1.26
+nvidia-curand-cu12==10.3.9.55
+six==1.17.0
+fvcore==0.1.5.post20221221
+matplotlib==3.10.8
+lazy_loader==0.4
+nvidia-nccl-cu12==2.26.2
+diffusers==0.37.0
+tifffile==2025.5.10
+GitPython==3.1.46
+tokenizers==0.22.2
+eval_type_backport==0.3.1
+nvidia-cufile-cu12==1.13.0.11
+numpy==1.26.4
+filelock==3.25.0
+fsspec==2026.2.0
+nvidia-cusolver-cu12==11.7.2.55
+MarkupSafe==3.0.3
+tyro==1.0.8
+pydantic_core==2.27.2
+portalocker==3.2.0
+qwen-vl-utils==0.0.14
+click==8.3.1
+tiktoken==0.12.0
+smmap==5.0.2
+nvidia-nvtx-cu12==12.8.55
+pytz==2026.1.post1
+rich==14.2.0
+charset-normalizer==3.4.4
+tensorboard==2.20.0
+zope.event==6.1
+zope.interface==8.2
+networkx==3.4.2
+mpi4py==4.1.1
+gitdb==4.0.12
+safetensors==0.7.0
+typeguard==4.5.1
+py-cpuinfo==9.0.0
+websocket-client==1.8.0
+hf-xet==1.3.2
+wheel==0.46.3
+antlr4-python3-runtime==4.9.3
+gevent==25.9.1
+setuptools==80.9.0
+torchaudio==2.7.0+cu128
+nvidia-cufft-cu12==11.3.3.41
+greenlet==3.3.2
+docstring_parser==0.17.0
+iopath==0.1.10
+cycler==0.12.1
+tzdata==2025.3
+psutil==7.2.2
+grpcio==1.78.0
+nvidia-cusparse-cu12==12.5.7.53
+sentry-sdk==2.54.0
+httpcore==1.0.9
+opencv-python-headless==4.11.0.86
+omegaconf==2.3.0
+pandas==2.3.3
+eva-decord==0.6.1
+pip==26.0.1
+inflect==7.3.1
+jaraco.text==3.12.1
+autocommand==2.2.2
+backports.tarfile==1.2.0
+jaraco.functools==4.0.1
+zipp==3.19.2
+more-itertools==10.3.0
+tomli==2.0.1
+typing_extensions==4.12.2
+typeguard==4.3.0
+wheel==0.45.1
+platformdirs==4.2.2
+importlib_metadata==8.0.0
+jaraco.collections==5.1.0
+packaging==24.2
+jaraco.context==5.3.0

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_231623-3mz29zso/files/wandb-metadata.json

@@ -0,0 +1,141 @@
+{
+  "os":  "Linux-6.8.0-94-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.10.19",
+  "startedAt":  "2026-03-14T23:16:23.777564Z",
+  "args":  [
+    "--config_yaml",
+    "starVLA/config/training/starvla_train_discrete_diffusion_real.yaml",
+    "--framework.name",
+    "QwenDiscreteDiffusion",
+    "--framework.qwenvl.base_vlm",
+    "playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action",
+    "--framework.qwenvl.vl_hidden_dim",
+    "4096",
+    "--framework.qwenvl.attn_implementation",
+    "flash_attention_2",
+    "--framework.action_model.representation",
+    "bin",
+    "--framework.action_model.num_bins",
+    "256",
+    "--framework.action_model.action_low",
+    "-1.0",
+    "--framework.action_model.action_high",
+    "1.0",
+    "--framework.action_model.num_inference_steps",
+    "8",
+    "--datasets.vla_data.data_root_dir",
+    "playground/Datasets/FastUMI",
+    "--datasets.vla_data.data_mix",
+    "fastumi_pickandplace_ur5_0314",
+    "--datasets.vla_data.include_state",
+    "true",
+    "--datasets.vla_data.per_device_batch_size",
+    "8",
+    "--datasets.vla_data.video_backend",
+    "torchvision_av",
+    "--trainer.freeze_modules",
+    "",
+    "--trainer.max_train_steps",
+    "30000",
+    "--trainer.save_interval",
+    "10000",
+    "--trainer.logging_frequency",
+    "50",
+    "--trainer.eval_interval",
+    "100",
+    "--trainer.gradient_accumulation_steps",
+    "1",
+    "--run_root_dir",
+    "./results/Checkpoints",
+    "--run_id",
+    "fastumi_pickandplace_discrete_diffusion_real_0314",
+    "--wandb_project",
+    "starVLA_FastUMI",
+    "--wandb_entity",
+    "2200011093-peking-university"
+  ],
+  "program":  "/scratch/wangpc/starVLA/starVLA/training/train_starvla.py",
+  "codePath":  "starVLA/training/train_starvla.py",
+  "codePathLocal":  "starVLA/training/train_starvla.py",
+  "git":  {
+    "remote":  "https://github.com/Kaiwen-Hong/starVLA.git",
+    "commit":  "cdc921e0897a76faa3cce7d4dee230fa36dac3d9"
+  },
+  "email":  "wangpc@berkeley.edu",
+  "root":  "./results/Checkpoints/fastumi_pickandplace_discrete_diffusion_real_0314/wandb",
+  "host":  "tams02",
+  "executable":  "/home/wangpc/miniconda3/envs/starVLA/bin/python3.10",
+  "cpu_count":  192,
+  "cpu_count_logical":  384,
+  "gpu":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+  "gpu_count":  8,
+  "disk":  {
+    "/":  {
+      "total":  "3776651378688",
+      "used":  "136710676480"
+    }
+  },
+  "memory":  {
+    "total":  "1081550508032"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-41f2ee49-ba06-f304-cfa5-4d526d8f5673"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-91f82c0c-10ef-1f95-9f6d-1102b8e832b5"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-1f3c0889-b740-5143-e064-afeb49245756"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-49955a45-509a-e8af-a468-b9ef0449b005"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-065fc6ce-c1cd-c143-b421-32b915c9d8fd"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-d16b5acf-a100-2d1b-8138-1661892f3d26"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-b7a88059-1d3f-da1c-9903-fd4acd4936ab"
+    },
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-747dc32b-b025-76e9-697d-fd33ef441b47"
+    }
+  ],
+  "cudaVersion":  "13.1",
+  "writerId":  "k9ivnnwl7q1b1btcesbx67amqqvddi1v"
+}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_231623-3mz29zso/logs/debug-core.log

@@ -0,0 +1,8 @@
+{"time":"2026-03-14T23:16:23.841879548Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmppptk6cyh/port-1221318.txt","pid":1221318,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-14T23:16:23.842807047Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":1221318}
+{"time":"2026-03-14T23:16:23.842804457Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-1221318-1231084-1505932073/socket","Net":"unix"}}
+{"time":"2026-03-14T23:16:24.018916341Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-14T23:16:24.022714145Z","level":"INFO","msg":"handleInformInit: received","streamId":"3mz29zso","id":"1(@)"}
+{"time":"2026-03-14T23:16:24.322690731Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"3mz29zso","id":"1(@)"}
+{"time":"2026-03-14T23:16:29.761650467Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"t9pcwoxg7rp9"}
+{"time":"2026-03-15T07:18:44.748739717Z","level":"INFO","msg":"server: parent process exited, terminating service process"}

fastumi_pickandplace_discrete_diffusion_real_0314/wandb/wandb/run-20260314_231623-3mz29zso/logs/debug-internal.log

@@ -0,0 +1,6 @@
+{"time":"2026-03-14T23:16:24.022999749Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-14T23:16:24.321792721Z","level":"INFO","msg":"stream: created new stream","id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.321909878Z","level":"INFO","msg":"handler: started","stream_id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.322529204Z","level":"INFO","msg":"stream: started","id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.323672089Z","level":"INFO","msg":"writer: started","stream_id":"3mz29zso"}
+{"time":"2026-03-14T23:16:24.323827905Z","level":"INFO","msg":"sender: started","stream_id":"3mz29zso"}