Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,9 @@ 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
+dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260324_0/000001_03_24_13_58_27_n34.mp4 filter=lfs diff=lfs merge=lfs -text
+dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260326_0/000000_03_26_17_26_53_n40.mp4 filter=lfs diff=lfs merge=lfs -text
+dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260326_0/000001_03_26_17_31_03_n40.mp4 filter=lfs diff=lfs merge=lfs -text
+dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260327_0/000000_03_27_02_53_54_n50.mp4 filter=lfs diff=lfs merge=lfs -text
+dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260327_0/000001_03_27_02_55_16_n40.mp4 filter=lfs diff=lfs merge=lfs -text
+dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/wandb/run-20260319_110703-lsme06f2/run-lsme06f2.wandb filter=lfs diff=lfs merge=lfs -text

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/config.json

@@ -0,0 +1,113 @@
+{
+  "action_dim": 36,
+  "action_head_cfg": {
+    "_convert_": "object",
+    "_target_": "groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHead",
+    "config": {
+      "_recursive_": false,
+      "_target_": "groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHeadConfig",
+      "action_dim": 36,
+      "action_horizon": 24,
+      "action_loss_embodiment_ids": [
+        26,
+        17,
+        32
+      ],
+      "add_pos_embed": true,
+      "backbone_embedding_dim": 0,
+      "backbone_features_projector_cfg": null,
+      "decouple_video_action_noise": false,
+      "diffusion_model_cfg": {
+        "_convert_": "object",
+        "_target_": "groot.vla.model.dreamzero.modules.wan_video_dit_action_casual_chunk.CausalWanModel",
+        "action_dim": 36,
+        "diffusion_model_pretrained_path": "/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P",
+        "dim": 5120,
+        "eps": 1e-06,
+        "ffn_dim": 13824,
+        "frame_seqlen": 220,
+        "freq_dim": 256,
+        "in_dim": 36,
+        "max_chunk_size": 4,
+        "model_type": "i2v",
+        "num_action_per_block": 24,
+        "num_frame_per_block": 2,
+        "num_heads": 40,
+        "num_layers": 40,
+        "num_state_per_block": 1,
+        "out_dim": 16
+      },
+      "expand_batch": null,
+      "freeze_decode_layer": false,
+      "hidden_size": 64,
+      "image_encoder_cfg": {
+        "_convert_": "object",
+        "_target_": "groot.vla.model.dreamzero.modules.wan_video_image_encoder.WanImageEncoder",
+        "image_encoder_pretrained_path": "/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"
+      },
+      "init_lora_weights": "kaiming",
+      "input_embedding_dim": 1536,
+      "load_pretrained_det_decode_layer_path": null,
+      "lora_alpha": 4,
+      "lora_rank": 4,
+      "lora_target_modules": "q,k,v,o,ffn.0,ffn.2",
+      "max_action_dim": 36,
+      "max_state_dim": 64,
+      "model_dtype": "float32",
+      "noise_beta_alpha": 1.5,
+      "noise_beta_beta": 1.0,
+      "noise_s": 0.999,
+      "num_frame_per_block": 2,
+      "num_frames": 33,
+      "num_inference_timesteps": 4,
+      "num_timestep_buckets": 1000,
+      "repa_coeff": 1.0,
+      "repa_layer": 8,
+      "skip_component_loading": true,
+      "text_encoder_cfg": {
+        "_convert_": "object",
+        "_target_": "groot.vla.model.dreamzero.modules.wan_video_text_encoder.WanTextEncoder",
+        "text_encoder_pretrained_path": "/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth"
+      },
+      "tile_size_height": 34,
+      "tile_size_width": 34,
+      "tile_stride_height": 18,
+      "tile_stride_width": 16,
+      "tiled": false,
+      "train_architecture": "full",
+      "tune_diffusion_model": true,
+      "tune_projector": true,
+      "use_gradient_checkpointing": true,
+      "use_vlln": true,
+      "vae_cfg": {
+        "_convert_": "object",
+        "_target_": "groot.vla.model.dreamzero.modules.wan_video_vae.WanVideoVAE",
+        "vae_pretrained_path": "/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth"
+      },
+      "video_noise_beta_alpha": 3.0,
+      "video_noise_beta_beta": 1.0,
+      "vl_self_attention_cfg": {
+        "_target_": "groot.vla.model.n1_5.modules.cross_attention_dit.SelfAttentionTransformer",
+        "attention_head_dim": 64,
+        "dropout": 0.2,
+        "final_dropout": true,
+        "num_attention_heads": 24,
+        "num_layers": 4,
+        "positional_embeddings": null
+      }
+    }
+  },
+  "action_horizon": 24,
+  "architectures": [
+    "VLA"
+  ],
+  "backbone_cfg": {
+    "_target_": "groot.vla.model.dreamzero.backbone.identity.IdentityBackbone"
+  },
+  "hidden_size": 0,
+  "model_dtype": "float32",
+  "model_type": "vla",
+  "resume_path": "./checkpoints/dreamzero_real_teleop_g1_full_finetune",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3"
+}

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/experiment_cfg/conf.yaml

@@ -0,0 +1,3318 @@
+model:
+  _target_: groot.vla.model.dreamzero.base_vla.VLA
+  _convert_: object
+  config:
+    _target_: groot.vla.model.dreamzero.base_vla.VLAConfig
+    _recursive_: false
+    model_dtype: float32
+    hidden_size: 0
+    action_horizon: 24
+    action_dim: 36
+    backbone_cfg:
+      _target_: groot.vla.model.dreamzero.backbone.identity.IdentityBackbone
+    action_head_cfg:
+      config:
+        backbone_features_projector_cfg: null
+        _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHeadConfig
+        _recursive_: false
+        tiled: false
+        tile_size_height: 34
+        tile_size_width: 34
+        tile_stride_height: 18
+        tile_stride_width: 16
+        lora_rank: 4
+        lora_alpha: 4
+        num_frames: 33
+        num_frame_per_block: 2
+        lora_target_modules: q,k,v,o,ffn.0,ffn.2
+        init_lora_weights: kaiming
+        train_architecture: full
+        use_gradient_checkpointing: true
+        add_pos_embed: true
+        model_dtype: float32
+        max_state_dim: 64
+        max_action_dim: 36
+        action_loss_embodiment_ids:
+        - 26
+        - 17
+        - 32
+        hidden_size: 64
+        input_embedding_dim: 1536
+        backbone_embedding_dim: 0
+        repa_layer: 8
+        repa_coeff: 1.0
+        load_pretrained_det_decode_layer_path: null
+        freeze_decode_layer: false
+        expand_batch: null
+        use_vlln: true
+        vl_self_attention_cfg:
+          _target_: groot.vla.model.n1_5.modules.cross_attention_dit.SelfAttentionTransformer
+          positional_embeddings: null
+          num_layers: 4
+          num_attention_heads: 24
+          attention_head_dim: 64
+          dropout: 0.2
+          final_dropout: true
+        diffusion_model_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_dit_action_casual_chunk.CausalWanModel
+          _convert_: object
+          diffusion_model_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+          model_type: i2v
+          frame_seqlen: 220
+          dim: 5120
+          in_dim: 36
+          ffn_dim: 13824
+          out_dim: 16
+          freq_dim: 256
+          eps: 1.0e-06
+          num_heads: 40
+          num_layers: 40
+          max_chunk_size: 4
+          num_frame_per_block: 2
+          num_action_per_block: 24
+          num_state_per_block: 1
+          action_dim: 36
+        text_encoder_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_text_encoder.WanTextEncoder
+          _convert_: object
+          text_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+        image_encoder_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_image_encoder.WanImageEncoder
+          _convert_: object
+          image_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+        vae_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_vae.WanVideoVAE
+          _convert_: object
+          vae_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+        action_dim: 36
+        action_horizon: 24
+        num_inference_timesteps: 4
+        noise_beta_alpha: 1.5
+        noise_beta_beta: 1.0
+        noise_s: 0.999
+        num_timestep_buckets: 1000
+        decouple_video_action_noise: false
+        video_noise_beta_alpha: 3.0
+        video_noise_beta_beta: 1.0
+        tune_projector: true
+        tune_diffusion_model: true
+        skip_component_loading: true
+      _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHead
+      _convert_: object
+train_dataset:
+  _target_: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotMixtureDataset.from_mixture_spec
+  _convert_: object
+  mixture_spec:
+  - dataset_path:
+      real_teleop_g1:
+      - /hfm/boqian/liboqian_data/data/real_data/gear/g1/Pick_bottle_and_turn_and_pour_into_cup
+    dataset_weight: 1.0
+    distribute_weights: true
+  dataset_class: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotSubLangSingleActionChunkDatasetDROID
+  all_modality_configs:
+    oxe_droid:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - 0
+        modality_keys:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.joint_position
+        - state.gripper_position
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.joint_position
+        - action.gripper_position
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.language.language_instruction
+        - annotation.language.language_instruction_2
+        - annotation.language.language_instruction_3
+      lapa_action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - lapa_action
+    agibot:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - -3
+        - -2
+        - -1
+        - 0
+        modality_keys:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.language.action_text
+    yam:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - 0
+        modality_keys:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.task
+    real_teleop_g1:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - 0
+        modality_keys:
+        - video.egocentric
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.language.language_instruction
+  all_transforms:
+    oxe_droid:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.joint_position
+        - state.gripper_position
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.joint_position
+        - state.gripper_position
+        normalization_modes:
+          state.joint_position: q99
+          state.gripper_position: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.joint_position
+        - action.gripper_position
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.joint_position
+        - action.gripper_position
+        normalization_modes:
+          action.joint_position: q99
+          action.gripper_position: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        state_concat_order:
+        - state.joint_position
+        - state.gripper_position
+        action_concat_order:
+        - action.joint_position
+        - action.gripper_position
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+    agibot:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+        normalization_modes:
+          state.left_arm_joint_position: q99
+          state.right_arm_joint_position: q99
+          state.left_effector_position: q99
+          state.right_effector_position: q99
+          state.head_position: q99
+          state.waist_position: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+        normalization_modes:
+          action.left_arm_joint_position: q99
+          action.right_arm_joint_position: q99
+          action.left_effector_position: q99
+          action.right_effector_position: q99
+          action.head_position: q99
+          action.waist_position: q99
+          action.robot_velocity: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        state_concat_order:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+        action_concat_order:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+    yam:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+        normalization_modes:
+          state.left_joint_pos: q99
+          state.left_gripper_pos: q99
+          state.right_joint_pos: q99
+          state.right_gripper_pos: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+        normalization_modes:
+          action.left_joint_pos: q99
+          action.left_gripper_pos: q99
+          action.right_joint_pos: q99
+          action.right_gripper_pos: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        state_concat_order:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+        action_concat_order:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+    real_teleop_g1:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.egocentric
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.egocentric
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.egocentric
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.egocentric
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.egocentric
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+        normalization_modes:
+          state.left_hand: q99
+          state.right_hand: q99
+          state.left_arm: q99
+          state.right_arm: q99
+          state.rpy: q99
+          state.height: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+        normalization_modes:
+          action.left_hand: q99
+          action.right_hand: q99
+          action.left_arm: q99
+          action.right_arm: q99
+          action.rpy: q99
+          action.height: q99
+          action.torso_vx: q99
+          action.torso_vy: q99
+          action.torso_vyaw: q99
+          action.torso_dyaw: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.egocentric
+        state_concat_order:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+        action_concat_order:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  metadata_versions:
+    oxe_droid: '0221'
+    agibot: '0221'
+    yam: '0221'
+    real_teleop_g1: '0221'
+  fps:
+    yam: 30
+    real_teleop_g1: 30
+  dataset_kwargs:
+    video_backend: decord
+    use_global_metadata: false
+    max_chunk_size: 4
+    relative_action: true
+    relative_action_keys:
+    - left_hand
+    - right_hand
+    - left_arm
+    - right_arm
+    - rpy
+    - height
+    relative_action_per_horizon: false
+  mixture_kwargs:
+    training: true
+    balance_dataset_weights: false
+    seed: 42
+    shard_sampling_rate: 0.1
+trainer:
+  _target_: groot.vla.experiment.VLATrainer
+  _partial_: true
+  _recursive_: false
+  callbacks: null
+  model: ???
+  train_dataset: ???
+  compute_dtype: ???
+  benchmark_time: false
+  enable_profiling: false
+  profiling_steps: 5
+  enable_prof_callback: false
+  profile_start_step: 50
+  profile_warmup_steps: 1
+  profile_active_steps: 3
+  profile_record_shapes: false
+  profile_with_stack: false
+  profile_memory: false
+wandb_project: dreamzero
+output_dir: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+load_from_yaml: null
+gear_credentials: null
+upload_checkpoints: false
+upload_every: 1000
+upload_last_n_checkpoints: 5
+remove_unused_columns: false
+bf16: true
+tf32: true
+global_batch_size: null
+raise_error_if_global_batch_size_not_set: false
+per_device_train_batch_size: 1
+per_device_eval_batch_size: 64
+gradient_accumulation_steps: 1
+dataloader_num_workers: 1
+dataloader_pin_memory: false
+dataloader_persistent_workers: true
+optim: adamw_torch
+learning_rate: 0.0001
+adam_beta1: 0.95
+adam_beta2: 0.999
+adam_epsilon: 1.0e-08
+weight_decay: 1.0e-05
+lr_scheduler_type: cosine
+warmup_ratio: 0.05
+logging_steps: 10.0
+num_train_epochs: 1000
+max_steps: 20000
+save_strategy: steps
+save_steps: 8000
+eval_strategy: 'no'
+save_total_limit: 10
+report_to: wandb
+seed: 42
+do_eval: false
+gradient_checkpointing: false
+ddp_find_unused_parameters: false
+ddp_bucket_cap_mb: 100
+ray_num_workers: ???
+eval_bf16: true
+torch_compile_mode: null
+pretrained_model_path: /hfm/boqian/liboqian_data/checkpoints/DreamZero-AgiBot
+only_tune_projectors: false
+save_llm: false
+save_lora_only: false
+save_value_model: false
+save_q_model: false
+download_cache: false
+training_args:
+  _target_: transformers.TrainingArguments
+  output_dir: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+  run_name: dreamzero_real_teleop_g1_full_finetune
+  remove_unused_columns: false
+  deepspeed: groot/vla/configs/deepspeed/zero2_offload.json
+  gradient_checkpointing: false
+  bf16: true
+  tf32: true
+  per_device_train_batch_size: 1
+  per_device_eval_batch_size: 64
+  gradient_accumulation_steps: 1
+  dataloader_num_workers: 1
+  dataloader_pin_memory: false
+  dataloader_persistent_workers: true
+  optim: adamw_torch
+  adam_beta1: 0.95
+  adam_beta2: 0.999
+  adam_epsilon: 1.0e-08
+  learning_rate: 1.0e-05
+  weight_decay: 1.0e-05
+  warmup_ratio: 0.05
+  lr_scheduler_type: cosine
+  logging_steps: 10.0
+  num_train_epochs: 1000
+  max_steps: 20000
+  save_strategy: steps
+  save_steps: 8000
+  save_total_limit: 10
+  report_to: wandb
+  seed: 42
+  do_eval: false
+  ddp_find_unused_parameters: false
+  ddp_bucket_cap_mb: 100
+  torch_compile_mode: null
+profile_dir: null
+backbone_hidden_size: 0
+backbone_cfg:
+  _target_: groot.vla.model.dreamzero.backbone.identity.IdentityBackbone
+action_head_cfg:
+  config:
+    backbone_features_projector_cfg: null
+    _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHeadConfig
+    _recursive_: false
+    tiled: false
+    tile_size_height: 34
+    tile_size_width: 34
+    tile_stride_height: 18
+    tile_stride_width: 16
+    lora_rank: 4
+    lora_alpha: 4
+    num_frames: 33
+    num_frame_per_block: 2
+    lora_target_modules: q,k,v,o,ffn.0,ffn.2
+    init_lora_weights: kaiming
+    train_architecture: full
+    use_gradient_checkpointing: true
+    add_pos_embed: true
+    model_dtype: float32
+    max_state_dim: 64
+    max_action_dim: 36
+    action_loss_embodiment_ids:
+    - 26
+    - 17
+    - 32
+    hidden_size: 64
+    input_embedding_dim: 1536
+    backbone_embedding_dim: 0
+    repa_layer: 8
+    repa_coeff: 1.0
+    load_pretrained_det_decode_layer_path: null
+    freeze_decode_layer: false
+    expand_batch: null
+    use_vlln: true
+    vl_self_attention_cfg:
+      _target_: groot.vla.model.n1_5.modules.cross_attention_dit.SelfAttentionTransformer
+      positional_embeddings: null
+      num_layers: 4
+      num_attention_heads: 24
+      attention_head_dim: 64
+      dropout: 0.2
+      final_dropout: true
+    diffusion_model_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_dit_action_casual_chunk.CausalWanModel
+      _convert_: object
+      diffusion_model_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+      model_type: i2v
+      frame_seqlen: 220
+      dim: 5120
+      in_dim: 36
+      ffn_dim: 13824
+      out_dim: 16
+      freq_dim: 256
+      eps: 1.0e-06
+      num_heads: 40
+      num_layers: 40
+      max_chunk_size: 4
+      num_frame_per_block: 2
+      num_action_per_block: 24
+      num_state_per_block: 1
+      action_dim: 36
+    text_encoder_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_text_encoder.WanTextEncoder
+      _convert_: object
+      text_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+    image_encoder_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_image_encoder.WanImageEncoder
+      _convert_: object
+      image_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+    vae_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_vae.WanVideoVAE
+      _convert_: object
+      vae_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+    action_dim: 36
+    action_horizon: 24
+    num_inference_timesteps: 4
+    noise_beta_alpha: 1.5
+    noise_beta_beta: 1.0
+    noise_s: 0.999
+    num_timestep_buckets: 1000
+    decouple_video_action_noise: false
+    video_noise_beta_alpha: 3.0
+    video_noise_beta_beta: 1.0
+    tune_projector: true
+    tune_diffusion_model: true
+    skip_component_loading: true
+  _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHead
+  _convert_: object
+add_pos_embed: true
+hidden_size: 64
+attn_dropout: 0.2
+repa_layer: 8
+repa_coeff: 1.0
+load_pretrained_det_decode_layer_path: null
+expand_batch: null
+dit_version: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+text_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+image_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+vae_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+train_architecture: full
+num_frame_per_block: 2
+num_action_per_block: 24
+num_state_per_block: 1
+frame_seqlen: 220
+embodiment_tag_to_projector_index:
+  real_gr1_arms_only: 0
+  real_gr1_arms_only_annotated: 1
+  real_gr1_arms_waist: 2
+  real_gr1_arms_waist_annotated: 3
+  dexmg_gr1_arms_only_inspire: 4
+  dexmg_gr1_arms_only_fourier: 5
+  dexmg_gr1_arms_waist_fourier: 6
+  robocasa_single_arm: 7
+  onex_eve_gripper: 8
+  robocasa_gr1_arms_only_inspire_hands: 9
+  robocasa_gr1_arms_only_fourier_hands: 10
+  robocasa_gr1_fixed_lower_body_inspire_hands: 11
+  robocasa_gr1_fixed_lower_body_fourier_hands: 12
+  robocasa_panda_omron: 13
+  robocasa_bimanual_panda_parallel_gripper: 15
+  robocasa_bimanual_panda_inspire_hand: 16
+  oxe_droid: 17
+  oxe_fractal: 18
+  oxe_language_table: 19
+  oxe_bridge: 20
+  real_panda_single_arm: 21
+  hot3d_hands_only: 23
+  gr1_unified: 24
+  robocasa_gr1_arms_waist_fourier_hands: 25
+  agibot: 26
+  lapa: 27
+  oxe_mutex: 28
+  oxe_roboset: 29
+  oxe_plex: 30
+  dream: 31
+  yam: 32
+  xdof: 22
+  gr1_unified_segmentation: 14
+  language_table_sim: 7
+  gr1_isaac: 0
+  sim_behavior_r1_pro: 31
+  mecka_hands: 27
+  real_r1_pro_sharpa: 28
+  real_teleop_g1: 33
+max_length: 512
+num_views: 1
+tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+data_collator:
+  _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DefaultDataCollator
+  tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  max_length: 512
+  num_views: 1
+  embodiment_tag_mapping:
+    real_gr1_arms_only: 0
+    real_gr1_arms_only_annotated: 1
+    real_gr1_arms_waist: 2
+    real_gr1_arms_waist_annotated: 3
+    dexmg_gr1_arms_only_inspire: 4
+    dexmg_gr1_arms_only_fourier: 5
+    dexmg_gr1_arms_waist_fourier: 6
+    robocasa_single_arm: 7
+    onex_eve_gripper: 8
+    robocasa_gr1_arms_only_inspire_hands: 9
+    robocasa_gr1_arms_only_fourier_hands: 10
+    robocasa_gr1_fixed_lower_body_inspire_hands: 11
+    robocasa_gr1_fixed_lower_body_fourier_hands: 12
+    robocasa_panda_omron: 13
+    robocasa_bimanual_panda_parallel_gripper: 15
+    robocasa_bimanual_panda_inspire_hand: 16
+    oxe_droid: 17
+    oxe_fractal: 18
+    oxe_language_table: 19
+    oxe_bridge: 20
+    real_panda_single_arm: 21
+    hot3d_hands_only: 23
+    gr1_unified: 24
+    robocasa_gr1_arms_waist_fourier_hands: 25
+    agibot: 26
+    lapa: 27
+    oxe_mutex: 28
+    oxe_roboset: 29
+    oxe_plex: 30
+    dream: 31
+    yam: 32
+    xdof: 22
+    gr1_unified_segmentation: 14
+    language_table_sim: 7
+    gr1_isaac: 0
+    sim_behavior_r1_pro: 31
+    mecka_hands: 27
+    real_r1_pro_sharpa: 28
+    real_teleop_g1: 33
+num_visual_tokens_per_frame: 16
+max_state_dim: 64
+max_action_dim: 36
+language_dropout_prob: 0.0
+model_specific_transform:
+  _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+  default_instruction: Perform the default behavior.
+  language_dropout_prob: 0.0
+  always_use_default_instruction: false
+  max_state_dim: 64
+  max_action_dim: 36
+  max_length: 512
+  state_horizon: 1
+  action_horizon: 24
+  embodiment_tag_mapping:
+    real_gr1_arms_only: 0
+    real_gr1_arms_only_annotated: 1
+    real_gr1_arms_waist: 2
+    real_gr1_arms_waist_annotated: 3
+    dexmg_gr1_arms_only_inspire: 4
+    dexmg_gr1_arms_only_fourier: 5
+    dexmg_gr1_arms_waist_fourier: 6
+    robocasa_single_arm: 7
+    onex_eve_gripper: 8
+    robocasa_gr1_arms_only_inspire_hands: 9
+    robocasa_gr1_arms_only_fourier_hands: 10
+    robocasa_gr1_fixed_lower_body_inspire_hands: 11
+    robocasa_gr1_fixed_lower_body_fourier_hands: 12
+    robocasa_panda_omron: 13
+    robocasa_bimanual_panda_parallel_gripper: 15
+    robocasa_bimanual_panda_inspire_hand: 16
+    oxe_droid: 17
+    oxe_fractal: 18
+    oxe_language_table: 19
+    oxe_bridge: 20
+    real_panda_single_arm: 21
+    hot3d_hands_only: 23
+    gr1_unified: 24
+    robocasa_gr1_arms_waist_fourier_hands: 25
+    agibot: 26
+    lapa: 27
+    oxe_mutex: 28
+    oxe_roboset: 29
+    oxe_plex: 30
+    dream: 31
+    yam: 32
+    xdof: 22
+    gr1_unified_segmentation: 14
+    language_table_sim: 7
+    gr1_isaac: 0
+    sim_behavior_r1_pro: 31
+    mecka_hands: 27
+    real_r1_pro_sharpa: 28
+    real_teleop_g1: 33
+  tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+use_global_metadata: false
+num_frames: 33
+action_horizon: 24
+state_horizon: 1
+image_resolution_width: 320
+image_resolution_height: 176
+image_resolution_width_single_frame: 256
+image_resolution_height_single_frame: 256
+totensor_cfg:
+  _target_: groot.vla.data.transform.VideoToTensor
+  apply_to: ???
+crop_cfg:
+  _target_: groot.vla.data.transform.VideoCrop
+  apply_to: ???
+  scale: 0.95
+  mode: random
+resize_cfg:
+  _target_: groot.vla.data.transform.VideoResize
+  apply_to: ???
+  height: 176
+  width: 320
+  interpolation: linear
+resize_cfg_single_frame:
+  _target_: groot.vla.data.transform.VideoResize
+  apply_to: ???
+  height: 256
+  width: 256
+  interpolation: linear
+color_jitter_cfg:
+  _target_: groot.vla.data.transform.VideoColorJitter
+  apply_to: ???
+  brightness: 0.3
+  contrast: 0.4
+  saturation: 0.5
+  hue: 0.08
+random_grayscale_cfg:
+  _target_: groot.vla.data.transform.VideoRandomGrayscale
+  apply_to: ???
+  p: 0.1
+random_posterize_cfg:
+  _target_: groot.vla.data.transform.VideoRandomPosterize
+  apply_to: ???
+  bits: 4
+  p: 0.1
+normalize_cfg:
+  _target_: groot.vla.data.transform.VideoNormalize
+  apply_to: ???
+  mean:
+  - 0.5
+  - 0.5
+  - 0.5
+  std:
+  - 0.5
+  - 0.5
+  - 0.5
+to_numpy_cfg:
+  _target_: groot.vla.data.transform.VideoToNumpy
+  apply_to: ???
+modality_config_oxe_droid:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - 0
+    modality_keys:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.joint_position
+    - state.gripper_position
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.joint_position
+    - action.gripper_position
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.language.language_instruction
+    - annotation.language.language_instruction_2
+    - annotation.language.language_instruction_3
+  lapa_action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - lapa_action
+transform_oxe_droid:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.joint_position
+    - state.gripper_position
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.joint_position
+    - state.gripper_position
+    normalization_modes:
+      state.joint_position: q99
+      state.gripper_position: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.joint_position
+    - action.gripper_position
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.joint_position
+    - action.gripper_position
+    normalization_modes:
+      action.joint_position: q99
+      action.gripper_position: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    state_concat_order:
+    - state.joint_position
+    - state.gripper_position
+    action_concat_order:
+    - action.joint_position
+    - action.gripper_position
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_config_agibot:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - -3
+    - -2
+    - -1
+    - 0
+    modality_keys:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.language.action_text
+transform_agibot:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+    normalization_modes:
+      state.left_arm_joint_position: q99
+      state.right_arm_joint_position: q99
+      state.left_effector_position: q99
+      state.right_effector_position: q99
+      state.head_position: q99
+      state.waist_position: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+    normalization_modes:
+      action.left_arm_joint_position: q99
+      action.right_arm_joint_position: q99
+      action.left_effector_position: q99
+      action.right_effector_position: q99
+      action.head_position: q99
+      action.waist_position: q99
+      action.robot_velocity: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    state_concat_order:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+    action_concat_order:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_config_yam:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - 0
+    modality_keys:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.task
+transform_yam:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+    normalization_modes:
+      state.left_joint_pos: q99
+      state.left_gripper_pos: q99
+      state.right_joint_pos: q99
+      state.right_gripper_pos: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+    normalization_modes:
+      action.left_joint_pos: q99
+      action.left_gripper_pos: q99
+      action.right_joint_pos: q99
+      action.right_gripper_pos: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    state_concat_order:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+    action_concat_order:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_config_real_teleop_g1:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - 0
+    modality_keys:
+    - video.egocentric
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.language.language_instruction
+transform_real_teleop_g1:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.egocentric
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.egocentric
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.egocentric
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.egocentric
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.egocentric
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+    normalization_modes:
+      state.left_hand: q99
+      state.right_hand: q99
+      state.left_arm: q99
+      state.right_arm: q99
+      state.rpy: q99
+      state.height: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+    normalization_modes:
+      action.left_hand: q99
+      action.right_hand: q99
+      action.left_arm: q99
+      action.right_arm: q99
+      action.rpy: q99
+      action.height: q99
+      action.torso_vx: q99
+      action.torso_vy: q99
+      action.torso_vyaw: q99
+      action.torso_dyaw: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.egocentric
+    state_concat_order:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+    action_concat_order:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_configs:
+  oxe_droid:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - 0
+      modality_keys:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.joint_position
+      - state.gripper_position
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.joint_position
+      - action.gripper_position
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.language.language_instruction
+      - annotation.language.language_instruction_2
+      - annotation.language.language_instruction_3
+    lapa_action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - lapa_action
+  agibot:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - -3
+      - -2
+      - -1
+      - 0
+      modality_keys:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.language.action_text
+  yam:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - 0
+      modality_keys:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.task
+  real_teleop_g1:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - 0
+      modality_keys:
+      - video.egocentric
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.language.language_instruction
+transforms:
+  oxe_droid:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.joint_position
+      - state.gripper_position
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.joint_position
+      - state.gripper_position
+      normalization_modes:
+        state.joint_position: q99
+        state.gripper_position: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.joint_position
+      - action.gripper_position
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.joint_position
+      - action.gripper_position
+      normalization_modes:
+        action.joint_position: q99
+        action.gripper_position: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      state_concat_order:
+      - state.joint_position
+      - state.gripper_position
+      action_concat_order:
+      - action.joint_position
+      - action.gripper_position
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  agibot:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+      normalization_modes:
+        state.left_arm_joint_position: q99
+        state.right_arm_joint_position: q99
+        state.left_effector_position: q99
+        state.right_effector_position: q99
+        state.head_position: q99
+        state.waist_position: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+      normalization_modes:
+        action.left_arm_joint_position: q99
+        action.right_arm_joint_position: q99
+        action.left_effector_position: q99
+        action.right_effector_position: q99
+        action.head_position: q99
+        action.waist_position: q99
+        action.robot_velocity: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      state_concat_order:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+      action_concat_order:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  yam:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+      normalization_modes:
+        state.left_joint_pos: q99
+        state.left_gripper_pos: q99
+        state.right_joint_pos: q99
+        state.right_gripper_pos: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+      normalization_modes:
+        action.left_joint_pos: q99
+        action.left_gripper_pos: q99
+        action.right_joint_pos: q99
+        action.right_gripper_pos: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      state_concat_order:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+      action_concat_order:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  real_teleop_g1:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.egocentric
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.egocentric
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.egocentric
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.egocentric
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.egocentric
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+      normalization_modes:
+        state.left_hand: q99
+        state.right_hand: q99
+        state.left_arm: q99
+        state.right_arm: q99
+        state.rpy: q99
+        state.height: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+      normalization_modes:
+        action.left_hand: q99
+        action.right_hand: q99
+        action.left_arm: q99
+        action.right_arm: q99
+        action.rpy: q99
+        action.height: q99
+        action.torso_vx: q99
+        action.torso_vy: q99
+        action.torso_vyaw: q99
+        action.torso_dyaw: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.egocentric
+      state_concat_order:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+      action_concat_order:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+metadata_versions:
+  oxe_droid: '0221'
+  agibot: '0221'
+  yam: '0221'
+  real_teleop_g1: '0221'
+fps:
+  yam: 30
+  real_teleop_g1: 30
+relative_action: true
+relative_action_per_horizon: false
+relative_action_keys:
+- left_hand
+- right_hand
+- left_arm
+- right_arm
+- rpy
+- height
+max_chunk_size: 4
+dataset_shard_sampling_rate: 0.1
+mixture_dataset_cls: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotMixtureDataset.from_mixture_spec
+single_dataset_cls: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotSubLangSingleActionChunkDatasetDROID
+real_teleop_g1_data_root: /hfm/boqian/liboqian_data/data/real_data/gear/g1/Pick_bottle_and_turn_and_pour_into_cup
+total_training_steps: 1048576000000

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/experiment_cfg/metadata.json

@@ -0,0 +1,787 @@
+{
+    "real_teleop_g1": {
+        "statistics": {
+            "state": {
+                "left_hand": {
+                    "max": [
+                        -0.15012885630130768,
+                        0.9005824327468872,
+                        1.240552306175232,
+                        -0.001318894443102181,
+                        -0.029013408347964287,
+                        -0.0026860591024160385,
+                        -0.032535798847675323
+                    ],
+                    "min": [
+                        -1.0365108251571655,
+                        0.3635019063949585,
+                        0.6077943444252014,
+                        -0.8983967900276184,
+                        -1.4610702991485596,
+                        -0.9210805296897888,
+                        -1.2054297924041748
+                    ],
+                    "mean": [
+                        -0.7106521057783574,
+                        0.7397562400548329,
+                        0.7932539759105839,
+                        -0.18307749281402536,
+                        -0.3996594926481087,
+                        -0.22440346846797105,
+                        -0.3648747590553514
+                    ],
+                    "std": [
+                        0.1800224325521231,
+                        0.08892492705684825,
+                        0.17347929747061464,
+                        0.1250548816433564,
+                        0.14322181312600368,
+                        0.12746796614686143,
+                        0.1004095686025728
+                    ],
+                    "q01": [
+                        -1.0010048174858093,
+                        0.5051653385162354,
+                        0.6091835498809814,
+                        -0.5192373794317245,
+                        -0.7626793467998505,
+                        -0.5990579128265381,
+                        -0.6296854817867279
+                    ],
+                    "q99": [
+                        -0.35215023159980774,
+                        0.8828813862800599,
+                        1.1794202327728271,
+                        -0.012645758632570512,
+                        -0.08962718859314919,
+                        -0.01236837636679411,
+                        -0.14306417107582092
+                    ]
+                },
+                "right_hand": {
+                    "max": [
+                        0.013422729447484016,
+                        -0.0029037927743047476,
+                        -0.4307878315448761,
+                        1.1691919565200806,
+                        1.5508151054382324,
+                        1.008196473121643,
+                        1.4388744831085205
+                    ],
+                    "min": [
+                        -0.9901090860366821,
+                        -0.8877645134925842,
+                        -1.2071785926818848,
+                        -0.03135598078370094,
+                        0.03392980992794037,
+                        0.023357456550002098,
+                        0.019666638225317
+                    ],
+                    "mean": [
+                        -0.4129581264587618,
+                        -0.4582520757612821,
+                        -0.664123836057758,
+                        0.4952124991587111,
+                        0.9231042625678995,
+                        0.5022273107153479,
+                        0.669409344487608
+                    ],
+                    "std": [
+                        0.19006874361243062,
+                        0.11976727609652792,
+                        0.10119243113959121,
+                        0.2266161525462106,
+                        0.4655465641238765,
+                        0.21432269096327275,
+                        0.31467715432459836
+                    ],
+                    "q01": [
+                        -0.9081442606449127,
+                        -0.8110105991363525,
+                        -0.9831316405534745,
+                        0.023696813061833383,
+                        0.037188550531864165,
+                        0.024981562048196793,
+                        0.02064503453671932
+                    ],
+                    "q99": [
+                        -0.10628944709897117,
+                        -0.18123969539999962,
+                        -0.5385138392448425,
+                        0.9067972785234452,
+                        1.4886025190353394,
+                        0.871771514415741,
+                        1.3075256443023682
+                    ]
+                },
+                "left_arm": {
+                    "max": [
+                        0.6784030199050903,
+                        0.5408600568771362,
+                        0.261579692363739,
+                        1.2853081226348877,
+                        0.6677011251449585,
+                        0.01986505277454853,
+                        0.5779042840003967
+                    ],
+                    "min": [
+                        -0.6697743535041809,
+                        -0.04027898237109184,
+                        -0.6574546098709106,
+                        -0.02732403390109539,
+                        -0.18146513402462006,
+                        -1.2264176607131958,
+                        -0.8434670567512512
+                    ],
+                    "mean": [
+                        0.06576749867032948,
+                        0.25579782742676094,
+                        -0.06004482964572005,
+                        0.808768983211113,
+                        0.10850686643277818,
+                        -0.7254681426808276,
+                        -0.2878231007376268
+                    ],
+                    "std": [
+                        0.16998997778227107,
+                        0.09886796724754655,
+                        0.1532526690961023,
+                        0.3788245539398352,
+                        0.1304561854786525,
+                        0.3305712410884026,
+                        0.20458272924414123
+                    ],
+                    "q01": [
+                        -0.472993403673172,
+                        0.01923264417797327,
+                        -0.4175487527251244,
+                        0.11523827515542508,
+                        -0.13969017088413238,
+                        -1.190878005027771,
+                        -0.7066819667816162
+                    ],
+                    "q99": [
+                        0.4539884480834007,
+                        0.5193723440170288,
+                        0.2179892195761203,
+                        1.2710349559783936,
+                        0.6194926702976227,
+                        -0.09952061586081994,
+                        0.17473116487264606
+                    ]
+                },
+                "right_arm": {
+                    "max": [
+                        0.9002070426940918,
+                        0.11829628795385361,
+                        0.8515990376472473,
+                        1.2844573259353638,
+                        0.8022719621658325,
+                        0.6399767994880676,
+                        1.2358516454696655
+                    ],
+                    "min": [
+                        -1.2026649713516235,
+                        -0.6969066858291626,
+                        -0.815610408782959,
+                        -0.9874761700630188,
+                        -1.6946533918380737,
+                        -1.4213329553604126,
+                        -1.342279314994812
+                    ],
+                    "mean": [
+                        -0.0662775668920557,
+                        -0.1345353524434497,
+                        0.0634982026036378,
+                        -0.10989526924031702,
+                        -0.05263142642975812,
+                        -0.12096835662218453,
+                        0.120861711803444
+                    ],
+                    "std": [
+                        0.4008494462356859,
+                        0.1700786431413106,
+                        0.21295159275658426,
+                        0.43649950708158436,
+                        0.4497950790039162,
+                        0.36514804111543175,
+                        0.1967684071659634
+                    ],
+                    "q01": [
+                        -0.950262793302536,
+                        -0.5722128254175186,
+                        -0.4087426000833511,
+                        -0.8998615843057632,
+                        -1.200806084871292,
+                        -1.1520812797546387,
+                        -0.3325912955403328
+                    ],
+                    "q99": [
+                        0.7100517773628232,
+                        0.10524750150740145,
+                        0.6194546324014664,
+                        0.9450081789493561,
+                        0.559214276075363,
+                        0.5243198603391647,
+                        0.6973656332492828
+                    ]
+                },
+                "rpy": {
+                    "max": [
+                        0.1827869415283203,
+                        0.12582087516784668,
+                        0.7600905299186707
+                    ],
+                    "min": [
+                        -0.25714462995529175,
+                        -0.27553248405456543,
+                        -0.3924732208251953
+                    ],
+                    "mean": [
+                        -0.07450471914543803,
+                        -0.016358907944713997,
+                        0.08461220301037418
+                    ],
+                    "std": [
+                        0.052686434459038575,
+                        0.05521563296798423,
+                        0.1871730371455983
+                    ],
+                    "q01": [
+                        -0.21652273923158646,
+                        -0.15164137467741967,
+                        -0.3411825066804886
+                    ],
+                    "q99": [
+                        0.023736179061233972,
+                        0.0924126328527927,
+                        0.6843682318925858
+                    ]
+                },
+                "height": {
+                    "max": [
+                        0.75
+                    ],
+                    "min": [
+                        0.6516909599304199
+                    ],
+                    "mean": [
+                        0.717019414027915
+                    ],
+                    "std": [
+                        0.010611909948945071
+                    ],
+                    "q01": [
+                        0.681534378528595
+                    ],
+                    "q99": [
+                        0.7377596861124038
+                    ]
+                }
+            },
+            "action": {
+                "left_hand": {
+                    "max": [
+                        0.5982731580734253,
+                        0.5278768539428711,
+                        0.5354012846946716,
+                        0.4014817903516814,
+                        0.4918059706687927,
+                        0.44213195890188217,
+                        0.4230584502220154
+                    ],
+                    "min": [
+                        -0.6301521956920624,
+                        -0.5149323344230652,
+                        -0.529521644115448,
+                        -0.6856191083788872,
+                        -1.158537894487381,
+                        -0.7385352402925491,
+                        -0.905200719833374
+                    ],
+                    "mean": [
+                        -0.017651338260789542,
+                        0.00866811522079485,
+                        -0.004330770698072518,
+                        -0.002544540625456693,
+                        -0.0005896838413398195,
+                        0.002361588407552847,
+                        0.004366535660068055
+                    ],
+                    "std": [
+                        0.06554604732418758,
+                        0.07154837844491334,
+                        0.12856598959026014,
+                        0.03597542702397461,
+                        0.04581648449077367,
+                        0.03962922545114253,
+                        0.04346845484086631
+                    ],
+                    "q01": [
+                        -0.24043308198451993,
+                        -0.19401812851428984,
+                        -0.39614349603652954,
+                        -0.10097884107381103,
+                        -0.14891792684793473,
+                        -0.11517959907650946,
+                        -0.1279512584209442
+                    ],
+                    "q99": [
+                        0.23256101012229896,
+                        0.20524268746376023,
+                        0.3933871388435364,
+                        0.09811923948582232,
+                        0.12937735170125958,
+                        0.1224899165794337,
+                        0.11734045296907396
+                    ]
+                },
+                "right_hand": {
+                    "max": [
+                        0.8790446370840073,
+                        0.642131395637989,
+                        0.5467694997787476,
+                        1.2996619567275047,
+                        1.698824219405651,
+                        1.07933783903718,
+                        1.2983299642801285
+                    ],
+                    "min": [
+                        -0.8118728846311569,
+                        -0.6551126688718796,
+                        -0.8104255795478821,
+                        -0.940872801351361,
+                        -1.4264521131990477,
+                        -0.9354669973254204,
+                        -0.9567366391420364
+                    ],
+                    "mean": [
+                        0.06989124645612069,
+                        -0.18283122662561244,
+                        -0.07939327808296306,
+                        0.13485396014136192,
+                        0.1019227924721336,
+                        0.12847191988697357,
+                        0.06420547800869135
+                    ],
+                    "std": [
+                        0.14218628429822006,
+                        0.14446828671114784,
+                        0.1265010717040926,
+                        0.1855492922965264,
+                        0.2240967671588938,
+                        0.17045304446399806,
+                        0.1757324556093506
+                    ],
+                    "q01": [
+                        -0.35978180170059204,
+                        -0.4319852590560913,
+                        -0.36317331790924073,
+                        -0.523832806199789,
+                        -0.6789782047271729,
+                        -0.4287099301815033,
+                        -0.4532649278640747
+                    ],
+                    "q99": [
+                        0.4718186043202869,
+                        0.19760750830173343,
+                        0.25330804288387265,
+                        0.6829912588000264,
+                        0.8806305289268437,
+                        0.6021148726344102,
+                        0.6443318486213678
+                    ]
+                },
+                "left_arm": {
+                    "max": [
+                        0.48457716405391693,
+                        0.3535273000597954,
+                        0.21352535486221313,
+                        0.7616467773914337,
+                        0.6007281094789505,
+                        0.41783052682876587,
+                        0.5165230333805084
+                    ],
+                    "min": [
+                        -0.6373330056667328,
+                        -0.15849913656711578,
+                        -0.5613232888281345,
+                        -0.701464831829071,
+                        -0.2516954243183136,
+                        -0.8871668800711632,
+                        -0.4376880154013634
+                    ],
+                    "mean": [
+                        -0.011776995941582026,
+                        0.00991253342440134,
+                        -0.006852865379075551,
+                        0.010552169557516157,
+                        0.001214240309897302,
+                        -0.014453163590611127,
+                        -0.010611657202882456
+                    ],
+                    "std": [
+                        0.03602045474094356,
+                        0.024036875417496602,
+                        0.02602605089260403,
+                        0.05981475365626236,
+                        0.028429534988341422,
+                        0.08916377427180806,
+                        0.03528055113076741
+                    ],
+                    "q01": [
+                        -0.1545458011329174,
+                        -0.031217777729034422,
+                        -0.11606558486819267,
+                        -0.0627044416964054,
+                        -0.056227393448352814,
+                        -0.5197785004973411,
+                        -0.156750052748248
+                    ],
+                    "q99": [
+                        0.05294616930186727,
+                        0.11603573272004716,
+                        0.02942010760307312,
+                        0.32693205326795444,
+                        0.11353727197274502,
+                        0.06767475279048074,
+                        0.06325854659080493
+                    ]
+                },
+                "right_arm": {
+                    "max": [
+                        1.0019388496875763,
+                        0.6778036952018738,
+                        0.8202316761016846,
+                        1.2868973910808563,
+                        1.6054365634918213,
+                        0.9513278678059578,
+                        1.0065587792778388
+                    ],
+                    "min": [
+                        -0.8566206395626068,
+                        -0.5352737102657557,
+                        -0.9050570726394653,
+                        -1.1766326129436493,
+                        -1.1665689051151276,
+                        -1.226991206407547,
+                        -1.4597684778273106
+                    ],
+                    "mean": [
+                        0.0011831602733654706,
+                        0.004668234701338991,
+                        -0.00464666805688978,
+                        -0.008819726561502726,
+                        0.01680292255365519,
+                        -0.023272413508623503,
+                        0.001828295062135536
+                    ],
+                    "std": [
+                        0.15114015459104163,
+                        0.10717935148199742,
+                        0.12210241645133885,
+                        0.18724624428113243,
+                        0.21873655609526446,
+                        0.18543346050682308,
+                        0.1150104843557695
+                    ],
+                    "q01": [
+                        -0.39952214881777764,
+                        -0.25475079081952573,
+                        -0.4339929953217506,
+                        -0.645755667425692,
+                        -0.5597020275890827,
+                        -0.640592061728239,
+                        -0.3658384716138244
+                    ],
+                    "q99": [
+                        0.47152970274910266,
+                        0.3625141793861984,
+                        0.3024886786937714,
+                        0.5137429375201458,
+                        0.7895279735326737,
+                        0.5061310570687053,
+                        0.30551581159233987
+                    ]
+                },
+                "rpy": {
+                    "max": [
+                        0.2602043002843857,
+                        0.0978180319070816,
+                        0.4442971870303154
+                    ],
+                    "min": [
+                        -0.2339760884642601,
+                        -0.27553248405456543,
+                        -0.44536447897553444
+                    ],
+                    "mean": [
+                        -0.0005869882495184713,
+                        0.0015875524968715301,
+                        0.0008563187639556434
+                    ],
+                    "std": [
+                        0.015963769360634213,
+                        0.009780051628911945,
+                        0.024366577111250988
+                    ],
+                    "q01": [
+                        -0.049293445050716395,
+                        -0.023501918860711155,
+                        -0.0595750443637371
+                    ],
+                    "q99": [
+                        0.05025968039408321,
+                        0.028102314658462975,
+                        0.0824947778135533
+                    ]
+                },
+                "height": {
+                    "max": [
+                        0.04315239191055298
+                    ],
+                    "min": [
+                        -0.08728241920471191
+                    ],
+                    "mean": [
+                        5.191507986045299e-05
+                    ],
+                    "std": [
+                        0.004163189516331377
+                    ],
+                    "q01": [
+                        -0.012046998739242552
+                    ],
+                    "q99": [
+                        0.012092745304107644
+                    ]
+                },
+                "torso_vx": {
+                    "max": [
+                        0.3499999940395355
+                    ],
+                    "min": [
+                        0.0
+                    ],
+                    "mean": [
+                        0.017269024247052768
+                    ],
+                    "std": [
+                        0.07580197349073346
+                    ],
+                    "q01": [
+                        0.0
+                    ],
+                    "q99": [
+                        0.3499999940395355
+                    ]
+                },
+                "torso_vy": {
+                    "max": [
+                        0.5
+                    ],
+                    "min": [
+                        0.0
+                    ],
+                    "mean": [
+                        0.0002320065992988245
+                    ],
+                    "std": [
+                        0.01076798368252486
+                    ],
+                    "q01": [
+                        0.0
+                    ],
+                    "q99": [
+                        0.0
+                    ]
+                },
+                "torso_vyaw": {
+                    "max": [
+                        0.5
+                    ],
+                    "min": [
+                        -0.5
+                    ],
+                    "mean": [
+                        -0.03733115137363985
+                    ],
+                    "std": [
+                        0.1199053455233051
+                    ],
+                    "q01": [
+                        -0.5
+                    ],
+                    "q99": [
+                        0.0
+                    ]
+                },
+                "torso_dyaw": {
+                    "max": [
+                        0.3072107136249542
+                    ],
+                    "min": [
+                        -0.1875186562538147
+                    ],
+                    "mean": [
+                        0.016342195046893805
+                    ],
+                    "std": [
+                        0.048962870720188865
+                    ],
+                    "q01": [
+                        -0.005849878406152129
+                    ],
+                    "q99": [
+                        0.22258896410465237
+                    ]
+                }
+            }
+        },
+        "modalities": {
+            "video": {
+                "egocentric": {
+                    "resolution": [
+                        640,
+                        480
+                    ],
+                    "channels": 3,
+                    "fps": 30.0
+                }
+            },
+            "state": {
+                "left_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "left_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "rpy": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        3
+                    ],
+                    "continuous": true
+                },
+                "height": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                }
+            },
+            "action": {
+                "left_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "left_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "rpy": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        3
+                    ],
+                    "continuous": true
+                },
+                "height": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_vx": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_vy": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_vyaw": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_dyaw": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                }
+            }
+        },
+        "embodiment_tag": "real_teleop_g1"
+    }
+}

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e54b0d78312d085adbb1b47afb1539a753e1648a42e3961d15383011d5d41dad
+size 24726494997

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e7d4159a5f78e298eea595df891039c8eb3e55150e04ecd7e6e77007cbccde8
+size 24726494869

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00f4cf817a1d4526c9127d904d8e30defe5fba49dfc8d3d18e1b5fd9afcf2b3a
+size 24726494933

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1064a887d1d9df9091382c220381cfe871b770e7878d74d20851111e3cfa7595
+size 24726495189

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ac6d1ec241aa1e4252c528f9fdfbeaa6edec9afab9a946c1d1d7c0edd4db60a
+size 24726494869

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3e8679f10db492ccc08552a1eadcc77563097a20661e5eef2ff9a923abf8ef5
+size 24726494805

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00228701cf4e15e5be08166f7c24db907a4f021ac28c178a92f4141e8c87ea2c
+size 24726494805

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1319970345100e63d9190905d280b43f52ae42bbc37c8210b8ddfbbdcb375c5
+size 24726495317

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/global_step20000/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5694cb3d62897774b80e800e5d8381c3db0afadcbcd18fff7660848e74473f6b
+size 58729254967

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/latest

@@ -0,0 +1 @@
+global_step20000

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00001-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0e0ff763ed8bb6b855782e10f1577d676bc7abce2df0e46cea5d340a05963a5
+size 4935816720

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00002-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dd6a6ed2735b920badc8b58864ac567f921b5e1ab537f9c755e35e6c65ebd8c
+size 4983107552

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00003-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b24734a620ebeda5fda8ba3860115a53ba609630eebeef8be48e9c585c8a80e
+size 4937616256

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00004-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25fa20cfabcc3e779f0d47b2f4dfec0e1746e5a5834a74124cf5fbb4cf75769c
+size 4987670360

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00005-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6de3521b2cca697af9c1be60f81359e9c2f21325ab881f5cb9a9da0beaeb85f7
+size 4987714608

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00006-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5659d860ff59c32f444ec88eac2f0e5b05d04d931d7bd83672057d43c5786155
+size 4950963456

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00007-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c27719ee648bb3595b6813d99678c6ea15b8d5f07953cbd5096863ce612a274
+size 4951004896

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00008-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c81c30844b10d612be6a677b7ee39c168f08eeef2e488b8943bbf87b19569c87
+size 4950963472

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00009-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0bdd073dfa80f90311e8690244be6960bd25dbf7231d07029afa3ce6531e030
+size 4950984200

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/model-00010-of-00010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2343d0348efd68e66fd5377236a217e0d8f3edd9213d37e815ab4200d26fb97
+size 1212823208

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260324_0/000001_03_24_13_58_27_n34.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:397a4f513fd768f0dda4f6e09307b4858693d60f759d3b17711d7b2b4c8a7e31
+size 664300

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260326_0/000000_03_26_17_26_53_n40.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d62daa6a188565dad50c8354387857de48434cea229883fd00af554245ac55a6
+size 793290

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260326_0/000001_03_26_17_31_03_n40.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d62daa6a188565dad50c8354387857de48434cea229883fd00af554245ac55a6
+size 793290

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260327_0/000000_03_27_02_53_54_n50.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a69d9c7826be63d6da2a38bba1529652c64dcabd1261e2d723cae4c035849d5
+size 995418

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/real_world_eval_gen_20260327_0/000001_03_27_02_55_16_n40.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9da95fd035e16e89c0ec661c85aa36416f007c6c79a9f36e87b27f7109e00818
+size 792418

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cae4b6de2acf6264a5c481dcbd7aa866d94dca1326cdbccff19f3fef8f7c53f
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fea41920daf9f3742ab29b10d916136a3d60dd6eb2a1e03f018d35e54ef2e868
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:099243b837d52dab61610fca2e997abd8a3c6e4ede0d8ccc83eee5e47383ded5
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be34cfa1b5e67c3e539fb03abbe775743b14e7fa4a17684d9400f13e8d16b341
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fec0da59f18a4474684abde6d1267e41f007667a870ca3c08f7467fd1381a20
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d71402cde94b72b47be2a664a4c005d140739f9eec545f070ca9677825a39dc2
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:841612644c79a5764b67f40bd1a9be5ee750c520b8b2a207b8bfd46f9cb4b5cf
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f970d4dde7f6abbd2c2a190732b19c85e7aafed87115303a61018145f87ecd7a
+size 16389

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28508b0a940eb7f712597070ad628ccf8634a0efd412740f51a02d7843f5d668
+size 1465

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/wandb_config.json

@@ -0,0 +1 @@
+{"project": "dreamzero", "run_id": ""}

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/checkpoint-20000/zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info("Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info("Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/experiment_cfg/conf.yaml

@@ -0,0 +1,3318 @@
+model:
+  _target_: groot.vla.model.dreamzero.base_vla.VLA
+  _convert_: object
+  config:
+    _target_: groot.vla.model.dreamzero.base_vla.VLAConfig
+    _recursive_: false
+    model_dtype: float32
+    hidden_size: 0
+    action_horizon: 24
+    action_dim: 36
+    backbone_cfg:
+      _target_: groot.vla.model.dreamzero.backbone.identity.IdentityBackbone
+    action_head_cfg:
+      config:
+        backbone_features_projector_cfg: null
+        _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHeadConfig
+        _recursive_: false
+        tiled: false
+        tile_size_height: 34
+        tile_size_width: 34
+        tile_stride_height: 18
+        tile_stride_width: 16
+        lora_rank: 4
+        lora_alpha: 4
+        num_frames: 33
+        num_frame_per_block: 2
+        lora_target_modules: q,k,v,o,ffn.0,ffn.2
+        init_lora_weights: kaiming
+        train_architecture: full
+        use_gradient_checkpointing: true
+        add_pos_embed: true
+        model_dtype: float32
+        max_state_dim: 64
+        max_action_dim: 36
+        action_loss_embodiment_ids:
+        - 26
+        - 17
+        - 32
+        hidden_size: 64
+        input_embedding_dim: 1536
+        backbone_embedding_dim: 0
+        repa_layer: 8
+        repa_coeff: 1.0
+        load_pretrained_det_decode_layer_path: null
+        freeze_decode_layer: false
+        expand_batch: null
+        use_vlln: true
+        vl_self_attention_cfg:
+          _target_: groot.vla.model.n1_5.modules.cross_attention_dit.SelfAttentionTransformer
+          positional_embeddings: null
+          num_layers: 4
+          num_attention_heads: 24
+          attention_head_dim: 64
+          dropout: 0.2
+          final_dropout: true
+        diffusion_model_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_dit_action_casual_chunk.CausalWanModel
+          _convert_: object
+          diffusion_model_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+          model_type: i2v
+          frame_seqlen: 220
+          dim: 5120
+          in_dim: 36
+          ffn_dim: 13824
+          out_dim: 16
+          freq_dim: 256
+          eps: 1.0e-06
+          num_heads: 40
+          num_layers: 40
+          max_chunk_size: 4
+          num_frame_per_block: 2
+          num_action_per_block: 24
+          num_state_per_block: 1
+          action_dim: 36
+        text_encoder_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_text_encoder.WanTextEncoder
+          _convert_: object
+          text_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+        image_encoder_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_image_encoder.WanImageEncoder
+          _convert_: object
+          image_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+        vae_cfg:
+          _target_: groot.vla.model.dreamzero.modules.wan_video_vae.WanVideoVAE
+          _convert_: object
+          vae_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+        action_dim: 36
+        action_horizon: 24
+        num_inference_timesteps: 4
+        noise_beta_alpha: 1.5
+        noise_beta_beta: 1.0
+        noise_s: 0.999
+        num_timestep_buckets: 1000
+        decouple_video_action_noise: false
+        video_noise_beta_alpha: 3.0
+        video_noise_beta_beta: 1.0
+        tune_projector: true
+        tune_diffusion_model: true
+        skip_component_loading: true
+      _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHead
+      _convert_: object
+train_dataset:
+  _target_: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotMixtureDataset.from_mixture_spec
+  _convert_: object
+  mixture_spec:
+  - dataset_path:
+      real_teleop_g1:
+      - /hfm/boqian/liboqian_data/data/real_data/gear/g1/Pick_bottle_and_turn_and_pour_into_cup
+    dataset_weight: 1.0
+    distribute_weights: true
+  dataset_class: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotSubLangSingleActionChunkDatasetDROID
+  all_modality_configs:
+    oxe_droid:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - 0
+        modality_keys:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.joint_position
+        - state.gripper_position
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.joint_position
+        - action.gripper_position
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.language.language_instruction
+        - annotation.language.language_instruction_2
+        - annotation.language.language_instruction_3
+      lapa_action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - lapa_action
+    agibot:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - -3
+        - -2
+        - -1
+        - 0
+        modality_keys:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.language.action_text
+    yam:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - 0
+        modality_keys:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.task
+    real_teleop_g1:
+      video:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        - 24
+        eval_delta_indices:
+        - 0
+        modality_keys:
+        - video.egocentric
+      state:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+      action:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        - 1
+        - 2
+        - 3
+        - 4
+        - 5
+        - 6
+        - 7
+        - 8
+        - 9
+        - 10
+        - 11
+        - 12
+        - 13
+        - 14
+        - 15
+        - 16
+        - 17
+        - 18
+        - 19
+        - 20
+        - 21
+        - 22
+        - 23
+        modality_keys:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+      language:
+        _target_: groot.vla.data.dataset.ModalityConfig
+        delta_indices:
+        - 0
+        modality_keys:
+        - annotation.language.language_instruction
+  all_transforms:
+    oxe_droid:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.joint_position
+        - state.gripper_position
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.joint_position
+        - state.gripper_position
+        normalization_modes:
+          state.joint_position: q99
+          state.gripper_position: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.joint_position
+        - action.gripper_position
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.joint_position
+        - action.gripper_position
+        normalization_modes:
+          action.joint_position: q99
+          action.gripper_position: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.exterior_image_1_left
+        - video.exterior_image_2_left
+        - video.wrist_image_left
+        state_concat_order:
+        - state.joint_position
+        - state.gripper_position
+        action_concat_order:
+        - action.joint_position
+        - action.gripper_position
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+    agibot:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+        normalization_modes:
+          state.left_arm_joint_position: q99
+          state.right_arm_joint_position: q99
+          state.left_effector_position: q99
+          state.right_effector_position: q99
+          state.head_position: q99
+          state.waist_position: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+        normalization_modes:
+          action.left_arm_joint_position: q99
+          action.right_arm_joint_position: q99
+          action.left_effector_position: q99
+          action.right_effector_position: q99
+          action.head_position: q99
+          action.waist_position: q99
+          action.robot_velocity: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.top_head
+        - video.hand_left
+        - video.hand_right
+        state_concat_order:
+        - state.left_arm_joint_position
+        - state.right_arm_joint_position
+        - state.left_effector_position
+        - state.right_effector_position
+        - state.head_position
+        - state.waist_position
+        action_concat_order:
+        - action.left_arm_joint_position
+        - action.right_arm_joint_position
+        - action.left_effector_position
+        - action.right_effector_position
+        - action.head_position
+        - action.waist_position
+        - action.robot_velocity
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+    yam:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+        normalization_modes:
+          state.left_joint_pos: q99
+          state.left_gripper_pos: q99
+          state.right_joint_pos: q99
+          state.right_gripper_pos: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+        normalization_modes:
+          action.left_joint_pos: q99
+          action.left_gripper_pos: q99
+          action.right_joint_pos: q99
+          action.right_gripper_pos: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.top_camera-images-rgb
+        - video.left_camera-images-rgb
+        - video.right_camera-images-rgb
+        state_concat_order:
+        - state.left_joint_pos
+        - state.left_gripper_pos
+        - state.right_joint_pos
+        - state.right_gripper_pos
+        action_concat_order:
+        - action.left_joint_pos
+        - action.left_gripper_pos
+        - action.right_joint_pos
+        - action.right_gripper_pos
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+    real_teleop_g1:
+      _target_: groot.vla.data.transform.ComposedModalityTransform
+      transforms:
+      - _target_: groot.vla.data.transform.VideoToTensor
+        apply_to:
+        - video.egocentric
+      - _target_: groot.vla.data.transform.VideoCrop
+        apply_to:
+        - video.egocentric
+        scale: 0.95
+        mode: random
+      - _target_: groot.vla.data.transform.VideoResize
+        apply_to:
+        - video.egocentric
+        height: 176
+        width: 320
+        interpolation: linear
+      - _target_: groot.vla.data.transform.VideoColorJitter
+        apply_to:
+        - video.egocentric
+        brightness: 0.3
+        contrast: 0.4
+        saturation: 0.5
+        hue: 0.08
+      - _target_: groot.vla.data.transform.VideoToNumpy
+        apply_to:
+        - video.egocentric
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+        normalization_modes:
+          state.left_hand: q99
+          state.right_hand: q99
+          state.left_arm: q99
+          state.right_arm: q99
+          state.rpy: q99
+          state.height: q99
+      - _target_: groot.vla.data.transform.StateActionToTensor
+        apply_to:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+      - _target_: groot.vla.data.transform.StateActionTransform
+        apply_to:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+        normalization_modes:
+          action.left_hand: q99
+          action.right_hand: q99
+          action.left_arm: q99
+          action.right_arm: q99
+          action.rpy: q99
+          action.height: q99
+          action.torso_vx: q99
+          action.torso_vy: q99
+          action.torso_vyaw: q99
+          action.torso_dyaw: q99
+      - _target_: groot.vla.data.transform.ConcatTransform
+        video_concat_order:
+        - video.egocentric
+        state_concat_order:
+        - state.left_hand
+        - state.right_hand
+        - state.left_arm
+        - state.right_arm
+        - state.rpy
+        - state.height
+        action_concat_order:
+        - action.left_hand
+        - action.right_hand
+        - action.left_arm
+        - action.right_arm
+        - action.rpy
+        - action.height
+        - action.torso_vx
+        - action.torso_vy
+        - action.torso_vyaw
+        - action.torso_dyaw
+      - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+        default_instruction: Perform the default behavior.
+        language_dropout_prob: 0.0
+        always_use_default_instruction: false
+        max_state_dim: 64
+        max_action_dim: 36
+        max_length: 512
+        state_horizon: 1
+        action_horizon: 24
+        embodiment_tag_mapping:
+          real_gr1_arms_only: 0
+          real_gr1_arms_only_annotated: 1
+          real_gr1_arms_waist: 2
+          real_gr1_arms_waist_annotated: 3
+          dexmg_gr1_arms_only_inspire: 4
+          dexmg_gr1_arms_only_fourier: 5
+          dexmg_gr1_arms_waist_fourier: 6
+          robocasa_single_arm: 7
+          onex_eve_gripper: 8
+          robocasa_gr1_arms_only_inspire_hands: 9
+          robocasa_gr1_arms_only_fourier_hands: 10
+          robocasa_gr1_fixed_lower_body_inspire_hands: 11
+          robocasa_gr1_fixed_lower_body_fourier_hands: 12
+          robocasa_panda_omron: 13
+          robocasa_bimanual_panda_parallel_gripper: 15
+          robocasa_bimanual_panda_inspire_hand: 16
+          oxe_droid: 17
+          oxe_fractal: 18
+          oxe_language_table: 19
+          oxe_bridge: 20
+          real_panda_single_arm: 21
+          hot3d_hands_only: 23
+          gr1_unified: 24
+          robocasa_gr1_arms_waist_fourier_hands: 25
+          agibot: 26
+          lapa: 27
+          oxe_mutex: 28
+          oxe_roboset: 29
+          oxe_plex: 30
+          dream: 31
+          yam: 32
+          xdof: 22
+          gr1_unified_segmentation: 14
+          language_table_sim: 7
+          gr1_isaac: 0
+          sim_behavior_r1_pro: 31
+          mecka_hands: 27
+          real_r1_pro_sharpa: 28
+          real_teleop_g1: 33
+        tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  metadata_versions:
+    oxe_droid: '0221'
+    agibot: '0221'
+    yam: '0221'
+    real_teleop_g1: '0221'
+  fps:
+    yam: 30
+    real_teleop_g1: 30
+  dataset_kwargs:
+    video_backend: decord
+    use_global_metadata: false
+    max_chunk_size: 4
+    relative_action: true
+    relative_action_keys:
+    - left_hand
+    - right_hand
+    - left_arm
+    - right_arm
+    - rpy
+    - height
+    relative_action_per_horizon: false
+  mixture_kwargs:
+    training: true
+    balance_dataset_weights: false
+    seed: 42
+    shard_sampling_rate: 0.1
+trainer:
+  _target_: groot.vla.experiment.VLATrainer
+  _partial_: true
+  _recursive_: false
+  callbacks: null
+  model: ???
+  train_dataset: ???
+  compute_dtype: ???
+  benchmark_time: false
+  enable_profiling: false
+  profiling_steps: 5
+  enable_prof_callback: false
+  profile_start_step: 50
+  profile_warmup_steps: 1
+  profile_active_steps: 3
+  profile_record_shapes: false
+  profile_with_stack: false
+  profile_memory: false
+wandb_project: dreamzero
+output_dir: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+load_from_yaml: null
+gear_credentials: null
+upload_checkpoints: false
+upload_every: 1000
+upload_last_n_checkpoints: 5
+remove_unused_columns: false
+bf16: true
+tf32: true
+global_batch_size: null
+raise_error_if_global_batch_size_not_set: false
+per_device_train_batch_size: 1
+per_device_eval_batch_size: 64
+gradient_accumulation_steps: 1
+dataloader_num_workers: 1
+dataloader_pin_memory: false
+dataloader_persistent_workers: true
+optim: adamw_torch
+learning_rate: 0.0001
+adam_beta1: 0.95
+adam_beta2: 0.999
+adam_epsilon: 1.0e-08
+weight_decay: 1.0e-05
+lr_scheduler_type: cosine
+warmup_ratio: 0.05
+logging_steps: 10.0
+num_train_epochs: 1000
+max_steps: 20000
+save_strategy: steps
+save_steps: 8000
+eval_strategy: 'no'
+save_total_limit: 10
+report_to: wandb
+seed: 42
+do_eval: false
+gradient_checkpointing: false
+ddp_find_unused_parameters: false
+ddp_bucket_cap_mb: 100
+ray_num_workers: ???
+eval_bf16: true
+torch_compile_mode: null
+pretrained_model_path: /hfm/boqian/liboqian_data/checkpoints/DreamZero-AgiBot
+only_tune_projectors: false
+save_llm: false
+save_lora_only: false
+save_value_model: false
+save_q_model: false
+download_cache: false
+training_args:
+  _target_: transformers.TrainingArguments
+  output_dir: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+  run_name: dreamzero_real_teleop_g1_full_finetune
+  remove_unused_columns: false
+  deepspeed: groot/vla/configs/deepspeed/zero2_offload.json
+  gradient_checkpointing: false
+  bf16: true
+  tf32: true
+  per_device_train_batch_size: 1
+  per_device_eval_batch_size: 64
+  gradient_accumulation_steps: 1
+  dataloader_num_workers: 1
+  dataloader_pin_memory: false
+  dataloader_persistent_workers: true
+  optim: adamw_torch
+  adam_beta1: 0.95
+  adam_beta2: 0.999
+  adam_epsilon: 1.0e-08
+  learning_rate: 1.0e-05
+  weight_decay: 1.0e-05
+  warmup_ratio: 0.05
+  lr_scheduler_type: cosine
+  logging_steps: 10.0
+  num_train_epochs: 1000
+  max_steps: 20000
+  save_strategy: steps
+  save_steps: 8000
+  save_total_limit: 10
+  report_to: wandb
+  seed: 42
+  do_eval: false
+  ddp_find_unused_parameters: false
+  ddp_bucket_cap_mb: 100
+  torch_compile_mode: null
+profile_dir: null
+backbone_hidden_size: 0
+backbone_cfg:
+  _target_: groot.vla.model.dreamzero.backbone.identity.IdentityBackbone
+action_head_cfg:
+  config:
+    backbone_features_projector_cfg: null
+    _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHeadConfig
+    _recursive_: false
+    tiled: false
+    tile_size_height: 34
+    tile_size_width: 34
+    tile_stride_height: 18
+    tile_stride_width: 16
+    lora_rank: 4
+    lora_alpha: 4
+    num_frames: 33
+    num_frame_per_block: 2
+    lora_target_modules: q,k,v,o,ffn.0,ffn.2
+    init_lora_weights: kaiming
+    train_architecture: full
+    use_gradient_checkpointing: true
+    add_pos_embed: true
+    model_dtype: float32
+    max_state_dim: 64
+    max_action_dim: 36
+    action_loss_embodiment_ids:
+    - 26
+    - 17
+    - 32
+    hidden_size: 64
+    input_embedding_dim: 1536
+    backbone_embedding_dim: 0
+    repa_layer: 8
+    repa_coeff: 1.0
+    load_pretrained_det_decode_layer_path: null
+    freeze_decode_layer: false
+    expand_batch: null
+    use_vlln: true
+    vl_self_attention_cfg:
+      _target_: groot.vla.model.n1_5.modules.cross_attention_dit.SelfAttentionTransformer
+      positional_embeddings: null
+      num_layers: 4
+      num_attention_heads: 24
+      attention_head_dim: 64
+      dropout: 0.2
+      final_dropout: true
+    diffusion_model_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_dit_action_casual_chunk.CausalWanModel
+      _convert_: object
+      diffusion_model_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+      model_type: i2v
+      frame_seqlen: 220
+      dim: 5120
+      in_dim: 36
+      ffn_dim: 13824
+      out_dim: 16
+      freq_dim: 256
+      eps: 1.0e-06
+      num_heads: 40
+      num_layers: 40
+      max_chunk_size: 4
+      num_frame_per_block: 2
+      num_action_per_block: 24
+      num_state_per_block: 1
+      action_dim: 36
+    text_encoder_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_text_encoder.WanTextEncoder
+      _convert_: object
+      text_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+    image_encoder_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_image_encoder.WanImageEncoder
+      _convert_: object
+      image_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+    vae_cfg:
+      _target_: groot.vla.model.dreamzero.modules.wan_video_vae.WanVideoVAE
+      _convert_: object
+      vae_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+    action_dim: 36
+    action_horizon: 24
+    num_inference_timesteps: 4
+    noise_beta_alpha: 1.5
+    noise_beta_beta: 1.0
+    noise_s: 0.999
+    num_timestep_buckets: 1000
+    decouple_video_action_noise: false
+    video_noise_beta_alpha: 3.0
+    video_noise_beta_beta: 1.0
+    tune_projector: true
+    tune_diffusion_model: true
+    skip_component_loading: true
+  _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHead
+  _convert_: object
+add_pos_embed: true
+hidden_size: 64
+attn_dropout: 0.2
+repa_layer: 8
+repa_coeff: 1.0
+load_pretrained_det_decode_layer_path: null
+expand_batch: null
+dit_version: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+text_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+image_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+vae_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+train_architecture: full
+num_frame_per_block: 2
+num_action_per_block: 24
+num_state_per_block: 1
+frame_seqlen: 220
+embodiment_tag_to_projector_index:
+  real_gr1_arms_only: 0
+  real_gr1_arms_only_annotated: 1
+  real_gr1_arms_waist: 2
+  real_gr1_arms_waist_annotated: 3
+  dexmg_gr1_arms_only_inspire: 4
+  dexmg_gr1_arms_only_fourier: 5
+  dexmg_gr1_arms_waist_fourier: 6
+  robocasa_single_arm: 7
+  onex_eve_gripper: 8
+  robocasa_gr1_arms_only_inspire_hands: 9
+  robocasa_gr1_arms_only_fourier_hands: 10
+  robocasa_gr1_fixed_lower_body_inspire_hands: 11
+  robocasa_gr1_fixed_lower_body_fourier_hands: 12
+  robocasa_panda_omron: 13
+  robocasa_bimanual_panda_parallel_gripper: 15
+  robocasa_bimanual_panda_inspire_hand: 16
+  oxe_droid: 17
+  oxe_fractal: 18
+  oxe_language_table: 19
+  oxe_bridge: 20
+  real_panda_single_arm: 21
+  hot3d_hands_only: 23
+  gr1_unified: 24
+  robocasa_gr1_arms_waist_fourier_hands: 25
+  agibot: 26
+  lapa: 27
+  oxe_mutex: 28
+  oxe_roboset: 29
+  oxe_plex: 30
+  dream: 31
+  yam: 32
+  xdof: 22
+  gr1_unified_segmentation: 14
+  language_table_sim: 7
+  gr1_isaac: 0
+  sim_behavior_r1_pro: 31
+  mecka_hands: 27
+  real_r1_pro_sharpa: 28
+  real_teleop_g1: 33
+max_length: 512
+num_views: 1
+tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+data_collator:
+  _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DefaultDataCollator
+  tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  max_length: 512
+  num_views: 1
+  embodiment_tag_mapping:
+    real_gr1_arms_only: 0
+    real_gr1_arms_only_annotated: 1
+    real_gr1_arms_waist: 2
+    real_gr1_arms_waist_annotated: 3
+    dexmg_gr1_arms_only_inspire: 4
+    dexmg_gr1_arms_only_fourier: 5
+    dexmg_gr1_arms_waist_fourier: 6
+    robocasa_single_arm: 7
+    onex_eve_gripper: 8
+    robocasa_gr1_arms_only_inspire_hands: 9
+    robocasa_gr1_arms_only_fourier_hands: 10
+    robocasa_gr1_fixed_lower_body_inspire_hands: 11
+    robocasa_gr1_fixed_lower_body_fourier_hands: 12
+    robocasa_panda_omron: 13
+    robocasa_bimanual_panda_parallel_gripper: 15
+    robocasa_bimanual_panda_inspire_hand: 16
+    oxe_droid: 17
+    oxe_fractal: 18
+    oxe_language_table: 19
+    oxe_bridge: 20
+    real_panda_single_arm: 21
+    hot3d_hands_only: 23
+    gr1_unified: 24
+    robocasa_gr1_arms_waist_fourier_hands: 25
+    agibot: 26
+    lapa: 27
+    oxe_mutex: 28
+    oxe_roboset: 29
+    oxe_plex: 30
+    dream: 31
+    yam: 32
+    xdof: 22
+    gr1_unified_segmentation: 14
+    language_table_sim: 7
+    gr1_isaac: 0
+    sim_behavior_r1_pro: 31
+    mecka_hands: 27
+    real_r1_pro_sharpa: 28
+    real_teleop_g1: 33
+num_visual_tokens_per_frame: 16
+max_state_dim: 64
+max_action_dim: 36
+language_dropout_prob: 0.0
+model_specific_transform:
+  _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+  default_instruction: Perform the default behavior.
+  language_dropout_prob: 0.0
+  always_use_default_instruction: false
+  max_state_dim: 64
+  max_action_dim: 36
+  max_length: 512
+  state_horizon: 1
+  action_horizon: 24
+  embodiment_tag_mapping:
+    real_gr1_arms_only: 0
+    real_gr1_arms_only_annotated: 1
+    real_gr1_arms_waist: 2
+    real_gr1_arms_waist_annotated: 3
+    dexmg_gr1_arms_only_inspire: 4
+    dexmg_gr1_arms_only_fourier: 5
+    dexmg_gr1_arms_waist_fourier: 6
+    robocasa_single_arm: 7
+    onex_eve_gripper: 8
+    robocasa_gr1_arms_only_inspire_hands: 9
+    robocasa_gr1_arms_only_fourier_hands: 10
+    robocasa_gr1_fixed_lower_body_inspire_hands: 11
+    robocasa_gr1_fixed_lower_body_fourier_hands: 12
+    robocasa_panda_omron: 13
+    robocasa_bimanual_panda_parallel_gripper: 15
+    robocasa_bimanual_panda_inspire_hand: 16
+    oxe_droid: 17
+    oxe_fractal: 18
+    oxe_language_table: 19
+    oxe_bridge: 20
+    real_panda_single_arm: 21
+    hot3d_hands_only: 23
+    gr1_unified: 24
+    robocasa_gr1_arms_waist_fourier_hands: 25
+    agibot: 26
+    lapa: 27
+    oxe_mutex: 28
+    oxe_roboset: 29
+    oxe_plex: 30
+    dream: 31
+    yam: 32
+    xdof: 22
+    gr1_unified_segmentation: 14
+    language_table_sim: 7
+    gr1_isaac: 0
+    sim_behavior_r1_pro: 31
+    mecka_hands: 27
+    real_r1_pro_sharpa: 28
+    real_teleop_g1: 33
+  tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+use_global_metadata: false
+num_frames: 33
+action_horizon: 24
+state_horizon: 1
+image_resolution_width: 320
+image_resolution_height: 176
+image_resolution_width_single_frame: 256
+image_resolution_height_single_frame: 256
+totensor_cfg:
+  _target_: groot.vla.data.transform.VideoToTensor
+  apply_to: ???
+crop_cfg:
+  _target_: groot.vla.data.transform.VideoCrop
+  apply_to: ???
+  scale: 0.95
+  mode: random
+resize_cfg:
+  _target_: groot.vla.data.transform.VideoResize
+  apply_to: ???
+  height: 176
+  width: 320
+  interpolation: linear
+resize_cfg_single_frame:
+  _target_: groot.vla.data.transform.VideoResize
+  apply_to: ???
+  height: 256
+  width: 256
+  interpolation: linear
+color_jitter_cfg:
+  _target_: groot.vla.data.transform.VideoColorJitter
+  apply_to: ???
+  brightness: 0.3
+  contrast: 0.4
+  saturation: 0.5
+  hue: 0.08
+random_grayscale_cfg:
+  _target_: groot.vla.data.transform.VideoRandomGrayscale
+  apply_to: ???
+  p: 0.1
+random_posterize_cfg:
+  _target_: groot.vla.data.transform.VideoRandomPosterize
+  apply_to: ???
+  bits: 4
+  p: 0.1
+normalize_cfg:
+  _target_: groot.vla.data.transform.VideoNormalize
+  apply_to: ???
+  mean:
+  - 0.5
+  - 0.5
+  - 0.5
+  std:
+  - 0.5
+  - 0.5
+  - 0.5
+to_numpy_cfg:
+  _target_: groot.vla.data.transform.VideoToNumpy
+  apply_to: ???
+modality_config_oxe_droid:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - 0
+    modality_keys:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.joint_position
+    - state.gripper_position
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.joint_position
+    - action.gripper_position
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.language.language_instruction
+    - annotation.language.language_instruction_2
+    - annotation.language.language_instruction_3
+  lapa_action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - lapa_action
+transform_oxe_droid:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.joint_position
+    - state.gripper_position
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.joint_position
+    - state.gripper_position
+    normalization_modes:
+      state.joint_position: q99
+      state.gripper_position: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.joint_position
+    - action.gripper_position
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.joint_position
+    - action.gripper_position
+    normalization_modes:
+      action.joint_position: q99
+      action.gripper_position: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.exterior_image_1_left
+    - video.exterior_image_2_left
+    - video.wrist_image_left
+    state_concat_order:
+    - state.joint_position
+    - state.gripper_position
+    action_concat_order:
+    - action.joint_position
+    - action.gripper_position
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_config_agibot:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - -3
+    - -2
+    - -1
+    - 0
+    modality_keys:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.language.action_text
+transform_agibot:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+    normalization_modes:
+      state.left_arm_joint_position: q99
+      state.right_arm_joint_position: q99
+      state.left_effector_position: q99
+      state.right_effector_position: q99
+      state.head_position: q99
+      state.waist_position: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+    normalization_modes:
+      action.left_arm_joint_position: q99
+      action.right_arm_joint_position: q99
+      action.left_effector_position: q99
+      action.right_effector_position: q99
+      action.head_position: q99
+      action.waist_position: q99
+      action.robot_velocity: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.top_head
+    - video.hand_left
+    - video.hand_right
+    state_concat_order:
+    - state.left_arm_joint_position
+    - state.right_arm_joint_position
+    - state.left_effector_position
+    - state.right_effector_position
+    - state.head_position
+    - state.waist_position
+    action_concat_order:
+    - action.left_arm_joint_position
+    - action.right_arm_joint_position
+    - action.left_effector_position
+    - action.right_effector_position
+    - action.head_position
+    - action.waist_position
+    - action.robot_velocity
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_config_yam:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - 0
+    modality_keys:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.task
+transform_yam:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+    normalization_modes:
+      state.left_joint_pos: q99
+      state.left_gripper_pos: q99
+      state.right_joint_pos: q99
+      state.right_gripper_pos: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+    normalization_modes:
+      action.left_joint_pos: q99
+      action.left_gripper_pos: q99
+      action.right_joint_pos: q99
+      action.right_gripper_pos: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.top_camera-images-rgb
+    - video.left_camera-images-rgb
+    - video.right_camera-images-rgb
+    state_concat_order:
+    - state.left_joint_pos
+    - state.left_gripper_pos
+    - state.right_joint_pos
+    - state.right_gripper_pos
+    action_concat_order:
+    - action.left_joint_pos
+    - action.left_gripper_pos
+    - action.right_joint_pos
+    - action.right_gripper_pos
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_config_real_teleop_g1:
+  video:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    - 24
+    eval_delta_indices:
+    - 0
+    modality_keys:
+    - video.egocentric
+  state:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+  action:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    - 11
+    - 12
+    - 13
+    - 14
+    - 15
+    - 16
+    - 17
+    - 18
+    - 19
+    - 20
+    - 21
+    - 22
+    - 23
+    modality_keys:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+  language:
+    _target_: groot.vla.data.dataset.ModalityConfig
+    delta_indices:
+    - 0
+    modality_keys:
+    - annotation.language.language_instruction
+transform_real_teleop_g1:
+  _target_: groot.vla.data.transform.ComposedModalityTransform
+  transforms:
+  - _target_: groot.vla.data.transform.VideoToTensor
+    apply_to:
+    - video.egocentric
+  - _target_: groot.vla.data.transform.VideoCrop
+    apply_to:
+    - video.egocentric
+    scale: 0.95
+    mode: random
+  - _target_: groot.vla.data.transform.VideoResize
+    apply_to:
+    - video.egocentric
+    height: 176
+    width: 320
+    interpolation: linear
+  - _target_: groot.vla.data.transform.VideoColorJitter
+    apply_to:
+    - video.egocentric
+    brightness: 0.3
+    contrast: 0.4
+    saturation: 0.5
+    hue: 0.08
+  - _target_: groot.vla.data.transform.VideoToNumpy
+    apply_to:
+    - video.egocentric
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+    normalization_modes:
+      state.left_hand: q99
+      state.right_hand: q99
+      state.left_arm: q99
+      state.right_arm: q99
+      state.rpy: q99
+      state.height: q99
+  - _target_: groot.vla.data.transform.StateActionToTensor
+    apply_to:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+  - _target_: groot.vla.data.transform.StateActionTransform
+    apply_to:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+    normalization_modes:
+      action.left_hand: q99
+      action.right_hand: q99
+      action.left_arm: q99
+      action.right_arm: q99
+      action.rpy: q99
+      action.height: q99
+      action.torso_vx: q99
+      action.torso_vy: q99
+      action.torso_vyaw: q99
+      action.torso_dyaw: q99
+  - _target_: groot.vla.data.transform.ConcatTransform
+    video_concat_order:
+    - video.egocentric
+    state_concat_order:
+    - state.left_hand
+    - state.right_hand
+    - state.left_arm
+    - state.right_arm
+    - state.rpy
+    - state.height
+    action_concat_order:
+    - action.left_hand
+    - action.right_hand
+    - action.left_arm
+    - action.right_arm
+    - action.rpy
+    - action.height
+    - action.torso_vx
+    - action.torso_vy
+    - action.torso_vyaw
+    - action.torso_dyaw
+  - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+    default_instruction: Perform the default behavior.
+    language_dropout_prob: 0.0
+    always_use_default_instruction: false
+    max_state_dim: 64
+    max_action_dim: 36
+    max_length: 512
+    state_horizon: 1
+    action_horizon: 24
+    embodiment_tag_mapping:
+      real_gr1_arms_only: 0
+      real_gr1_arms_only_annotated: 1
+      real_gr1_arms_waist: 2
+      real_gr1_arms_waist_annotated: 3
+      dexmg_gr1_arms_only_inspire: 4
+      dexmg_gr1_arms_only_fourier: 5
+      dexmg_gr1_arms_waist_fourier: 6
+      robocasa_single_arm: 7
+      onex_eve_gripper: 8
+      robocasa_gr1_arms_only_inspire_hands: 9
+      robocasa_gr1_arms_only_fourier_hands: 10
+      robocasa_gr1_fixed_lower_body_inspire_hands: 11
+      robocasa_gr1_fixed_lower_body_fourier_hands: 12
+      robocasa_panda_omron: 13
+      robocasa_bimanual_panda_parallel_gripper: 15
+      robocasa_bimanual_panda_inspire_hand: 16
+      oxe_droid: 17
+      oxe_fractal: 18
+      oxe_language_table: 19
+      oxe_bridge: 20
+      real_panda_single_arm: 21
+      hot3d_hands_only: 23
+      gr1_unified: 24
+      robocasa_gr1_arms_waist_fourier_hands: 25
+      agibot: 26
+      lapa: 27
+      oxe_mutex: 28
+      oxe_roboset: 29
+      oxe_plex: 30
+      dream: 31
+      yam: 32
+      xdof: 22
+      gr1_unified_segmentation: 14
+      language_table_sim: 7
+      gr1_isaac: 0
+      sim_behavior_r1_pro: 31
+      mecka_hands: 27
+      real_r1_pro_sharpa: 28
+      real_teleop_g1: 33
+    tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+modality_configs:
+  oxe_droid:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - 0
+      modality_keys:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.joint_position
+      - state.gripper_position
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.joint_position
+      - action.gripper_position
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.language.language_instruction
+      - annotation.language.language_instruction_2
+      - annotation.language.language_instruction_3
+    lapa_action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - lapa_action
+  agibot:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - -3
+      - -2
+      - -1
+      - 0
+      modality_keys:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.language.action_text
+  yam:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - 0
+      modality_keys:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.task
+  real_teleop_g1:
+    video:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      - 24
+      eval_delta_indices:
+      - 0
+      modality_keys:
+      - video.egocentric
+    state:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+    action:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      - 4
+      - 5
+      - 6
+      - 7
+      - 8
+      - 9
+      - 10
+      - 11
+      - 12
+      - 13
+      - 14
+      - 15
+      - 16
+      - 17
+      - 18
+      - 19
+      - 20
+      - 21
+      - 22
+      - 23
+      modality_keys:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+    language:
+      _target_: groot.vla.data.dataset.ModalityConfig
+      delta_indices:
+      - 0
+      modality_keys:
+      - annotation.language.language_instruction
+transforms:
+  oxe_droid:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.joint_position
+      - state.gripper_position
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.joint_position
+      - state.gripper_position
+      normalization_modes:
+        state.joint_position: q99
+        state.gripper_position: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.joint_position
+      - action.gripper_position
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.joint_position
+      - action.gripper_position
+      normalization_modes:
+        action.joint_position: q99
+        action.gripper_position: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.exterior_image_1_left
+      - video.exterior_image_2_left
+      - video.wrist_image_left
+      state_concat_order:
+      - state.joint_position
+      - state.gripper_position
+      action_concat_order:
+      - action.joint_position
+      - action.gripper_position
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  agibot:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+      normalization_modes:
+        state.left_arm_joint_position: q99
+        state.right_arm_joint_position: q99
+        state.left_effector_position: q99
+        state.right_effector_position: q99
+        state.head_position: q99
+        state.waist_position: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+      normalization_modes:
+        action.left_arm_joint_position: q99
+        action.right_arm_joint_position: q99
+        action.left_effector_position: q99
+        action.right_effector_position: q99
+        action.head_position: q99
+        action.waist_position: q99
+        action.robot_velocity: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.top_head
+      - video.hand_left
+      - video.hand_right
+      state_concat_order:
+      - state.left_arm_joint_position
+      - state.right_arm_joint_position
+      - state.left_effector_position
+      - state.right_effector_position
+      - state.head_position
+      - state.waist_position
+      action_concat_order:
+      - action.left_arm_joint_position
+      - action.right_arm_joint_position
+      - action.left_effector_position
+      - action.right_effector_position
+      - action.head_position
+      - action.waist_position
+      - action.robot_velocity
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  yam:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+      normalization_modes:
+        state.left_joint_pos: q99
+        state.left_gripper_pos: q99
+        state.right_joint_pos: q99
+        state.right_gripper_pos: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+      normalization_modes:
+        action.left_joint_pos: q99
+        action.left_gripper_pos: q99
+        action.right_joint_pos: q99
+        action.right_gripper_pos: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.top_camera-images-rgb
+      - video.left_camera-images-rgb
+      - video.right_camera-images-rgb
+      state_concat_order:
+      - state.left_joint_pos
+      - state.left_gripper_pos
+      - state.right_joint_pos
+      - state.right_gripper_pos
+      action_concat_order:
+      - action.left_joint_pos
+      - action.left_gripper_pos
+      - action.right_joint_pos
+      - action.right_gripper_pos
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+  real_teleop_g1:
+    _target_: groot.vla.data.transform.ComposedModalityTransform
+    transforms:
+    - _target_: groot.vla.data.transform.VideoToTensor
+      apply_to:
+      - video.egocentric
+    - _target_: groot.vla.data.transform.VideoCrop
+      apply_to:
+      - video.egocentric
+      scale: 0.95
+      mode: random
+    - _target_: groot.vla.data.transform.VideoResize
+      apply_to:
+      - video.egocentric
+      height: 176
+      width: 320
+      interpolation: linear
+    - _target_: groot.vla.data.transform.VideoColorJitter
+      apply_to:
+      - video.egocentric
+      brightness: 0.3
+      contrast: 0.4
+      saturation: 0.5
+      hue: 0.08
+    - _target_: groot.vla.data.transform.VideoToNumpy
+      apply_to:
+      - video.egocentric
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+      normalization_modes:
+        state.left_hand: q99
+        state.right_hand: q99
+        state.left_arm: q99
+        state.right_arm: q99
+        state.rpy: q99
+        state.height: q99
+    - _target_: groot.vla.data.transform.StateActionToTensor
+      apply_to:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+    - _target_: groot.vla.data.transform.StateActionTransform
+      apply_to:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+      normalization_modes:
+        action.left_hand: q99
+        action.right_hand: q99
+        action.left_arm: q99
+        action.right_arm: q99
+        action.rpy: q99
+        action.height: q99
+        action.torso_vx: q99
+        action.torso_vy: q99
+        action.torso_vyaw: q99
+        action.torso_dyaw: q99
+    - _target_: groot.vla.data.transform.ConcatTransform
+      video_concat_order:
+      - video.egocentric
+      state_concat_order:
+      - state.left_hand
+      - state.right_hand
+      - state.left_arm
+      - state.right_arm
+      - state.rpy
+      - state.height
+      action_concat_order:
+      - action.left_hand
+      - action.right_hand
+      - action.left_arm
+      - action.right_arm
+      - action.rpy
+      - action.height
+      - action.torso_vx
+      - action.torso_vy
+      - action.torso_vyaw
+      - action.torso_dyaw
+    - _target_: groot.vla.model.dreamzero.transform.dreamzero_cotrain.DreamTransform
+      default_instruction: Perform the default behavior.
+      language_dropout_prob: 0.0
+      always_use_default_instruction: false
+      max_state_dim: 64
+      max_action_dim: 36
+      max_length: 512
+      state_horizon: 1
+      action_horizon: 24
+      embodiment_tag_mapping:
+        real_gr1_arms_only: 0
+        real_gr1_arms_only_annotated: 1
+        real_gr1_arms_waist: 2
+        real_gr1_arms_waist_annotated: 3
+        dexmg_gr1_arms_only_inspire: 4
+        dexmg_gr1_arms_only_fourier: 5
+        dexmg_gr1_arms_waist_fourier: 6
+        robocasa_single_arm: 7
+        onex_eve_gripper: 8
+        robocasa_gr1_arms_only_inspire_hands: 9
+        robocasa_gr1_arms_only_fourier_hands: 10
+        robocasa_gr1_fixed_lower_body_inspire_hands: 11
+        robocasa_gr1_fixed_lower_body_fourier_hands: 12
+        robocasa_panda_omron: 13
+        robocasa_bimanual_panda_parallel_gripper: 15
+        robocasa_bimanual_panda_inspire_hand: 16
+        oxe_droid: 17
+        oxe_fractal: 18
+        oxe_language_table: 19
+        oxe_bridge: 20
+        real_panda_single_arm: 21
+        hot3d_hands_only: 23
+        gr1_unified: 24
+        robocasa_gr1_arms_waist_fourier_hands: 25
+        agibot: 26
+        lapa: 27
+        oxe_mutex: 28
+        oxe_roboset: 29
+        oxe_plex: 30
+        dream: 31
+        yam: 32
+        xdof: 22
+        gr1_unified_segmentation: 14
+        language_table_sim: 7
+        gr1_isaac: 0
+        sim_behavior_r1_pro: 31
+        mecka_hands: 27
+        real_r1_pro_sharpa: 28
+        real_teleop_g1: 33
+      tokenizer_path: /hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+metadata_versions:
+  oxe_droid: '0221'
+  agibot: '0221'
+  yam: '0221'
+  real_teleop_g1: '0221'
+fps:
+  yam: 30
+  real_teleop_g1: 30
+relative_action: true
+relative_action_per_horizon: false
+relative_action_keys:
+- left_hand
+- right_hand
+- left_arm
+- right_arm
+- rpy
+- height
+max_chunk_size: 4
+dataset_shard_sampling_rate: 0.1
+mixture_dataset_cls: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotMixtureDataset.from_mixture_spec
+single_dataset_cls: groot.vla.data.dataset.lerobot_sharded.ShardedLeRobotSubLangSingleActionChunkDatasetDROID
+real_teleop_g1_data_root: /hfm/boqian/liboqian_data/data/real_data/gear/g1/Pick_bottle_and_turn_and_pour_into_cup
+total_training_steps: 1048576000000

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/experiment_cfg/metadata.json

@@ -0,0 +1,787 @@
+{
+    "real_teleop_g1": {
+        "statistics": {
+            "state": {
+                "left_hand": {
+                    "max": [
+                        -0.15012885630130768,
+                        0.9005824327468872,
+                        1.240552306175232,
+                        -0.001318894443102181,
+                        -0.029013408347964287,
+                        -0.0026860591024160385,
+                        -0.032535798847675323
+                    ],
+                    "min": [
+                        -1.0365108251571655,
+                        0.3635019063949585,
+                        0.6077943444252014,
+                        -0.8983967900276184,
+                        -1.4610702991485596,
+                        -0.9210805296897888,
+                        -1.2054297924041748
+                    ],
+                    "mean": [
+                        -0.7106521057783574,
+                        0.7397562400548329,
+                        0.7932539759105839,
+                        -0.18307749281402536,
+                        -0.3996594926481087,
+                        -0.22440346846797105,
+                        -0.3648747590553514
+                    ],
+                    "std": [
+                        0.1800224325521231,
+                        0.08892492705684825,
+                        0.17347929747061464,
+                        0.1250548816433564,
+                        0.14322181312600368,
+                        0.12746796614686143,
+                        0.1004095686025728
+                    ],
+                    "q01": [
+                        -1.0010048174858093,
+                        0.5051653385162354,
+                        0.6091835498809814,
+                        -0.5192373794317245,
+                        -0.7626793467998505,
+                        -0.5990579128265381,
+                        -0.6296854817867279
+                    ],
+                    "q99": [
+                        -0.35215023159980774,
+                        0.8828813862800599,
+                        1.1794202327728271,
+                        -0.012645758632570512,
+                        -0.08962718859314919,
+                        -0.01236837636679411,
+                        -0.14306417107582092
+                    ]
+                },
+                "right_hand": {
+                    "max": [
+                        0.013422729447484016,
+                        -0.0029037927743047476,
+                        -0.4307878315448761,
+                        1.1691919565200806,
+                        1.5508151054382324,
+                        1.008196473121643,
+                        1.4388744831085205
+                    ],
+                    "min": [
+                        -0.9901090860366821,
+                        -0.8877645134925842,
+                        -1.2071785926818848,
+                        -0.03135598078370094,
+                        0.03392980992794037,
+                        0.023357456550002098,
+                        0.019666638225317
+                    ],
+                    "mean": [
+                        -0.4129581264587618,
+                        -0.4582520757612821,
+                        -0.664123836057758,
+                        0.4952124991587111,
+                        0.9231042625678995,
+                        0.5022273107153479,
+                        0.669409344487608
+                    ],
+                    "std": [
+                        0.19006874361243062,
+                        0.11976727609652792,
+                        0.10119243113959121,
+                        0.2266161525462106,
+                        0.4655465641238765,
+                        0.21432269096327275,
+                        0.31467715432459836
+                    ],
+                    "q01": [
+                        -0.9081442606449127,
+                        -0.8110105991363525,
+                        -0.9831316405534745,
+                        0.023696813061833383,
+                        0.037188550531864165,
+                        0.024981562048196793,
+                        0.02064503453671932
+                    ],
+                    "q99": [
+                        -0.10628944709897117,
+                        -0.18123969539999962,
+                        -0.5385138392448425,
+                        0.9067972785234452,
+                        1.4886025190353394,
+                        0.871771514415741,
+                        1.3075256443023682
+                    ]
+                },
+                "left_arm": {
+                    "max": [
+                        0.6784030199050903,
+                        0.5408600568771362,
+                        0.261579692363739,
+                        1.2853081226348877,
+                        0.6677011251449585,
+                        0.01986505277454853,
+                        0.5779042840003967
+                    ],
+                    "min": [
+                        -0.6697743535041809,
+                        -0.04027898237109184,
+                        -0.6574546098709106,
+                        -0.02732403390109539,
+                        -0.18146513402462006,
+                        -1.2264176607131958,
+                        -0.8434670567512512
+                    ],
+                    "mean": [
+                        0.06576749867032948,
+                        0.25579782742676094,
+                        -0.06004482964572005,
+                        0.808768983211113,
+                        0.10850686643277818,
+                        -0.7254681426808276,
+                        -0.2878231007376268
+                    ],
+                    "std": [
+                        0.16998997778227107,
+                        0.09886796724754655,
+                        0.1532526690961023,
+                        0.3788245539398352,
+                        0.1304561854786525,
+                        0.3305712410884026,
+                        0.20458272924414123
+                    ],
+                    "q01": [
+                        -0.472993403673172,
+                        0.01923264417797327,
+                        -0.4175487527251244,
+                        0.11523827515542508,
+                        -0.13969017088413238,
+                        -1.190878005027771,
+                        -0.7066819667816162
+                    ],
+                    "q99": [
+                        0.4539884480834007,
+                        0.5193723440170288,
+                        0.2179892195761203,
+                        1.2710349559783936,
+                        0.6194926702976227,
+                        -0.09952061586081994,
+                        0.17473116487264606
+                    ]
+                },
+                "right_arm": {
+                    "max": [
+                        0.9002070426940918,
+                        0.11829628795385361,
+                        0.8515990376472473,
+                        1.2844573259353638,
+                        0.8022719621658325,
+                        0.6399767994880676,
+                        1.2358516454696655
+                    ],
+                    "min": [
+                        -1.2026649713516235,
+                        -0.6969066858291626,
+                        -0.815610408782959,
+                        -0.9874761700630188,
+                        -1.6946533918380737,
+                        -1.4213329553604126,
+                        -1.342279314994812
+                    ],
+                    "mean": [
+                        -0.0662775668920557,
+                        -0.1345353524434497,
+                        0.0634982026036378,
+                        -0.10989526924031702,
+                        -0.05263142642975812,
+                        -0.12096835662218453,
+                        0.120861711803444
+                    ],
+                    "std": [
+                        0.4008494462356859,
+                        0.1700786431413106,
+                        0.21295159275658426,
+                        0.43649950708158436,
+                        0.4497950790039162,
+                        0.36514804111543175,
+                        0.1967684071659634
+                    ],
+                    "q01": [
+                        -0.950262793302536,
+                        -0.5722128254175186,
+                        -0.4087426000833511,
+                        -0.8998615843057632,
+                        -1.200806084871292,
+                        -1.1520812797546387,
+                        -0.3325912955403328
+                    ],
+                    "q99": [
+                        0.7100517773628232,
+                        0.10524750150740145,
+                        0.6194546324014664,
+                        0.9450081789493561,
+                        0.559214276075363,
+                        0.5243198603391647,
+                        0.6973656332492828
+                    ]
+                },
+                "rpy": {
+                    "max": [
+                        0.1827869415283203,
+                        0.12582087516784668,
+                        0.7600905299186707
+                    ],
+                    "min": [
+                        -0.25714462995529175,
+                        -0.27553248405456543,
+                        -0.3924732208251953
+                    ],
+                    "mean": [
+                        -0.07450471914543803,
+                        -0.016358907944713997,
+                        0.08461220301037418
+                    ],
+                    "std": [
+                        0.052686434459038575,
+                        0.05521563296798423,
+                        0.1871730371455983
+                    ],
+                    "q01": [
+                        -0.21652273923158646,
+                        -0.15164137467741967,
+                        -0.3411825066804886
+                    ],
+                    "q99": [
+                        0.023736179061233972,
+                        0.0924126328527927,
+                        0.6843682318925858
+                    ]
+                },
+                "height": {
+                    "max": [
+                        0.75
+                    ],
+                    "min": [
+                        0.6516909599304199
+                    ],
+                    "mean": [
+                        0.717019414027915
+                    ],
+                    "std": [
+                        0.010611909948945071
+                    ],
+                    "q01": [
+                        0.681534378528595
+                    ],
+                    "q99": [
+                        0.7377596861124038
+                    ]
+                }
+            },
+            "action": {
+                "left_hand": {
+                    "max": [
+                        0.5982731580734253,
+                        0.5278768539428711,
+                        0.5354012846946716,
+                        0.4014817903516814,
+                        0.4918059706687927,
+                        0.44213195890188217,
+                        0.4230584502220154
+                    ],
+                    "min": [
+                        -0.6301521956920624,
+                        -0.5149323344230652,
+                        -0.529521644115448,
+                        -0.6856191083788872,
+                        -1.158537894487381,
+                        -0.7385352402925491,
+                        -0.905200719833374
+                    ],
+                    "mean": [
+                        -0.017651338260789542,
+                        0.00866811522079485,
+                        -0.004330770698072518,
+                        -0.002544540625456693,
+                        -0.0005896838413398195,
+                        0.002361588407552847,
+                        0.004366535660068055
+                    ],
+                    "std": [
+                        0.06554604732418758,
+                        0.07154837844491334,
+                        0.12856598959026014,
+                        0.03597542702397461,
+                        0.04581648449077367,
+                        0.03962922545114253,
+                        0.04346845484086631
+                    ],
+                    "q01": [
+                        -0.24043308198451993,
+                        -0.19401812851428984,
+                        -0.39614349603652954,
+                        -0.10097884107381103,
+                        -0.14891792684793473,
+                        -0.11517959907650946,
+                        -0.1279512584209442
+                    ],
+                    "q99": [
+                        0.23256101012229896,
+                        0.20524268746376023,
+                        0.3933871388435364,
+                        0.09811923948582232,
+                        0.12937735170125958,
+                        0.1224899165794337,
+                        0.11734045296907396
+                    ]
+                },
+                "right_hand": {
+                    "max": [
+                        0.8790446370840073,
+                        0.642131395637989,
+                        0.5467694997787476,
+                        1.2996619567275047,
+                        1.698824219405651,
+                        1.07933783903718,
+                        1.2983299642801285
+                    ],
+                    "min": [
+                        -0.8118728846311569,
+                        -0.6551126688718796,
+                        -0.8104255795478821,
+                        -0.940872801351361,
+                        -1.4264521131990477,
+                        -0.9354669973254204,
+                        -0.9567366391420364
+                    ],
+                    "mean": [
+                        0.06989124645612069,
+                        -0.18283122662561244,
+                        -0.07939327808296306,
+                        0.13485396014136192,
+                        0.1019227924721336,
+                        0.12847191988697357,
+                        0.06420547800869135
+                    ],
+                    "std": [
+                        0.14218628429822006,
+                        0.14446828671114784,
+                        0.1265010717040926,
+                        0.1855492922965264,
+                        0.2240967671588938,
+                        0.17045304446399806,
+                        0.1757324556093506
+                    ],
+                    "q01": [
+                        -0.35978180170059204,
+                        -0.4319852590560913,
+                        -0.36317331790924073,
+                        -0.523832806199789,
+                        -0.6789782047271729,
+                        -0.4287099301815033,
+                        -0.4532649278640747
+                    ],
+                    "q99": [
+                        0.4718186043202869,
+                        0.19760750830173343,
+                        0.25330804288387265,
+                        0.6829912588000264,
+                        0.8806305289268437,
+                        0.6021148726344102,
+                        0.6443318486213678
+                    ]
+                },
+                "left_arm": {
+                    "max": [
+                        0.48457716405391693,
+                        0.3535273000597954,
+                        0.21352535486221313,
+                        0.7616467773914337,
+                        0.6007281094789505,
+                        0.41783052682876587,
+                        0.5165230333805084
+                    ],
+                    "min": [
+                        -0.6373330056667328,
+                        -0.15849913656711578,
+                        -0.5613232888281345,
+                        -0.701464831829071,
+                        -0.2516954243183136,
+                        -0.8871668800711632,
+                        -0.4376880154013634
+                    ],
+                    "mean": [
+                        -0.011776995941582026,
+                        0.00991253342440134,
+                        -0.006852865379075551,
+                        0.010552169557516157,
+                        0.001214240309897302,
+                        -0.014453163590611127,
+                        -0.010611657202882456
+                    ],
+                    "std": [
+                        0.03602045474094356,
+                        0.024036875417496602,
+                        0.02602605089260403,
+                        0.05981475365626236,
+                        0.028429534988341422,
+                        0.08916377427180806,
+                        0.03528055113076741
+                    ],
+                    "q01": [
+                        -0.1545458011329174,
+                        -0.031217777729034422,
+                        -0.11606558486819267,
+                        -0.0627044416964054,
+                        -0.056227393448352814,
+                        -0.5197785004973411,
+                        -0.156750052748248
+                    ],
+                    "q99": [
+                        0.05294616930186727,
+                        0.11603573272004716,
+                        0.02942010760307312,
+                        0.32693205326795444,
+                        0.11353727197274502,
+                        0.06767475279048074,
+                        0.06325854659080493
+                    ]
+                },
+                "right_arm": {
+                    "max": [
+                        1.0019388496875763,
+                        0.6778036952018738,
+                        0.8202316761016846,
+                        1.2868973910808563,
+                        1.6054365634918213,
+                        0.9513278678059578,
+                        1.0065587792778388
+                    ],
+                    "min": [
+                        -0.8566206395626068,
+                        -0.5352737102657557,
+                        -0.9050570726394653,
+                        -1.1766326129436493,
+                        -1.1665689051151276,
+                        -1.226991206407547,
+                        -1.4597684778273106
+                    ],
+                    "mean": [
+                        0.0011831602733654706,
+                        0.004668234701338991,
+                        -0.00464666805688978,
+                        -0.008819726561502726,
+                        0.01680292255365519,
+                        -0.023272413508623503,
+                        0.001828295062135536
+                    ],
+                    "std": [
+                        0.15114015459104163,
+                        0.10717935148199742,
+                        0.12210241645133885,
+                        0.18724624428113243,
+                        0.21873655609526446,
+                        0.18543346050682308,
+                        0.1150104843557695
+                    ],
+                    "q01": [
+                        -0.39952214881777764,
+                        -0.25475079081952573,
+                        -0.4339929953217506,
+                        -0.645755667425692,
+                        -0.5597020275890827,
+                        -0.640592061728239,
+                        -0.3658384716138244
+                    ],
+                    "q99": [
+                        0.47152970274910266,
+                        0.3625141793861984,
+                        0.3024886786937714,
+                        0.5137429375201458,
+                        0.7895279735326737,
+                        0.5061310570687053,
+                        0.30551581159233987
+                    ]
+                },
+                "rpy": {
+                    "max": [
+                        0.2602043002843857,
+                        0.0978180319070816,
+                        0.4442971870303154
+                    ],
+                    "min": [
+                        -0.2339760884642601,
+                        -0.27553248405456543,
+                        -0.44536447897553444
+                    ],
+                    "mean": [
+                        -0.0005869882495184713,
+                        0.0015875524968715301,
+                        0.0008563187639556434
+                    ],
+                    "std": [
+                        0.015963769360634213,
+                        0.009780051628911945,
+                        0.024366577111250988
+                    ],
+                    "q01": [
+                        -0.049293445050716395,
+                        -0.023501918860711155,
+                        -0.0595750443637371
+                    ],
+                    "q99": [
+                        0.05025968039408321,
+                        0.028102314658462975,
+                        0.0824947778135533
+                    ]
+                },
+                "height": {
+                    "max": [
+                        0.04315239191055298
+                    ],
+                    "min": [
+                        -0.08728241920471191
+                    ],
+                    "mean": [
+                        5.191507986045299e-05
+                    ],
+                    "std": [
+                        0.004163189516331377
+                    ],
+                    "q01": [
+                        -0.012046998739242552
+                    ],
+                    "q99": [
+                        0.012092745304107644
+                    ]
+                },
+                "torso_vx": {
+                    "max": [
+                        0.3499999940395355
+                    ],
+                    "min": [
+                        0.0
+                    ],
+                    "mean": [
+                        0.017269024247052768
+                    ],
+                    "std": [
+                        0.07580197349073346
+                    ],
+                    "q01": [
+                        0.0
+                    ],
+                    "q99": [
+                        0.3499999940395355
+                    ]
+                },
+                "torso_vy": {
+                    "max": [
+                        0.5
+                    ],
+                    "min": [
+                        0.0
+                    ],
+                    "mean": [
+                        0.0002320065992988245
+                    ],
+                    "std": [
+                        0.01076798368252486
+                    ],
+                    "q01": [
+                        0.0
+                    ],
+                    "q99": [
+                        0.0
+                    ]
+                },
+                "torso_vyaw": {
+                    "max": [
+                        0.5
+                    ],
+                    "min": [
+                        -0.5
+                    ],
+                    "mean": [
+                        -0.03733115137363985
+                    ],
+                    "std": [
+                        0.1199053455233051
+                    ],
+                    "q01": [
+                        -0.5
+                    ],
+                    "q99": [
+                        0.0
+                    ]
+                },
+                "torso_dyaw": {
+                    "max": [
+                        0.3072107136249542
+                    ],
+                    "min": [
+                        -0.1875186562538147
+                    ],
+                    "mean": [
+                        0.016342195046893805
+                    ],
+                    "std": [
+                        0.048962870720188865
+                    ],
+                    "q01": [
+                        -0.005849878406152129
+                    ],
+                    "q99": [
+                        0.22258896410465237
+                    ]
+                }
+            }
+        },
+        "modalities": {
+            "video": {
+                "egocentric": {
+                    "resolution": [
+                        640,
+                        480
+                    ],
+                    "channels": 3,
+                    "fps": 30.0
+                }
+            },
+            "state": {
+                "left_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "left_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "rpy": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        3
+                    ],
+                    "continuous": true
+                },
+                "height": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                }
+            },
+            "action": {
+                "left_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_hand": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "left_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "right_arm": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        7
+                    ],
+                    "continuous": true
+                },
+                "rpy": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        3
+                    ],
+                    "continuous": true
+                },
+                "height": {
+                    "absolute": true,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_vx": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_vy": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_vyaw": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                },
+                "torso_dyaw": {
+                    "absolute": false,
+                    "rotation_type": null,
+                    "shape": [
+                        1
+                    ],
+                    "continuous": true
+                }
+            }
+        },
+        "embodiment_tag": "real_teleop_g1"
+    }
+}

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/wandb/debug-internal.log

@@ -0,0 +1,20 @@
+{"time":"2026-03-19T11:07:03.568457075Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-19T11:07:03.817295367Z","level":"INFO","msg":"stream: created new stream","id":"lsme06f2"}
+{"time":"2026-03-19T11:07:03.817394477Z","level":"INFO","msg":"handler: started","stream_id":"lsme06f2"}
+{"time":"2026-03-19T11:07:03.817531838Z","level":"INFO","msg":"stream: started","id":"lsme06f2"}
+{"time":"2026-03-19T11:07:03.817558849Z","level":"INFO","msg":"sender: started","stream_id":"lsme06f2"}
+{"time":"2026-03-19T11:07:03.817564649Z","level":"INFO","msg":"writer: started","stream_id":"lsme06f2"}
+{"time":"2026-03-20T18:00:34.85408512Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"\n<html><head>\n<meta http-equiv=\"content-type\" content=\"text/html;charset=utf-8\">\n<title>502 Server Error</title>\n</head>\n<body text=#000000 bgcolor=#ffffff>\n<h1>Error: Server Error</h1>\n<h2>The server encountered a temporary error and could not complete your request.<p>Please try again in 30 seconds.</h2>\n<h2></h2>\n</body></html>\n"}
+{"time":"2026-03-20T20:47:26.570329788Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"\n<html><head>\n<meta http-equiv=\"content-type\" content=\"text/html;charset=utf-8\">\n<title>502 Server Error</title>\n</head>\n<body text=#000000 bgcolor=#ffffff>\n<h1>Error: Server Error</h1>\n<h2>The server encountered a temporary error and could not complete your request.<p>Please try again in 30 seconds.</h2>\n<h2></h2>\n</body></html>\n"}
+{"time":"2026-03-20T20:48:53.069085604Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"{\"error\":\"context deadline exceeded\"}"}
+{"time":"2026-03-20T20:56:21.788979982Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"{\"error\":\"context deadline exceeded\"}"}
+{"time":"2026-03-20T20:58:30.788847673Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"{\"error\":\"context deadline exceeded\"}"}
+{"time":"2026-03-20T21:01:49.472533676Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"{\"error\":\"context deadline exceeded\"}"}
+{"time":"2026-03-20T21:03:33.794344816Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"{\"error\":\"context deadline exceeded\"}"}
+{"time":"2026-03-20T21:04:30.913563045Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"{\"error\":\"context deadline exceeded\"}"}
+{"time":"2026-03-20T21:06:13.266987441Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/files/boqianli/dreamzero/lsme06f2/file_stream","body":"{\"error\":\"context deadline exceeded\"}"}
+{"time":"2026-03-23T07:12:08.820179122Z","level":"INFO","msg":"stream: closing","id":"lsme06f2"}
+{"time":"2026-03-23T07:12:10.471712011Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-03-23T07:12:10.6442904Z","level":"INFO","msg":"handler: closed","stream_id":"lsme06f2"}
+{"time":"2026-03-23T07:12:10.644415291Z","level":"INFO","msg":"sender: closed","stream_id":"lsme06f2"}
+{"time":"2026-03-23T07:12:10.644451071Z","level":"INFO","msg":"stream: closed","id":"lsme06f2"}

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/wandb/debug.log

@@ -0,0 +1,24 @@
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_setup.py:_flush():81] Current SDK version is 0.25.0
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_setup.py:_flush():81] Configure stats pid to 1976195
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_init.py:setup_run_log_directory():717] Logging user logs to checkpoints/dreamzero_real_teleop_g1_full_finetune/wandb/run-20260319_110703-lsme06f2/logs/debug.log
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_init.py:setup_run_log_directory():718] Logging internal logs to checkpoints/dreamzero_real_teleop_g1_full_finetune/wandb/run-20260319_110703-lsme06f2/logs/debug-internal.log
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_init.py:init():844] calling init triggers
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_init.py:init():849] wandb.init called with sweep_config: {}
+config: {'_wandb': {}}
+2026-03-19 11:07:03,312 INFO    MainThread:1976195 [wandb_init.py:init():892] starting backend
+2026-03-19 11:07:03,562 INFO    MainThread:1976195 [wandb_init.py:init():895] sending inform_init request
+2026-03-19 11:07:03,564 INFO    MainThread:1976195 [wandb_init.py:init():903] backend started and connected
+2026-03-19 11:07:03,565 INFO    MainThread:1976195 [wandb_init.py:init():973] updated telemetry
+2026-03-19 11:07:03,570 INFO    MainThread:1976195 [wandb_init.py:init():997] communicating run to backend with 90.0 second timeout
+2026-03-19 11:07:04,794 INFO    MainThread:1976195 [wandb_init.py:init():1042] starting run threads in backend
+2026-03-19 11:07:04,941 INFO    MainThread:1976195 [wandb_run.py:_console_start():2524] atexit reg
+2026-03-19 11:07:04,942 INFO    MainThread:1976195 [wandb_run.py:_redirect():2373] redirect: wrap_raw
+2026-03-19 11:07:04,942 INFO    MainThread:1976195 [wandb_run.py:_redirect():2442] Wrapping output streams.
+2026-03-19 11:07:04,942 INFO    MainThread:1976195 [wandb_run.py:_redirect():2465] Redirects installed.
+2026-03-19 11:07:04,944 INFO    MainThread:1976195 [wandb_init.py:init():1082] run started, returning control to user process
+2026-03-19 11:07:04,946 INFO    MainThread:1976195 [wandb_run.py:_config_callback():1403] config_cb None None {'return_dict': True, 'output_hidden_states': False, 'output_attentions': False, 'torchscript': False, 'torch_dtype': None, 'use_bfloat16': False, 'tf_legacy_loss': False, 'pruned_heads': {}, 'tie_word_embeddings': True, 'chunk_size_feed_forward': 0, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'num_beam_groups': 1, 'diversity_penalty': 0.0, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'suppress_tokens': None, 'begin_suppress_tokens': None, 'architectures': None, 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': None, 'pad_token_id': None, 'eos_token_id': None, 'sep_token_id': None, 'decoder_start_token_id': None, 'task_specific_params': None, 'problem_type': None, '_name_or_path': './checkpoints/dreamzero_real_teleop_g1_full_finetune', '_attn_implementation_autoset': True, 'transformers_version': '4.51.3', 'model_dtype': 'float32', 'hidden_size': 0, 'action_horizon': 24, 'action_dim': 36, 'backbone_cfg': {'_target_': 'groot.vla.model.dreamzero.backbone.identity.IdentityBackbone'}, 'action_head_cfg': {'config': {'backbone_features_projector_cfg': None, '_target_': 'groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHeadConfig', '_recursive_': False, 'tiled': False, 'tile_size_height': 34, 'tile_size_width': 34, 'tile_stride_height': 18, 'tile_stride_width': 16, 'lora_rank': 4, 'lora_alpha': 4, 'num_frames': 33, 'num_frame_per_block': 2, 'lora_target_modules': 'q,k,v,o,ffn.0,ffn.2', 'init_lora_weights': 'kaiming', 'train_architecture': 'full', 'use_gradient_checkpointing': True, 'add_pos_embed': True, 'model_dtype': 'float32', 'max_state_dim': 64, 'max_action_dim': 36, 'action_loss_embodiment_ids': [26, 17, 32], 'hidden_size': 64, 'input_embedding_dim': 1536, 'backbone_embedding_dim': 0, 'repa_layer': 8, 'repa_coeff': 1.0, 'load_pretrained_det_decode_layer_path': None, 'freeze_decode_layer': False, 'expand_batch': None, 'use_vlln': True, 'vl_self_attention_cfg': {'_target_': 'groot.vla.model.n1_5.modules.cross_attention_dit.SelfAttentionTransformer', 'positional_embeddings': None, 'num_layers': 4, 'num_attention_heads': 24, 'attention_head_dim': 64, 'dropout': 0.2, 'final_dropout': True}, 'diffusion_model_cfg': {'_target_': 'groot.vla.model.dreamzero.modules.wan_video_dit_action_casual_chunk.CausalWanModel', '_convert_': 'object', 'diffusion_model_pretrained_path': '/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P', 'model_type': 'i2v', 'frame_seqlen': 220, 'dim': 5120, 'in_dim': 36, 'ffn_dim': 13824, 'out_dim': 16, 'freq_dim': 256, 'eps': 1e-06, 'num_heads': 40, 'num_layers': 40, 'max_chunk_size': 4, 'num_frame_per_block': 2, 'num_action_per_block': 24, 'num_state_per_block': 1, 'action_dim': 36}, 'text_encoder_cfg': {'_target_': 'groot.vla.model.dreamzero.modules.wan_video_text_encoder.WanTextEncoder', '_convert_': 'object', 'text_encoder_pretrained_path': '/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth'}, 'image_encoder_cfg': {'_target_': 'groot.vla.model.dreamzero.modules.wan_video_image_encoder.WanImageEncoder', '_convert_': 'object', 'image_encoder_pretrained_path': '/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth'}, 'vae_cfg': {'_target_': 'groot.vla.model.dreamzero.modules.wan_video_vae.WanVideoVAE', '_convert_': 'object', 'vae_pretrained_path': '/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth'}, 'action_dim': 36, 'action_horizon': 24, 'num_inference_timesteps': 4, 'noise_beta_alpha': 1.5, 'noise_beta_beta': 1.0, 'noise_s': 0.999, 'num_timestep_buckets': 1000, 'decouple_video_action_noise': False, 'video_noise_beta_alpha': 3.0, 'video_noise_beta_beta': 1.0, 'tune_projector': True, 'tune_diffusion_model': True, 'skip_component_loading': True}, '_target_': 'groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHead', '_convert_': 'object'}, 'resume_path': './checkpoints/dreamzero_real_teleop_g1_full_finetune', 'model_type': 'vla', 'output_dir': './checkpoints/dreamzero_real_teleop_g1_full_finetune', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': 'no', 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 64, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 1, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 1e-05, 'adam_beta1': 0.95, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 1000, 'max_steps': 20000, 'lr_scheduler_type': 'cosine', 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': './checkpoints/dreamzero_real_teleop_g1_full_finetune/runs/Mar19_11-00-49_nebula100', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 10, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 8000, 'save_total_limit': 10, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': True, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 1, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'dreamzero_real_teleop_g1_full_finetune', 'disable_tqdm': False, 'remove_unused_columns': False, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': True, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}, 'deepspeed': 'groot/vla/configs/deepspeed/zero2_offload.json', 'label_smoothing_factor': 0.0, 'optim': 'adamw_torch', 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': False, 'ddp_bucket_cap_mb': 100, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': False, 'dataloader_persistent_workers': True, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': None, 'hub_model_id': None, 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': False, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules': None, 'batch_eval_metrics': False, 'eval_on_start': False, 'use_liger_kernel': False, 'eval_use_gather_object': False, 'average_tokens_across_devices': False}
+2026-03-19 11:07:04,952 INFO    MainThread:1976195 [wandb_config.py:__setitem__():155] [no run ID] config set model/num_parameters = 22924196696 - <bound method Run._config_callback of <wandb.sdk.wandb_run.Run object at 0x76265c247450>>
+2026-03-19 11:07:04,952 INFO    MainThread:1976195 [wandb_run.py:_config_callback():1403] config_cb model/num_parameters 22924196696 None
+2026-03-23 07:12:08,813 INFO    wandb-AsyncioManager-main:1976195 [service_client.py:_forward_responses():134] Reached EOF.
+2026-03-23 07:12:08,814 INFO    wandb-AsyncioManager-main:1976195 [mailbox.py:close():155] Closing mailbox, abandoning 1 handles.

dreamzero_real_teleop_g1_full_finetune_relative_Pick_bottle_and_turn_and_pour_into_cup/wandb/run-20260319_110703-lsme06f2/files/config.yaml

@@ -0,0 +1,651 @@
+_attn_implementation_autoset:
+    value: true
+_name_or_path:
+    value: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+_wandb:
+    value:
+        cli_version: 0.25.0
+        e:
+            w0fhsrag0fgp0xlyv0somzfdgnau53vg:
+                args:
+                    - report_to=wandb
+                    - data=dreamzero/real_teleop_g1_relative
+                    - wandb_project=dreamzero
+                    - train_architecture=full
+                    - num_frames=33
+                    - action_horizon=24
+                    - num_views=1
+                    - model=dreamzero/vla
+                    - model/dreamzero/action_head=wan_flow_matching_action_tf
+                    - model/dreamzero/transform=dreamzero_cotrain
+                    - num_frame_per_block=2
+                    - num_action_per_block=24
+                    - num_state_per_block=1
+                    - seed=42
+                    - training_args.learning_rate=1e-5
+                    - training_args.deepspeed=groot/vla/configs/deepspeed/zero2_offload.json
+                    - save_steps=8000
+                    - training_args.warmup_ratio=0.05
+                    - output_dir=./checkpoints/dreamzero_real_teleop_g1_full_finetune
+                    - per_device_train_batch_size=1
+                    - max_steps=20000
+                    - weight_decay=1e-5
+                    - save_total_limit=10
+                    - upload_checkpoints=false
+                    - bf16=true
+                    - tf32=true
+                    - eval_bf16=true
+                    - dataloader_pin_memory=false
+                    - dataloader_num_workers=1
+                    - image_resolution_width=320
+                    - image_resolution_height=176
+                    - save_lora_only=false
+                    - max_chunk_size=4
+                    - frame_seqlen=220
+                    - save_strategy=steps
+                    - real_teleop_g1_data_root=/hfm/boqian/liboqian_data/data/real_data/gear/g1/Pick_bottle_and_turn_and_pour_into_cup
+                    - dit_version=/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+                    - text_encoder_pretrained_path=/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+                    - image_encoder_pretrained_path=/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+                    - vae_pretrained_path=/hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+                    - tokenizer_path=/hfm/boqian/liboqian_data/checkpoints/umt5-xxl
+                    - pretrained_model_path=/hfm/boqian/liboqian_data/checkpoints/DreamZero-AgiBot
+                    - max_action_dim=36
+                    - ++action_head_cfg.config.skip_component_loading=true
+                codePath: groot/vla/experiment/experiment.py
+                codePathLocal: groot/vla/experiment/experiment.py
+                cpu_count: 128
+                cpu_count_logical: 128
+                cudaVersion: "12.9"
+                disk:
+                    /:
+                        total: "3934171283456"
+                        used: "3671307268096"
+                email: 2074894208@qq.com
+                executable: /hfm/boqian/miniconda3/envs/dreamzero/bin/python3.11
+                git:
+                    commit: 71e794d8192d689c24bc5ae85a1bb28a3d423d2f
+                    remote: git@github.com:boqian-li/DreamZero-private.git
+                gpu: NVIDIA A100-SXM4-80GB
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-95155859-2785-a5fa-d493-0c9e9eb3332f
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-3829db0c-5c30-974b-5c88-9ed4f984eddf
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-0e260d0d-d121-6d8c-4c97-6ca0dfce506f
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-b99492cb-2347-0f80-e94f-a95e105eaffc
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-8a143efc-044d-2012-5820-afb258793ffd
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-9fedd571-c6c4-fd5a-1dd7-f82e048386ad
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-8cd44ee0-3344-490d-bb51-ba5b38f2b924
+                    - architecture: Ampere
+                      cudaCores: 6912
+                      memoryTotal: "85899345920"
+                      name: NVIDIA A100-SXM4-80GB
+                      uuid: GPU-55a7cb99-494a-2d02-b9d8-6f85a0614e8c
+                host: nebula100
+                memory:
+                    total: "1082011422720"
+                os: Linux-6.8.0-100-generic-x86_64-with-glibc2.39
+                program: /hfm/boqian/liboqian_code/DreamZero-private/groot/vla/experiment/experiment.py
+                python: CPython 3.11.14
+                root: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+                startedAt: "2026-03-19T11:07:03.311236Z"
+                writerId: w0fhsrag0fgp0xlyv0somzfdgnau53vg
+        m:
+            - "1": train/global_step
+              "6":
+                - 3
+              "7": []
+            - "2": '*'
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.11.14
+        t:
+            "1":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 50
+                - 51
+                - 71
+                - 80
+                - 83
+                - 98
+            "2":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 50
+                - 51
+                - 71
+                - 80
+                - 83
+                - 98
+            "3":
+                - 7
+                - 13
+                - 19
+                - 62
+                - 66
+            "4": 3.11.14
+            "5": 0.25.0
+            "6": 4.51.3
+            "9":
+                "1": transformers_trainer
+            "12": 0.25.0
+            "13": linux-x86_64
+accelerator_config:
+    value:
+        dispatch_batches: null
+        even_batches: true
+        gradient_accumulation_kwargs: null
+        non_blocking: false
+        split_batches: false
+        use_seedable_sampler: true
+action_dim:
+    value: 36
+action_head_cfg:
+    value:
+        _convert_: object
+        _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHead
+        config:
+            _recursive_: false
+            _target_: groot.vla.model.dreamzero.action_head.wan_flow_matching_action_tf.WANPolicyHeadConfig
+            action_dim: 36
+            action_horizon: 24
+            action_loss_embodiment_ids:
+                - 26
+                - 17
+                - 32
+            add_pos_embed: true
+            backbone_embedding_dim: 0
+            backbone_features_projector_cfg: null
+            decouple_video_action_noise: false
+            diffusion_model_cfg:
+                _convert_: object
+                _target_: groot.vla.model.dreamzero.modules.wan_video_dit_action_casual_chunk.CausalWanModel
+                action_dim: 36
+                diffusion_model_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P
+                dim: 5120
+                eps: 1e-06
+                ffn_dim: 13824
+                frame_seqlen: 220
+                freq_dim: 256
+                in_dim: 36
+                max_chunk_size: 4
+                model_type: i2v
+                num_action_per_block: 24
+                num_frame_per_block: 2
+                num_heads: 40
+                num_layers: 40
+                num_state_per_block: 1
+                out_dim: 16
+            expand_batch: null
+            freeze_decode_layer: false
+            hidden_size: 64
+            image_encoder_cfg:
+                _convert_: object
+                _target_: groot.vla.model.dreamzero.modules.wan_video_image_encoder.WanImageEncoder
+                image_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth
+            init_lora_weights: kaiming
+            input_embedding_dim: 1536
+            load_pretrained_det_decode_layer_path: null
+            lora_alpha: 4
+            lora_rank: 4
+            lora_target_modules: q,k,v,o,ffn.0,ffn.2
+            max_action_dim: 36
+            max_state_dim: 64
+            model_dtype: float32
+            noise_beta_alpha: 1.5
+            noise_beta_beta: 1
+            noise_s: 0.999
+            num_frame_per_block: 2
+            num_frames: 33
+            num_inference_timesteps: 4
+            num_timestep_buckets: 1000
+            repa_coeff: 1
+            repa_layer: 8
+            skip_component_loading: true
+            text_encoder_cfg:
+                _convert_: object
+                _target_: groot.vla.model.dreamzero.modules.wan_video_text_encoder.WanTextEncoder
+                text_encoder_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth
+            tile_size_height: 34
+            tile_size_width: 34
+            tile_stride_height: 18
+            tile_stride_width: 16
+            tiled: false
+            train_architecture: full
+            tune_diffusion_model: true
+            tune_projector: true
+            use_gradient_checkpointing: true
+            use_vlln: true
+            vae_cfg:
+                _convert_: object
+                _target_: groot.vla.model.dreamzero.modules.wan_video_vae.WanVideoVAE
+                vae_pretrained_path: /hfm/boqian/liboqian_data/checkpoints/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth
+            video_noise_beta_alpha: 3
+            video_noise_beta_beta: 1
+            vl_self_attention_cfg:
+                _target_: groot.vla.model.n1_5.modules.cross_attention_dit.SelfAttentionTransformer
+                attention_head_dim: 64
+                dropout: 0.2
+                final_dropout: true
+                num_attention_heads: 24
+                num_layers: 4
+                positional_embeddings: null
+action_horizon:
+    value: 24
+adafactor:
+    value: false
+adam_beta1:
+    value: 0.95
+adam_beta2:
+    value: 0.999
+adam_epsilon:
+    value: 1e-08
+add_cross_attention:
+    value: false
+architectures:
+    value: null
+auto_find_batch_size:
+    value: false
+average_tokens_across_devices:
+    value: false
+backbone_cfg:
+    value:
+        _target_: groot.vla.model.dreamzero.backbone.identity.IdentityBackbone
+bad_words_ids:
+    value: null
+batch_eval_metrics:
+    value: false
+begin_suppress_tokens:
+    value: null
+bf16:
+    value: true
+bf16_full_eval:
+    value: false
+bos_token_id:
+    value: null
+chunk_size_feed_forward:
+    value: 0
+cross_attention_hidden_size:
+    value: null
+data_seed:
+    value: null
+dataloader_drop_last:
+    value: false
+dataloader_num_workers:
+    value: 1
+dataloader_persistent_workers:
+    value: true
+dataloader_pin_memory:
+    value: false
+dataloader_prefetch_factor:
+    value: null
+ddp_backend:
+    value: null
+ddp_broadcast_buffers:
+    value: null
+ddp_bucket_cap_mb:
+    value: 100
+ddp_find_unused_parameters:
+    value: false
+ddp_timeout:
+    value: 1800
+debug:
+    value: []
+decoder_start_token_id:
+    value: null
+deepspeed:
+    value: groot/vla/configs/deepspeed/zero2_offload.json
+disable_tqdm:
+    value: false
+diversity_penalty:
+    value: 0
+do_eval:
+    value: false
+do_predict:
+    value: false
+do_sample:
+    value: false
+do_train:
+    value: false
+early_stopping:
+    value: false
+encoder_no_repeat_ngram_size:
+    value: 0
+eos_token_id:
+    value: null
+eval_accumulation_steps:
+    value: null
+eval_delay:
+    value: 0
+eval_do_concat_batches:
+    value: true
+eval_on_start:
+    value: false
+eval_steps:
+    value: null
+eval_strategy:
+    value: "no"
+eval_use_gather_object:
+    value: false
+exponential_decay_length_penalty:
+    value: null
+finetuning_task:
+    value: null
+forced_bos_token_id:
+    value: null
+forced_eos_token_id:
+    value: null
+fp16:
+    value: false
+fp16_backend:
+    value: auto
+fp16_full_eval:
+    value: false
+fp16_opt_level:
+    value: O1
+fsdp:
+    value: []
+fsdp_config:
+    value:
+        min_num_params: 0
+        xla: false
+        xla_fsdp_grad_ckpt: false
+        xla_fsdp_v2: false
+fsdp_min_num_params:
+    value: 0
+fsdp_transformer_layer_cls_to_wrap:
+    value: null
+full_determinism:
+    value: false
+gradient_accumulation_steps:
+    value: 1
+gradient_checkpointing:
+    value: false
+gradient_checkpointing_kwargs:
+    value: null
+greater_is_better:
+    value: null
+group_by_length:
+    value: false
+half_precision_backend:
+    value: auto
+hidden_size:
+    value: 0
+hub_always_push:
+    value: false
+hub_model_id:
+    value: null
+hub_private_repo:
+    value: null
+hub_strategy:
+    value: every_save
+hub_token:
+    value: <HUB_TOKEN>
+id2label:
+    value:
+        "0": LABEL_0
+        "1": LABEL_1
+ignore_data_skip:
+    value: true
+include_for_metrics:
+    value: []
+include_inputs_for_metrics:
+    value: false
+include_num_input_tokens_seen:
+    value: false
+include_tokens_per_second:
+    value: false
+is_decoder:
+    value: false
+is_encoder_decoder:
+    value: false
+jit_mode_eval:
+    value: false
+label_names:
+    value: null
+label_smoothing_factor:
+    value: 0
+label2id:
+    value:
+        LABEL_0: 0
+        LABEL_1: 1
+learning_rate:
+    value: 1e-05
+length_column_name:
+    value: length
+length_penalty:
+    value: 1
+load_best_model_at_end:
+    value: false
+local_rank:
+    value: 0
+log_level:
+    value: passive
+log_level_replica:
+    value: warning
+log_on_each_node:
+    value: true
+logging_dir:
+    value: ./checkpoints/dreamzero_real_teleop_g1_full_finetune/runs/Mar19_11-00-49_nebula100
+logging_first_step:
+    value: false
+logging_nan_inf_filter:
+    value: true
+logging_steps:
+    value: 10
+logging_strategy:
+    value: steps
+lr_scheduler_type:
+    value: cosine
+max_grad_norm:
+    value: 1
+max_length:
+    value: 20
+max_steps:
+    value: 20000
+metric_for_best_model:
+    value: null
+min_length:
+    value: 0
+model/num_parameters:
+    value: 22924196696
+model_dtype:
+    value: float32
+model_type:
+    value: vla
+mp_parameters:
+    value: ""
+neftune_noise_alpha:
+    value: null
+no_cuda:
+    value: false
+no_repeat_ngram_size:
+    value: 0
+num_beam_groups:
+    value: 1
+num_beams:
+    value: 1
+num_return_sequences:
+    value: 1
+num_train_epochs:
+    value: 1000
+optim:
+    value: adamw_torch
+optim_args:
+    value: null
+optim_target_modules:
+    value: null
+output_attentions:
+    value: false
+output_dir:
+    value: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+output_hidden_states:
+    value: false
+output_scores:
+    value: false
+overwrite_output_dir:
+    value: false
+pad_token_id:
+    value: null
+past_index:
+    value: -1
+per_device_eval_batch_size:
+    value: 64
+per_device_train_batch_size:
+    value: 1
+per_gpu_eval_batch_size:
+    value: null
+per_gpu_train_batch_size:
+    value: null
+prediction_loss_only:
+    value: false
+prefix:
+    value: null
+problem_type:
+    value: null
+push_to_hub:
+    value: false
+push_to_hub_model_id:
+    value: null
+push_to_hub_organization:
+    value: null
+push_to_hub_token:
+    value: <PUSH_TO_HUB_TOKEN>
+ray_scope:
+    value: last
+remove_invalid_values:
+    value: false
+remove_unused_columns:
+    value: false
+repetition_penalty:
+    value: 1
+report_to:
+    value:
+        - wandb
+restore_callback_states_from_checkpoint:
+    value: false
+resume_from_checkpoint:
+    value: null
+resume_path:
+    value: ./checkpoints/dreamzero_real_teleop_g1_full_finetune
+return_dict:
+    value: true
+return_dict_in_generate:
+    value: false
+run_name:
+    value: dreamzero_real_teleop_g1_full_finetune
+save_on_each_node:
+    value: false
+save_only_model:
+    value: false
+save_safetensors:
+    value: true
+save_steps:
+    value: 8000
+save_strategy:
+    value: steps
+save_total_limit:
+    value: 10
+seed:
+    value: 42
+sep_token_id:
+    value: null
+skip_memory_metrics:
+    value: true
+suppress_tokens:
+    value: null
+task_specific_params:
+    value: null
+temperature:
+    value: 1
+tf_legacy_loss:
+    value: false
+tf32:
+    value: true
+tie_encoder_decoder:
+    value: false
+tie_word_embeddings:
+    value: true
+tokenizer_class:
+    value: null
+top_k:
+    value: 50
+top_p:
+    value: 1
+torch_compile:
+    value: false
+torch_compile_backend:
+    value: null
+torch_compile_mode:
+    value: null
+torch_dtype:
+    value: null
+torch_empty_cache_steps:
+    value: null
+torchdynamo:
+    value: null
+torchscript:
+    value: false
+tp_size:
+    value: 0
+tpu_metrics_debug:
+    value: false
+tpu_num_cores:
+    value: null
+transformers_version:
+    value: 4.51.3
+typical_p:
+    value: 1
+use_bfloat16:
+    value: false
+use_cpu:
+    value: false
+use_ipex:
+    value: false
+use_legacy_prediction_loop:
+    value: false
+use_liger_kernel:
+    value: false
+use_mps_device:
+    value: false
+warmup_ratio:
+    value: 0.05
+warmup_steps:
+    value: 0
+weight_decay:
+    value: 1e-05