Add files using upload-large-folder tool

config.json

@@ -0,0 +1,59 @@
+{
+  "action_dim": 7,
+  "action_model_type": "DiT-L",
+  "consolidate_type": "tome",
+  "data_root_dir": "./data/openx_256",
+  "dataloader_type": "stream",
+  "fusion_type": "gate",
+  "future_action_window_size": 15,
+  "group_size": 16,
+  "hf_token": "hf_xxx",
+  "image_aug": true,
+  "is_resume": false,
+  "load_all_data_for_training": true,
+  "mem_length": 16,
+  "per_token_size": 256,
+  "pretrained_checkpoint": "./pretrained/CogACT-Large/checkpoints/CogACT-Large.pt",
+  "repeated_diffusion_steps": 4,
+  "resume_epoch": 0,
+  "resume_step": 0,
+  "retrieval_layers": 2,
+  "run_id": "memvla-bridge",
+  "run_id_note": null,
+  "run_root_dir": "./log/bridge",
+  "save_interval": 2500,
+  "seed": 42,
+  "trackers": [
+    "jsonl",
+    "wandb"
+  ],
+  "update_fused": false,
+  "use_ema": false,
+  "use_timestep_pe": true,
+  "vla": {
+    "base_vlm": "prism-dinosiglip-224px+7b",
+    "data_mix": "bridge",
+    "enable_gradient_checkpointing": true,
+    "enable_mixed_precision_training": true,
+    "epochs": 100,
+    "expected_world_size": 8,
+    "freeze_llm_backbone": false,
+    "freeze_vision_backbone": false,
+    "global_batch_size": 256,
+    "learning_rate": 2e-05,
+    "lr_scheduler_type": "constant",
+    "max_grad_norm": 1.0,
+    "max_steps": 50000,
+    "per_device_batch_size": 32,
+    "reduce_in_full_precision": true,
+    "shuffle_buffer_size": 128000,
+    "train_strategy": "fsdp-full-shard",
+    "type": "prism-dinosiglip-224px+oxe+diffusion",
+    "unfreeze_last_llm_layer": false,
+    "vla_id": "prism-dinosiglip-224px+oxe+diffusion",
+    "warmup_ratio": 0.0,
+    "weight_decay": 0.0
+  },
+  "wandb_entity": "shihao-thu",
+  "wandb_project": "memvla"
+}

config.yaml

@@ -0,0 +1,55 @@
+action_dim: 7
+action_model_type: DiT-L
+consolidate_type: tome
+data_root_dir: ./data/openx_256
+dataloader_type: stream
+fusion_type: gate
+future_action_window_size: 15
+group_size: 16
+hf_token: hf_xxx
+image_aug: true
+is_resume: false
+load_all_data_for_training: true
+mem_length: 16
+per_token_size: 256
+pretrained_checkpoint: ./pretrained/CogACT-Large/checkpoints/CogACT-Large.pt
+repeated_diffusion_steps: 4
+resume_epoch: 0
+resume_step: 0
+retrieval_layers: 2
+run_id: memvla-bridge
+run_id_note: null
+run_root_dir: ./log/bridge
+save_interval: 2500
+seed: 42
+trackers:
+- jsonl
+- wandb
+update_fused: false
+use_ema: false
+use_timestep_pe: true
+vla:
+  base_vlm: prism-dinosiglip-224px+7b
+  data_mix: bridge
+  enable_gradient_checkpointing: true
+  enable_mixed_precision_training: true
+  epochs: 100
+  expected_world_size: 8
+  freeze_llm_backbone: false
+  freeze_vision_backbone: false
+  global_batch_size: 256
+  learning_rate: 2.0e-05
+  lr_scheduler_type: constant
+  max_grad_norm: 1.0
+  max_steps: 50000
+  per_device_batch_size: 32
+  reduce_in_full_precision: true
+  shuffle_buffer_size: 128000
+  train_strategy: fsdp-full-shard
+  type: prism-dinosiglip-224px+oxe+diffusion
+  unfreeze_last_llm_layer: false
+  vla_id: prism-dinosiglip-224px+oxe+diffusion
+  warmup_ratio: 0.0
+  weight_decay: 0.0
+wandb_entity: shihao-thu
+wandb_project: memvla

dataset_statistics.json

@@ -0,0 +1,127 @@
+{
+  "bridge_orig": {
+    "action": {
+      "mean": [
+        0.00023341714404523373,
+        0.00013004250649828464,
+        -0.00012762725236825645,
+        -0.00015565899957437068,
+        -0.0004039369523525238,
+        0.0002355837932555005,
+        0.576458215713501
+      ],
+      "std": [
+        0.009765729308128357,
+        0.01368947234004736,
+        0.012667056173086166,
+        0.02853463403880596,
+        0.03063785843551159,
+        0.07691734284162521,
+        0.4973647892475128
+      ],
+      "max": [
+        0.41691166162490845,
+        0.25864794850349426,
+        0.21218234300613403,
+        3.122201919555664,
+        1.8618112802505493,
+        6.280478477478027,
+        1.0
+      ],
+      "min": [
+        -0.4007510244846344,
+        -0.13874775171279907,
+        -0.22553899884223938,
+        -3.2010786533355713,
+        -1.8618112802505493,
+        -6.279075622558594,
+        0.0
+      ],
+      "q01": [
+        -0.02872725307941437,
+        -0.04170349963009357,
+        -0.026093858778476715,
+        -0.08092105075716972,
+        -0.09288699507713317,
+        -0.20718276381492615,
+        0.0
+      ],
+      "q99": [
+        0.028309678435325586,
+        0.040855254605412394,
+        0.040161586627364146,
+        0.08192047759890528,
+        0.07792850524187081,
+        0.20382574498653397,
+        1.0
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        false
+      ]
+    },
+    "proprio": {
+      "mean": [
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0
+      ],
+      "std": [
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0
+      ],
+      "max": [
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0
+      ],
+      "min": [
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0
+      ],
+      "q01": [
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0
+      ],
+      "q99": [
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0,
+        0.0
+      ]
+    },
+    "num_transitions": 2135463,
+    "num_trajectories": 60064
+  }
+}

eval_simpler/memvla-bridge.pt/Carrot.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfe8266385bca362f22dec1d5803c0f13e99cc0141b0cd9a8eb6874b60c7ff55
+size 7253

eval_simpler/memvla-bridge.pt/Cube.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cb174658fa5f532db10a79e3cf42059e2e1441c11819873537bc6f424b01b82
+size 7799

eval_simpler/memvla-bridge.pt/Eggplant.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:628e0e563ec73fcc1e3c42ceadecc63e73b809d7ce63d4f3b42a211fccd99a40
+size 7535

eval_simpler/memvla-bridge.pt/Spoon.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:477fa605e6a8e1ce205f3949a541916a6f1ee69ba6d876815353c393c53272e7
+size 7425

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/failure_obj_episode_13_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_False.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:665391a037360b2bed8fb75c945f78317b8433f03e072c004eb98c85f489bb5b
+size 334949

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/failure_obj_episode_18_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_False_src_on_target_False.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c5305997291567b76685279e732f76d97a2399854330d8a5867a01fcb60ce44a
+size 523610

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/failure_obj_episode_1_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_False.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f113404b01b7e23f7f7ecf8d3b9263ab8e79d1f061bce7818442a79f0bba663a
+size 329671

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/failure_obj_episode_23_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_False.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7509914fe6e576351e86d27e3f055123e5c718f0eb17144ad751a19b152d6261
+size 273599

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_0_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9dd9309db5e56c3a4d7c7d7bb85768472c74b24ba1a99313c95fcff7d21c32ed
+size 217887

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_10_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f416cf699788479f22753390ea0238a013b70acb93b0e5e1b218de39c2153d1d
+size 210737

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_11_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ea1287caeff41cfc145a449961d11362b7959d80f32a68faa4acb72d1223772
+size 260320

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_12_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:997345bd6f2207f577aa4e991a989411be4604cec4a095770a9ff172ed0f989f
+size 191534

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_14_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41645943d18dd548d88528dde70fe9a2221f1998e09843cf12c7aeebe8de7f63
+size 235750

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_15_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:499312f4a59af685571c28a704bedecec70bf93571057e66ac3387dc64c177d1
+size 242354

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_16_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52380b32c1769f5132844abdb54a099bf55c621c3debb89c5d04bb8583915f47
+size 229821

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_17_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1edd1543c7ba5605a81ded47e7d833108d687e4d3b48594440c0208463a111a
+size 312467

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_19_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98ef9ba4a802690c9f996d0ad87c0ea6efc0d194b00d9344077e394e5ac8b82a
+size 272916

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_20_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:828339b28a79a6c75d646b646a1e030bb3259b90ae7b17e4ffab7eba42184774
+size 237292

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_21_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1e06094bf045f13a8f1d9192e62d79ada656c457ac341c4d149386b61ef16ab
+size 384580

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_22_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:257a4d444163897f05b7b8bf2063028b3c846bf7cdf1c1cadff22ec3f8317ada
+size 236742

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_2_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c7f33b21e3d4bf2e9cdbe9c52ef472ff5f5ea9c4bb5ee40d8f2b64cf8eed0fa
+size 169938

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_3_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03a3c36bcd99ca8e5ea598e2e04b979a0e4361f032e130571be0d2a9a5b8a3f8
+size 218065

eval_simpler/memvla-bridge.pt/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0/rob_0.147_0.028_rot_0.000_-0.000_3.142_rgb_overlay_bridge_real_eval_1/success_obj_episode_4_moved_correct_obj_True_moved_wrong_obj_False_is_src_obj_grasped_True_consecutive_grasp_True_src_on_target_True.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e51e6d8198704d7212c08b9a7bbb34a6d6b754aa726b1824867c1c497265b6d0
+size 209978