diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed7.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..eff928d59cc5d589c638a4a1ee7b58917509734b --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed7.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage1_clip_seed7 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 7 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage1_seed7.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage1_seed7.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 7 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.2 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed8.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed8.yaml new file mode 100644 index 0000000000000000000000000000000000000000..eb7af2bd2871f10553592409954b935d914cf98d --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed8.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage1_clip_seed8 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 8 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage1_seed8.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage1_seed8.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 8 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.2 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed9.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed9.yaml new file mode 100644 index 0000000000000000000000000000000000000000..426f4d7345307a7df8fe8fff3536b2bef0b1a763 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage1_clip_seed9.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage1_clip_seed9 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 9 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage1_seed9.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage1_seed9.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 9 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.2 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed13.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed13.yaml new file mode 100644 index 0000000000000000000000000000000000000000..441cbf8ec8fa39123f486d3ba1787de5632aa000 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed13.yaml @@ -0,0 +1,147 @@ +experiment_name: proxy_interaction_r3d_stage1_dummy_seed13 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 13 +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 96 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_v6_rgbd_stage1_dummy_seed13.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_v6_rgbd_stage1_dummy_seed13.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 13 +optim: + epochs: 4 + batch_size: 16 + num_workers: 4 + lr: 0.001 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: false + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 192 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: true + fusion: + hidden_dim: 192 + num_cameras: 3 + num_layers: 2 + num_heads: 4 + ff_dim: 384 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 192 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 4 + max_history_steps: 8 + decoder: + hidden_dim: 192 + num_heads: 4 + num_layers: 2 + ff_dim: 384 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 192 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 4 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 192 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 4 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 192 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 4 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.15 + arm_role: 0.2 + support_mode: 0.15 + corridor: 0.2 + persistence: 0.1 + disturbance: 0.1 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.2 + planner_risk: 0.1 + planner_ranking: 0.1 + proposal_reconstruction: 0.2 + proposal_success: 0.1 + proposal_ranking: 0.1 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed14.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed14.yaml new file mode 100644 index 0000000000000000000000000000000000000000..51fa0f37d7643addb961bc9cea308c2069430f10 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed14.yaml @@ -0,0 +1,147 @@ +experiment_name: proxy_interaction_r3d_stage1_dummy_seed14 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 14 +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 96 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_v6_rgbd_stage1_dummy_seed14.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_v6_rgbd_stage1_dummy_seed14.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 14 +optim: + epochs: 4 + batch_size: 16 + num_workers: 4 + lr: 0.001 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: false + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 192 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: true + fusion: + hidden_dim: 192 + num_cameras: 3 + num_layers: 2 + num_heads: 4 + ff_dim: 384 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 192 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 4 + max_history_steps: 8 + decoder: + hidden_dim: 192 + num_heads: 4 + num_layers: 2 + ff_dim: 384 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 192 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 4 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 192 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 4 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 192 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 4 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.15 + arm_role: 0.2 + support_mode: 0.15 + corridor: 0.2 + persistence: 0.1 + disturbance: 0.1 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.2 + planner_risk: 0.1 + planner_ranking: 0.1 + proposal_reconstruction: 0.2 + proposal_success: 0.1 + proposal_ranking: 0.1 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed15.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed15.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bf93049674d59f6ad7937203233ea51c2cdbbaed --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage1_dummy_seed15.yaml @@ -0,0 +1,147 @@ +experiment_name: proxy_interaction_r3d_stage1_dummy_seed15 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 15 +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 96 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_v6_rgbd_stage1_dummy_seed15.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_v6_rgbd_stage1_dummy_seed15.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 15 +optim: + epochs: 4 + batch_size: 16 + num_workers: 4 + lr: 0.001 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: false + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 192 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: true + fusion: + hidden_dim: 192 + num_cameras: 3 + num_layers: 2 + num_heads: 4 + ff_dim: 384 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 192 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 4 + max_history_steps: 8 + decoder: + hidden_dim: 192 + num_heads: 4 + num_layers: 2 + ff_dim: 384 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 192 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 4 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 192 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 4 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 192 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 4 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.15 + arm_role: 0.2 + support_mode: 0.15 + corridor: 0.2 + persistence: 0.1 + disturbance: 0.1 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.2 + planner_risk: 0.1 + planner_ranking: 0.1 + proposal_reconstruction: 0.2 + proposal_success: 0.1 + proposal_ranking: 0.1 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed11.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed11.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f8259a4b910eccd10954ce134823179a566fdb9f --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed11.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage2_clip_seed11 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 11 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage2_seed11.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage2_seed11.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 11 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed12.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed12.yaml new file mode 100644 index 0000000000000000000000000000000000000000..51410160afba1bc080f013abc06d9c7e4edfc9f1 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed12.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage2_clip_seed12 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 12 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage2_seed12.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage2_seed12.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 12 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed13.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed13.yaml new file mode 100644 index 0000000000000000000000000000000000000000..cfd523ac65ba14b15a6abef147c17da31bc6968c --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage2_clip_seed13.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage2_clip_seed13 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 13 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage2_seed13.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage2_seed13.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 13 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed21.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed21.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a696d5c3f8c9d9af76f2eebe257f06c3312751b3 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed21.yaml @@ -0,0 +1,148 @@ +experiment_name: proxy_interaction_r3d_stage2_dummy_seed21 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 21 +defaults: [] +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 96 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_v6_rgbd_stage2_dummy_seed21.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_v6_rgbd_stage2_dummy_seed21.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 21 +optim: + epochs: 4 + batch_size: 16 + num_workers: 4 + lr: 0.001 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: false + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 192 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: true + fusion: + hidden_dim: 192 + num_cameras: 3 + num_layers: 2 + num_heads: 4 + ff_dim: 384 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 192 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 4 + max_history_steps: 8 + decoder: + hidden_dim: 192 + num_heads: 4 + num_layers: 2 + ff_dim: 384 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 192 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 4 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 192 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 4 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 192 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 4 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.15 + arm_role: 0.2 + support_mode: 0.15 + corridor: 0.2 + persistence: 0.1 + disturbance: 0.1 + world_model: 0.3 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.2 + planner_risk: 0.1 + planner_ranking: 0.1 + proposal_reconstruction: 0.2 + proposal_success: 0.1 + proposal_ranking: 0.1 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed22.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed22.yaml new file mode 100644 index 0000000000000000000000000000000000000000..120a05c579a02cb5837dea6e0e837736552231bf --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed22.yaml @@ -0,0 +1,148 @@ +experiment_name: proxy_interaction_r3d_stage2_dummy_seed22 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 22 +defaults: [] +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 96 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_v6_rgbd_stage2_dummy_seed22.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_v6_rgbd_stage2_dummy_seed22.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 22 +optim: + epochs: 4 + batch_size: 16 + num_workers: 4 + lr: 0.001 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: false + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 192 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: true + fusion: + hidden_dim: 192 + num_cameras: 3 + num_layers: 2 + num_heads: 4 + ff_dim: 384 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 192 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 4 + max_history_steps: 8 + decoder: + hidden_dim: 192 + num_heads: 4 + num_layers: 2 + ff_dim: 384 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 192 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 4 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 192 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 4 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 192 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 4 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.15 + arm_role: 0.2 + support_mode: 0.15 + corridor: 0.2 + persistence: 0.1 + disturbance: 0.1 + world_model: 0.3 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.2 + planner_risk: 0.1 + planner_ranking: 0.1 + proposal_reconstruction: 0.2 + proposal_success: 0.1 + proposal_ranking: 0.1 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed23.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed23.yaml new file mode 100644 index 0000000000000000000000000000000000000000..00c657eb23e26c38238200a9ac782819c1acebf3 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage2_dummy_seed23.yaml @@ -0,0 +1,148 @@ +experiment_name: proxy_interaction_r3d_stage2_dummy_seed23 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 23 +defaults: [] +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 96 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_v6_rgbd_stage2_dummy_seed23.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_v6_rgbd_stage2_dummy_seed23.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 23 +optim: + epochs: 4 + batch_size: 16 + num_workers: 4 + lr: 0.001 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: false + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: false + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 192 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: true + fusion: + hidden_dim: 192 + num_cameras: 3 + num_layers: 2 + num_heads: 4 + ff_dim: 384 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 192 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 4 + max_history_steps: 8 + decoder: + hidden_dim: 192 + num_heads: 4 + num_layers: 2 + ff_dim: 384 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 192 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 4 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 192 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 4 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 192 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 4 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.15 + arm_role: 0.2 + support_mode: 0.15 + corridor: 0.2 + persistence: 0.1 + disturbance: 0.1 + world_model: 0.3 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.2 + planner_risk: 0.1 + planner_ranking: 0.1 + proposal_reconstruction: 0.2 + proposal_success: 0.1 + proposal_ranking: 0.1 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed17.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed17.yaml new file mode 100644 index 0000000000000000000000000000000000000000..79cd3d6e38e4b0a081d397b9d5130b5ca69707bc --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed17.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage3_clip_rgbd_seed17 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 17 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage3_seed17.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage3_seed17.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 17 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: true + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed18.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed18.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f95b269273891af66f01204fdc3d8d102945fa27 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed18.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage3_clip_rgbd_seed18 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 18 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage3_seed18.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage3_seed18.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 18 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: true + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed19.yaml b/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed19.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a02b2cf736be139e064cb0834562eee7b9210c49 --- /dev/null +++ b/artifacts/generated_configs/proxy_interaction_r3d_stage3_clip_rgbd_seed19.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage3_clip_rgbd_seed19 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 19 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage3_seed19.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage3_seed19.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 19 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: true + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..a728c113c22d156317a29d76f521b5703da9183b --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_dual_push_buttons --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..28d44fe0ebbaf50c6997a93240788ec4e4cc04a4 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_dual_push_buttons/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualDualPushButtons", + "description": "push the olive and the orange buttons", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..a1097997bedab5fd569f3ea63eb7fe1444a8c37e --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_handover_item --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..13d371011df9418bee1a5f1b94d2654572894907 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualHandoverItem", + "description": "bring me the red item", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..d7355d431868290c1d7fa1648384f23245bad2e4 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_handover_item_easy --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..0f9d3c1e7ca5d87a57edd5d0bd8914558037614b --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_handover_item_easy/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualHandoverItemEasy", + "description": "bring me the item", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..9b469763faaf25b322b37dd0b6b3fa788a4c6fa9 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_lift_ball --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..6339602138f0992a15c39e7904e293bdfe1bdd62 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_ball/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualLiftBall", + "description": "Lift the ball", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..2de31091a7209f2ce30537210dc184e5ca7052e6 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_lift_tray --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..416e0e36b8378f5691255806da0037b4eb822738 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_lift_tray/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualLiftTray", + "description": "Lift the tray", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..ddecb81725bdb853688f500fb9b12b6b4e8bdfd7 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_pick_laptop --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3b0d67a6a8756c1d220e36224062c7951d06d800 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/stderr.txt @@ -0,0 +1,6 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. +WARNING:root:Waypoints are not reachable right=(False, 'waypoint1 - DualPanda') left=(True, '') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_pick_laptop. Infeasible episode. Can't reach waypoint ((False, 'waypoint1 - DualPanda'), (True, '')). +WARNING:root:Waypoints are not reachable right=(False, 'waypoint1 - DualPanda') left=(True, '') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_pick_laptop. Infeasible episode. Can't reach waypoint ((False, 'waypoint1 - DualPanda'), (True, '')). diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..26b8f13f5acad936f90b06d6688c82ec60244baa --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_laptop/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualPickLaptop", + "description": "pick up the laptop", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..e168a9df1d2b6f84d4a585c736e20f1a6d7a2014 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_pick_plate --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..48774aec4da572b37f528e7e75113b41f6261232 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_pick_plate/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualPickPlate", + "description": "pick up the plate", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..328737b7efe0e6babbef31ddd7e79119fd9e9eb3 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_push_box --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..d5e207a5132ae70695dcce13fec5547a4b2c1b99 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_push_box/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualPushBox", + "description": "push the box to the red area", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..3638f874c1bb56234f6728e59da6db7d61c9f5be --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_put_bottle_in_fridge --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..f2a40ca618fe3319e0b1c76f7a29f78315887a51 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_bottle_in_fridge/stdout.txt @@ -0,0 +1,39 @@ +[ 0.05 -0.27500001 0.75199997] +{ + "display": ":99", + "headless": true, + "task": "BimanualPutBottleInFridge", + "description": "put bottle in fridge", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..bae6bc8478301ecfde6369a85ed53eea023b63f3 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_put_item_in_drawer --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..e1207d2389fd6fd5fa15cba12bc1be2d6d55657d --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/stderr.txt @@ -0,0 +1,12 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. +WARNING:root:Waypoints are not reachable right=(False, 'waypoint4 - DualPanda') left=(False, 'waypoint2 - DualPanda') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_put_item_in_drawer. Infeasible episode. Can't reach waypoint ((False, 'waypoint4 - DualPanda'), (False, 'waypoint2 - DualPanda')). +WARNING:root:Waypoints are not reachable right=(False, 'waypoint4 - DualPanda') left=(False, 'waypoint2 - DualPanda') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_put_item_in_drawer. Infeasible episode. Can't reach waypoint ((False, 'waypoint4 - DualPanda'), (False, 'waypoint2 - DualPanda')). +WARNING:root:Waypoints are not reachable right=(False, 'waypoint4 - DualPanda') left=(True, '') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_put_item_in_drawer. Infeasible episode. Can't reach waypoint ((False, 'waypoint4 - DualPanda'), (True, '')). +WARNING:root:Waypoints are not reachable right=(False, 'waypoint4 - DualPanda') left=(False, 'waypoint2 - DualPanda') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_put_item_in_drawer. Infeasible episode. Can't reach waypoint ((False, 'waypoint4 - DualPanda'), (False, 'waypoint2 - DualPanda')). +WARNING:root:Waypoints are not reachable right=(False, 'waypoint4 - DualPanda') left=(True, '') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_put_item_in_drawer. Infeasible episode. Can't reach waypoint ((False, 'waypoint4 - DualPanda'), (True, '')). diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..51c84d886cce725ed1843d955040da6e23454d6b --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_put_item_in_drawer/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualPutItemInDrawer", + "description": "put the item in the bottom drawer", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..20cc9c2dba9687ef498ce2f3cda18664c88b5eb3 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_straighten_rope --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..54aa8d15b06ce33a32e826dd7afda1c72139da21 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/stderr.txt @@ -0,0 +1,4 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. +WARNING:root:Waypoints are not reachable right=(False, 'waypoint3 - DualPanda') left=(True, '') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_straighten_rope. Infeasible episode. Can't reach waypoint ((False, 'waypoint3 - DualPanda'), (True, '')). diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..60edb34aff776cb8d9b60bed8ef6cad346e68307 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_straighten_rope/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualStraightenRope", + "description": "straighten rope", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..04b0c82b7ab0818bba30fb5b08dcf4412b1c0748 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_sweep_to_dustpan --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..3a8c9d15613dda07b91a2d3d81eaeb75e3e7b1fc --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/stderr.txt @@ -0,0 +1,2 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..82c6677e7f5d6907c112dd5229efa57057100267 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_sweep_to_dustpan/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualSweepToDustpan", + "description": "sweep dirt to dustpan", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/command.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..837597dcc586a5b31d64a99d0bf5bfb51649b976 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m sim_rlbench.launch_smoke --task bimanual_take_tray_out_of_oven --resolution 224 --headless diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/stderr.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..7b9b4b306546c313ef6348bb7265541d797ea034 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/stderr.txt @@ -0,0 +1,4 @@ +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. +WARNING:root:Waypoints are not reachable right=(False, 'waypoint0 - DualPanda') left=(True, '') +ERROR:root:Error when checking waypoints. Exception is: Error in task bimanual_take_tray_out_of_oven. Infeasible episode. Can't reach waypoint ((False, 'waypoint0 - DualPanda'), (True, '')). diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/stdout.txt b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..9775a456db4e41e439005d6d7fb844390c5a023d --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/bimanual_take_tray_out_of_oven/stdout.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "BimanualTakeTrayOutOfOven", + "description": "take tray out of oven", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/launch_smoke_summary.json b/artifacts/outputs/r3d/peract2_13_launch_smoke/launch_smoke_summary.json new file mode 100644 index 0000000000000000000000000000000000000000..20d43a766561c89fbafe8db5d227652791ccc006 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/launch_smoke_summary.json @@ -0,0 +1,517 @@ +{ + "resolution": 224, + "headless": true, + "tasks": { + "bimanual_push_box": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualPushBox", + "description": "push the box to the red area", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_lift_ball": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualLiftBall", + "description": "Lift the ball", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_dual_push_buttons": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualDualPushButtons", + "description": "push the olive and the orange buttons", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_pick_plate": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualPickPlate", + "description": "pick up the plate", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_put_item_in_drawer": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualPutItemInDrawer", + "description": "put the item in the bottom drawer", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_put_bottle_in_fridge": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualPutBottleInFridge", + "description": "put bottle in fridge", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_handover_item": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualHandoverItem", + "description": "bring me the red item", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_pick_laptop": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualPickLaptop", + "description": "pick up the laptop", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_straighten_rope": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualStraightenRope", + "description": "straighten rope", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_sweep_to_dustpan": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualSweepToDustpan", + "description": "sweep dirt to dustpan", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_lift_tray": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualLiftTray", + "description": "Lift the tray", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_handover_item_easy": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualHandoverItemEasy", + "description": "bring me the item", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + }, + "bimanual_take_tray_out_of_oven": { + "subprocess_returncode": 0, + "launch_ok": true, + "display": ":99", + "headless": true, + "task": "BimanualTakeTrayOutOfOven", + "description": "take tray out of oven", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] + } + }, + "task_count": 13, + "launch_successes": 13, + "finite_action_tasks": 13, + "error_tasks": [] +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/peract2_13_launch_smoke/launch_smoke_summary.md b/artifacts/outputs/r3d/peract2_13_launch_smoke/launch_smoke_summary.md new file mode 100644 index 0000000000000000000000000000000000000000..ea6de2b888a8d99a983fa2916e8a5018ed0025c2 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_launch_smoke/launch_smoke_summary.md @@ -0,0 +1,24 @@ +# PerAct2 13-Task Launch Smoke + +- Resolution: `224` +- Headless: `True` +- Task count: `13` +- Launch successes: `13` +- Finite-action tasks: `13` +- Error tasks: `[]` + +## Per-task + +- `bimanual_push_box`: launch_ok=True, action_finite=True, task_class=BimanualPushBox, reward=0.0 +- `bimanual_lift_ball`: launch_ok=True, action_finite=True, task_class=BimanualLiftBall, reward=0.0 +- `bimanual_dual_push_buttons`: launch_ok=True, action_finite=True, task_class=BimanualDualPushButtons, reward=0.0 +- `bimanual_pick_plate`: launch_ok=True, action_finite=True, task_class=BimanualPickPlate, reward=0.0 +- `bimanual_put_item_in_drawer`: launch_ok=True, action_finite=True, task_class=BimanualPutItemInDrawer, reward=0.0 +- `bimanual_put_bottle_in_fridge`: launch_ok=True, action_finite=True, task_class=BimanualPutBottleInFridge, reward=0.0 +- `bimanual_handover_item`: launch_ok=True, action_finite=True, task_class=BimanualHandoverItem, reward=0.0 +- `bimanual_pick_laptop`: launch_ok=True, action_finite=True, task_class=BimanualPickLaptop, reward=0.0 +- `bimanual_straighten_rope`: launch_ok=True, action_finite=True, task_class=BimanualStraightenRope, reward=0.0 +- `bimanual_sweep_to_dustpan`: launch_ok=True, action_finite=True, task_class=BimanualSweepToDustpan, reward=0.0 +- `bimanual_lift_tray`: launch_ok=True, action_finite=True, task_class=BimanualLiftTray, reward=0.0 +- `bimanual_handover_item_easy`: launch_ok=True, action_finite=True, task_class=BimanualHandoverItemEasy, reward=0.0 +- `bimanual_take_tray_out_of_oven`: launch_ok=True, action_finite=True, task_class=BimanualTakeTrayOutOfOven, reward=0.0 diff --git a/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/command.txt b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..2f4c27cd3991b5bd5c0119f9834920bfd4c7676e --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m eval.run_rlbench_rollout_eval --checkpoint /workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt --output-dir /workspace/VLAarchtests/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box --tasks bimanual_push_box --episodes-per-task 1 --episode-length 2 --resolution 224 --device cuda --chunk-commit-steps 1 --headless --allow-unsupervised-planning diff --git a/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.json b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.json new file mode 100644 index 0000000000000000000000000000000000000000..8264fc776737445dd30ea1470014c0075edfe4e6 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.json @@ -0,0 +1,35 @@ +{ + "checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt", + "plan_requested": false, + "plan_applied": false, + "planner_mode": "trainable", + "support_mode_conditioning": true, + "episodes_per_task": 1, + "episode_length": 2, + "resolution": 224, + "reset_retries": 20, + "cameras": [ + "front", + "wrist_left", + "wrist_right" + ], + "tasks": { + "bimanual_push_box": { + "task_class": "BimanualPushBox", + "successes": [ + 0.0 + ], + "returns": [ + 0.0 + ], + "path_recoveries": 0, + "noop_fallbacks": 0, + "reset_retries": [ + 0 + ], + "mean_success": 0.0, + "mean_return": 0.0 + } + }, + "mean_success": 0.0 +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.md b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.md new file mode 100644 index 0000000000000000000000000000000000000000..1cd91cdc8960748f070009c5a1cd47a5a2735dd9 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.md @@ -0,0 +1,11 @@ +# RLBench Rollout Eval + +- Checkpoint: `/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt` +- Plan requested: `False` +- Plan applied: `False` +- Support-mode conditioning: `True` +- Mean success: `0.000` + +## Per-task + +- `bimanual_push_box`: mean_success=0.000, returns=[0.0] diff --git a/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/stderr.txt b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..5e59d7b97d75398177c479be5f31b06dbac14246 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/stderr.txt @@ -0,0 +1,4 @@ +/workspace/envs/rlbench/lib/python3.10/site-packages/torch/nn/modules/transformer.py:306: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.norm_first was True + warnings.warn(f"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}") +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/stdout.txt b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..74d7009fa89d34be2a0d414e14f42e90fdbd7f14 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_smoke_short/peract2_13_rollout_noplan_split/bimanual_push_box/stdout.txt @@ -0,0 +1,36 @@ +{ + "checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt", + "plan_requested": false, + "plan_applied": false, + "planner_mode": "trainable", + "support_mode_conditioning": true, + "episodes_per_task": 1, + "episode_length": 2, + "resolution": 224, + "reset_retries": 20, + "cameras": [ + "front", + "wrist_left", + "wrist_right" + ], + "tasks": { + "bimanual_push_box": { + "task_class": "BimanualPushBox", + "successes": [ + 0.0 + ], + "returns": [ + 0.0 + ], + "path_recoveries": 0, + "noop_fallbacks": 0, + "reset_retries": [ + 0 + ], + "mean_success": 0.0, + "mean_return": 0.0 + } + }, + "mean_success": 0.0 +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/command.txt b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..a51848c962daebe148b36fda2df58dfa25a85a82 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m eval.run_rlbench_rollout_eval --checkpoint /workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt --output-dir /workspace/VLAarchtests/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball --tasks bimanual_lift_ball --episodes-per-task 1 --episode-length 20 --resolution 224 --device cpu --chunk-commit-steps 4 --headless --allow-unsupervised-planning diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/rollout_eval.json b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/rollout_eval.json new file mode 100644 index 0000000000000000000000000000000000000000..527cc51b07248dfde4c862aaf600e5437f4b2254 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/rollout_eval.json @@ -0,0 +1,35 @@ +{ + "checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt", + "plan_requested": false, + "plan_applied": false, + "planner_mode": "trainable", + "support_mode_conditioning": true, + "episodes_per_task": 1, + "episode_length": 20, + "resolution": 224, + "reset_retries": 20, + "cameras": [ + "front", + "wrist_left", + "wrist_right" + ], + "tasks": { + "bimanual_lift_ball": { + "task_class": "BimanualLiftBall", + "successes": [ + 0.0 + ], + "returns": [ + 0.0 + ], + "path_recoveries": 0, + "noop_fallbacks": 0, + "reset_retries": [ + 0 + ], + "mean_success": 0.0, + "mean_return": 0.0 + } + }, + "mean_success": 0.0 +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/rollout_eval.md b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/rollout_eval.md new file mode 100644 index 0000000000000000000000000000000000000000..6f0c11cab5e220c58ee8cf7a3143e76c82acb692 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/rollout_eval.md @@ -0,0 +1,11 @@ +# RLBench Rollout Eval + +- Checkpoint: `/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt` +- Plan requested: `False` +- Plan applied: `False` +- Support-mode conditioning: `True` +- Mean success: `0.000` + +## Per-task + +- `bimanual_lift_ball`: mean_success=0.000, returns=[0.0] diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/stderr.txt b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..5e59d7b97d75398177c479be5f31b06dbac14246 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/stderr.txt @@ -0,0 +1,4 @@ +/workspace/envs/rlbench/lib/python3.10/site-packages/torch/nn/modules/transformer.py:306: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.norm_first was True + warnings.warn(f"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}") +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/stdout.txt b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..a8c64313dbc48c13b16d3dcdac48a416e896aa46 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_lift_ball/stdout.txt @@ -0,0 +1,36 @@ +{ + "checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt", + "plan_requested": false, + "plan_applied": false, + "planner_mode": "trainable", + "support_mode_conditioning": true, + "episodes_per_task": 1, + "episode_length": 20, + "resolution": 224, + "reset_retries": 20, + "cameras": [ + "front", + "wrist_left", + "wrist_right" + ], + "tasks": { + "bimanual_lift_ball": { + "task_class": "BimanualLiftBall", + "successes": [ + 0.0 + ], + "returns": [ + 0.0 + ], + "path_recoveries": 0, + "noop_fallbacks": 0, + "reset_retries": [ + 0 + ], + "mean_success": 0.0, + "mean_return": 0.0 + } + }, + "mean_success": 0.0 +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/command.txt b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/command.txt new file mode 100644 index 0000000000000000000000000000000000000000..4621dfebd4c2fb08b12e49bf54aef40fea267df9 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/command.txt @@ -0,0 +1 @@ +/workspace/envs/rlbench/bin/python -m eval.run_rlbench_rollout_eval --checkpoint /workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt --output-dir /workspace/VLAarchtests/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box --tasks bimanual_push_box --episodes-per-task 1 --episode-length 20 --resolution 224 --device cpu --chunk-commit-steps 4 --headless --allow-unsupervised-planning diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.json b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.json new file mode 100644 index 0000000000000000000000000000000000000000..6c79ac43dbdeff4b602c44d6690ec1a6055f6d54 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.json @@ -0,0 +1,35 @@ +{ + "checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt", + "plan_requested": false, + "plan_applied": false, + "planner_mode": "trainable", + "support_mode_conditioning": true, + "episodes_per_task": 1, + "episode_length": 20, + "resolution": 224, + "reset_retries": 20, + "cameras": [ + "front", + "wrist_left", + "wrist_right" + ], + "tasks": { + "bimanual_push_box": { + "task_class": "BimanualPushBox", + "successes": [ + 0.0 + ], + "returns": [ + 0.0 + ], + "path_recoveries": 0, + "noop_fallbacks": 0, + "reset_retries": [ + 0 + ], + "mean_success": 0.0, + "mean_return": 0.0 + } + }, + "mean_success": 0.0 +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.md b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.md new file mode 100644 index 0000000000000000000000000000000000000000..1cd91cdc8960748f070009c5a1cd47a5a2735dd9 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/rollout_eval.md @@ -0,0 +1,11 @@ +# RLBench Rollout Eval + +- Checkpoint: `/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt` +- Plan requested: `False` +- Plan applied: `False` +- Support-mode conditioning: `True` +- Mean success: `0.000` + +## Per-task + +- `bimanual_push_box`: mean_success=0.000, returns=[0.0] diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/stderr.txt b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/stderr.txt new file mode 100644 index 0000000000000000000000000000000000000000..5e59d7b97d75398177c479be5f31b06dbac14246 --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/stderr.txt @@ -0,0 +1,4 @@ +/workspace/envs/rlbench/lib/python3.10/site-packages/torch/nn/modules/transformer.py:306: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.norm_first was True + warnings.warn(f"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}") +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/stdout.txt b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/stdout.txt new file mode 100644 index 0000000000000000000000000000000000000000..38fe983312533730155921afec241fd5b1fe726a --- /dev/null +++ b/artifacts/outputs/r3d/peract2_13_r3d_sweep/peract2_13_rollout_noplan_split/bimanual_push_box/stdout.txt @@ -0,0 +1,36 @@ +{ + "checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt", + "plan_requested": false, + "plan_applied": false, + "planner_mode": "trainable", + "support_mode_conditioning": true, + "episodes_per_task": 1, + "episode_length": 20, + "resolution": 224, + "reset_retries": 20, + "cameras": [ + "front", + "wrist_left", + "wrist_right" + ], + "tasks": { + "bimanual_push_box": { + "task_class": "BimanualPushBox", + "successes": [ + 0.0 + ], + "returns": [ + 0.0 + ], + "path_recoveries": 0, + "noop_fallbacks": 0, + "reset_retries": [ + 0 + ], + "mean_success": 0.0, + "mean_return": 0.0 + } + }, + "mean_success": 0.0 +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed18/summary.json b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed18/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..38e9b8ba0a338c57efbf5b2d362182a910047301 --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed18/summary.json @@ -0,0 +1,557 @@ +{ + "experiment_name": "proxy_interaction_r3d_stage3_clip_rgbd_seed18", + "device": "cuda", + "best_checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed18/checkpoint_best.pt", + "final_train_total": 1.4377938396643595, + "final_val_total": 1.4556549372095051, + "train_time_sec": 150.84560704231262, + "peak_gpu_memory_mb": 1962.224609375, + "num_train_samples": 381, + "num_val_samples": 132, + "planner_mode": "trainable", + "frozen_modules": [], + "init_info": { + "path": "/workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt", + "loaded_keys": 461, + "skipped_shape_mismatch_keys": [ + "memory.gru.weight_ih_l0", + "memory.gru.weight_hh_l0", + "memory.gru.bias_ih_l0", + "memory.gru.bias_hh_l0", + "memory.token_proj.0.weight", + "memory.token_proj.0.bias", + "memory.token_proj.1.weight", + "memory.token_proj.1.bias", + "decoder.actor_role_bias", + "decoder.revealer_decoder.layers.0.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.0.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.0.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.0.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.0.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.0.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.0.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.0.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.0.linear1.weight", + "decoder.revealer_decoder.layers.0.linear1.bias", + "decoder.revealer_decoder.layers.0.linear2.weight", + "decoder.revealer_decoder.layers.0.linear2.bias", + "decoder.revealer_decoder.layers.0.norm1.weight", + "decoder.revealer_decoder.layers.0.norm1.bias", + "decoder.revealer_decoder.layers.0.norm2.weight", + "decoder.revealer_decoder.layers.0.norm2.bias", + "decoder.revealer_decoder.layers.0.norm3.weight", + "decoder.revealer_decoder.layers.0.norm3.bias", + "decoder.revealer_decoder.layers.1.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.1.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.1.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.1.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.1.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.1.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.1.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.1.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.1.linear1.weight", + "decoder.revealer_decoder.layers.1.linear1.bias", + "decoder.revealer_decoder.layers.1.linear2.weight", + "decoder.revealer_decoder.layers.1.linear2.bias", + "decoder.revealer_decoder.layers.1.norm1.weight", + "decoder.revealer_decoder.layers.1.norm1.bias", + "decoder.revealer_decoder.layers.1.norm2.weight", + "decoder.revealer_decoder.layers.1.norm2.bias", + "decoder.revealer_decoder.layers.1.norm3.weight", + "decoder.revealer_decoder.layers.1.norm3.bias", + "decoder.revealer_decoder.layers.2.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.2.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.2.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.2.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.2.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.2.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.2.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.2.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.2.linear1.weight", + "decoder.revealer_decoder.layers.2.linear1.bias", + "decoder.revealer_decoder.layers.2.linear2.weight", + "decoder.revealer_decoder.layers.2.linear2.bias", + "decoder.revealer_decoder.layers.2.norm1.weight", + "decoder.revealer_decoder.layers.2.norm1.bias", + "decoder.revealer_decoder.layers.2.norm2.weight", + "decoder.revealer_decoder.layers.2.norm2.bias", + "decoder.revealer_decoder.layers.2.norm3.weight", + "decoder.revealer_decoder.layers.2.norm3.bias", + "decoder.revealer_decoder.layers.3.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.3.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.3.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.3.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.3.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.3.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.3.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.3.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.3.linear1.weight", + "decoder.revealer_decoder.layers.3.linear1.bias", + "decoder.revealer_decoder.layers.3.linear2.weight", + "decoder.revealer_decoder.layers.3.linear2.bias", + "decoder.revealer_decoder.layers.3.norm1.weight", + "decoder.revealer_decoder.layers.3.norm1.bias", + "decoder.revealer_decoder.layers.3.norm2.weight", + "decoder.revealer_decoder.layers.3.norm2.bias", + "decoder.revealer_decoder.layers.3.norm3.weight", + "decoder.revealer_decoder.layers.3.norm3.bias", + "decoder.actor_decoder.layers.0.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.0.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.0.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.0.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.0.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.0.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.0.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.0.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.0.linear1.weight", + "decoder.actor_decoder.layers.0.linear1.bias", + "decoder.actor_decoder.layers.0.linear2.weight", + "decoder.actor_decoder.layers.0.linear2.bias", + "decoder.actor_decoder.layers.0.norm1.weight", + "decoder.actor_decoder.layers.0.norm1.bias", + "decoder.actor_decoder.layers.0.norm2.weight", + "decoder.actor_decoder.layers.0.norm2.bias", + "decoder.actor_decoder.layers.0.norm3.weight", + "decoder.actor_decoder.layers.0.norm3.bias", + "decoder.actor_decoder.layers.1.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.1.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.1.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.1.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.1.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.1.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.1.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.1.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.1.linear1.weight", + "decoder.actor_decoder.layers.1.linear1.bias", + "decoder.actor_decoder.layers.1.linear2.weight", + "decoder.actor_decoder.layers.1.linear2.bias", + "decoder.actor_decoder.layers.1.norm1.weight", + "decoder.actor_decoder.layers.1.norm1.bias", + "decoder.actor_decoder.layers.1.norm2.weight", + "decoder.actor_decoder.layers.1.norm2.bias", + "decoder.actor_decoder.layers.1.norm3.weight", + "decoder.actor_decoder.layers.1.norm3.bias", + "decoder.actor_decoder.layers.2.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.2.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.2.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.2.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.2.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.2.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.2.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.2.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.2.linear1.weight", + "decoder.actor_decoder.layers.2.linear1.bias", + "decoder.actor_decoder.layers.2.linear2.weight", + "decoder.actor_decoder.layers.2.linear2.bias", + "decoder.actor_decoder.layers.2.norm1.weight", + "decoder.actor_decoder.layers.2.norm1.bias", + "decoder.actor_decoder.layers.2.norm2.weight", + "decoder.actor_decoder.layers.2.norm2.bias", + "decoder.actor_decoder.layers.2.norm3.weight", + "decoder.actor_decoder.layers.2.norm3.bias", + "decoder.actor_decoder.layers.3.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.3.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.3.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.3.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.3.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.3.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.3.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.3.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.3.linear1.weight", + "decoder.actor_decoder.layers.3.linear1.bias", + "decoder.actor_decoder.layers.3.linear2.weight", + "decoder.actor_decoder.layers.3.linear2.bias", + "decoder.actor_decoder.layers.3.norm1.weight", + "decoder.actor_decoder.layers.3.norm1.bias", + "decoder.actor_decoder.layers.3.norm2.weight", + "decoder.actor_decoder.layers.3.norm2.bias", + "decoder.actor_decoder.layers.3.norm3.weight", + "decoder.actor_decoder.layers.3.norm3.bias", + "decoder.revealer_mean.weight", + "decoder.revealer_mean.bias", + "decoder.revealer_log_std.weight", + "decoder.revealer_log_std.bias", + "decoder.actor_mean.weight", + "decoder.actor_mean.bias", + "decoder.actor_log_std.weight", + "decoder.actor_log_std.bias", + "decoder.proposal_score.0.weight", + "decoder.proposal_score.0.bias", + "decoder.proposal_score.1.weight", + "decoder.proposal_score.1.bias" + ], + "missing_keys": [ + "backbone.depth_adapter.depth_proj.0.weight", + "backbone.depth_adapter.depth_proj.0.bias", + "backbone.depth_adapter.depth_proj.1.weight", + "backbone.depth_adapter.depth_proj.1.bias", + "backbone.depth_adapter.depth_proj.3.weight", + "backbone.depth_adapter.depth_proj.3.bias", + "backbone.depth_adapter.geometry_proj.0.weight", + "backbone.depth_adapter.geometry_proj.0.bias", + "backbone.depth_adapter.geometry_proj.1.weight", + "backbone.depth_adapter.geometry_proj.1.bias", + "backbone.depth_adapter.camera_proj.0.weight", + "backbone.depth_adapter.camera_proj.0.bias", + "backbone.depth_adapter.camera_proj.1.weight", + "backbone.depth_adapter.camera_proj.1.bias", + "fusion.geometry_fusion.attn.in_proj_weight", + "fusion.geometry_fusion.attn.in_proj_bias", + "fusion.geometry_fusion.attn.out_proj.weight", + "fusion.geometry_fusion.attn.out_proj.bias", + "fusion.geometry_fusion.gate.0.weight", + "fusion.geometry_fusion.gate.0.bias", + "fusion.geometry_fusion.gate.1.weight", + "fusion.geometry_fusion.gate.1.bias", + "fusion.geometry_fusion.gate.3.weight", + "fusion.geometry_fusion.gate.3.bias", + "fusion.geometry_fusion.out.0.weight", + "fusion.geometry_fusion.out.0.bias", + "fusion.geometry_fusion.out.1.weight", + "fusion.geometry_fusion.out.1.bias", + "memory.scene_memory.position_embedding", + "memory.scene_memory.bank_queries", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.in_proj_weight", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.in_proj_bias", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.out_proj.weight", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.out_proj.bias", + "memory.scene_memory.sequence_encoder.layers.0.linear1.weight", + "memory.scene_memory.sequence_encoder.layers.0.linear1.bias", + "memory.scene_memory.sequence_encoder.layers.0.linear2.weight", + "memory.scene_memory.sequence_encoder.layers.0.linear2.bias", + "memory.scene_memory.sequence_encoder.layers.0.norm1.weight", + "memory.scene_memory.sequence_encoder.layers.0.norm1.bias", + "memory.scene_memory.sequence_encoder.layers.0.norm2.weight", + "memory.scene_memory.sequence_encoder.layers.0.norm2.bias", + "memory.scene_memory.bank_attention.in_proj_weight", + "memory.scene_memory.bank_attention.in_proj_bias", + "memory.scene_memory.bank_attention.out_proj.weight", + "memory.scene_memory.bank_attention.out_proj.bias", + "memory.scene_memory.action_proj.0.weight", + "memory.scene_memory.action_proj.0.bias", + "memory.scene_memory.action_proj.1.weight", + "memory.scene_memory.action_proj.1.bias", + "memory.scene_memory.write_gate.0.weight", + "memory.scene_memory.write_gate.0.bias", + "memory.scene_memory.write_gate.1.weight", + "memory.scene_memory.write_gate.1.bias", + "memory.scene_memory.write_gate.3.weight", + "memory.scene_memory.write_gate.3.bias", + "memory.scene_memory.token_proj.0.weight", + "memory.scene_memory.token_proj.0.bias", + "memory.scene_memory.token_proj.1.weight", + "memory.scene_memory.token_proj.1.bias", + "memory.belief_memory.position_embedding", + "memory.belief_memory.bank_queries", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.in_proj_weight", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.in_proj_bias", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.out_proj.weight", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.out_proj.bias", + "memory.belief_memory.sequence_encoder.layers.0.linear1.weight", + "memory.belief_memory.sequence_encoder.layers.0.linear1.bias", + "memory.belief_memory.sequence_encoder.layers.0.linear2.weight", + "memory.belief_memory.sequence_encoder.layers.0.linear2.bias", + "memory.belief_memory.sequence_encoder.layers.0.norm1.weight", + "memory.belief_memory.sequence_encoder.layers.0.norm1.bias", + "memory.belief_memory.sequence_encoder.layers.0.norm2.weight", + "memory.belief_memory.sequence_encoder.layers.0.norm2.bias", + "memory.belief_memory.bank_attention.in_proj_weight", + "memory.belief_memory.bank_attention.in_proj_bias", + "memory.belief_memory.bank_attention.out_proj.weight", + "memory.belief_memory.bank_attention.out_proj.bias", + "memory.belief_memory.action_proj.0.weight", + "memory.belief_memory.action_proj.0.bias", + "memory.belief_memory.action_proj.1.weight", + "memory.belief_memory.action_proj.1.bias", + "memory.belief_memory.write_gate.0.weight", + "memory.belief_memory.write_gate.0.bias", + "memory.belief_memory.write_gate.1.weight", + "memory.belief_memory.write_gate.1.bias", + "memory.belief_memory.write_gate.3.weight", + "memory.belief_memory.write_gate.3.bias", + "memory.belief_memory.token_proj.0.weight", + "memory.belief_memory.token_proj.0.bias", + "memory.belief_memory.token_proj.1.weight", + "memory.belief_memory.token_proj.1.bias", + "decoder.arm_decoder.layers.0.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.0.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.0.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.0.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.0.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.0.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.0.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.0.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.0.linear1.weight", + "decoder.arm_decoder.layers.0.linear1.bias", + "decoder.arm_decoder.layers.0.linear2.weight", + "decoder.arm_decoder.layers.0.linear2.bias", + "decoder.arm_decoder.layers.0.norm1.weight", + "decoder.arm_decoder.layers.0.norm1.bias", + "decoder.arm_decoder.layers.0.norm2.weight", + "decoder.arm_decoder.layers.0.norm2.bias", + "decoder.arm_decoder.layers.0.norm3.weight", + "decoder.arm_decoder.layers.0.norm3.bias", + "decoder.arm_decoder.layers.1.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.1.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.1.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.1.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.1.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.1.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.1.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.1.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.1.linear1.weight", + "decoder.arm_decoder.layers.1.linear1.bias", + "decoder.arm_decoder.layers.1.linear2.weight", + "decoder.arm_decoder.layers.1.linear2.bias", + "decoder.arm_decoder.layers.1.norm1.weight", + "decoder.arm_decoder.layers.1.norm1.bias", + "decoder.arm_decoder.layers.1.norm2.weight", + "decoder.arm_decoder.layers.1.norm2.bias", + "decoder.arm_decoder.layers.1.norm3.weight", + "decoder.arm_decoder.layers.1.norm3.bias", + "decoder.arm_decoder.layers.2.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.2.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.2.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.2.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.2.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.2.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.2.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.2.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.2.linear1.weight", + "decoder.arm_decoder.layers.2.linear1.bias", + "decoder.arm_decoder.layers.2.linear2.weight", + "decoder.arm_decoder.layers.2.linear2.bias", + "decoder.arm_decoder.layers.2.norm1.weight", + "decoder.arm_decoder.layers.2.norm1.bias", + "decoder.arm_decoder.layers.2.norm2.weight", + "decoder.arm_decoder.layers.2.norm2.bias", + "decoder.arm_decoder.layers.2.norm3.weight", + "decoder.arm_decoder.layers.2.norm3.bias", + "decoder.arm_decoder.layers.3.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.3.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.3.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.3.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.3.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.3.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.3.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.3.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.3.linear1.weight", + "decoder.arm_decoder.layers.3.linear1.bias", + "decoder.arm_decoder.layers.3.linear2.weight", + "decoder.arm_decoder.layers.3.linear2.bias", + "decoder.arm_decoder.layers.3.norm1.weight", + "decoder.arm_decoder.layers.3.norm1.bias", + "decoder.arm_decoder.layers.3.norm2.weight", + "decoder.arm_decoder.layers.3.norm2.bias", + "decoder.arm_decoder.layers.3.norm3.weight", + "decoder.arm_decoder.layers.3.norm3.bias", + "decoder.arm_identity.weight", + "decoder.phase_adapter.weight", + "decoder.phase_adapter.bias", + "decoder.role_adapter.weight", + "decoder.role_adapter.bias", + "decoder.context_proj.0.weight", + "decoder.context_proj.0.bias", + "decoder.context_proj.1.weight", + "decoder.context_proj.1.bias", + "decoder.arm_head.0.weight", + "decoder.arm_head.0.bias", + "decoder.arm_head.1.weight", + "decoder.arm_head.1.bias", + "decoder.arm_mean.weight", + "decoder.arm_mean.bias", + "decoder.arm_log_std.weight", + "decoder.arm_log_std.bias", + "decoder.proposal_mode_head.0.weight", + "decoder.proposal_mode_head.0.bias", + "decoder.proposal_mode_head.1.weight", + "decoder.proposal_mode_head.1.bias", + "decoder.proposal_mode_head.3.weight", + "decoder.proposal_mode_head.3.bias", + "decoder.proposal_mode_embeddings.weight", + "decoder.proposal_slot_embeddings.weight", + "decoder.mode_residual_heads.0.0.weight", + "decoder.mode_residual_heads.0.0.bias", + "decoder.mode_residual_heads.0.1.weight", + "decoder.mode_residual_heads.0.1.bias", + "decoder.mode_residual_heads.0.3.weight", + "decoder.mode_residual_heads.0.3.bias", + "decoder.mode_residual_heads.1.0.weight", + "decoder.mode_residual_heads.1.0.bias", + "decoder.mode_residual_heads.1.1.weight", + "decoder.mode_residual_heads.1.1.bias", + "decoder.mode_residual_heads.1.3.weight", + "decoder.mode_residual_heads.1.3.bias", + "decoder.mode_residual_heads.2.0.weight", + "decoder.mode_residual_heads.2.0.bias", + "decoder.mode_residual_heads.2.1.weight", + "decoder.mode_residual_heads.2.1.bias", + "decoder.mode_residual_heads.2.3.weight", + "decoder.mode_residual_heads.2.3.bias", + "decoder.mode_residual_heads.3.0.weight", + "decoder.mode_residual_heads.3.0.bias", + "decoder.mode_residual_heads.3.1.weight", + "decoder.mode_residual_heads.3.1.bias", + "decoder.mode_residual_heads.3.3.weight", + "decoder.mode_residual_heads.3.3.bias", + "decoder.mode_residual_heads.4.0.weight", + "decoder.mode_residual_heads.4.0.bias", + "decoder.mode_residual_heads.4.1.weight", + "decoder.mode_residual_heads.4.1.bias", + "decoder.mode_residual_heads.4.3.weight", + "decoder.mode_residual_heads.4.3.bias", + "decoder.mode_residual_heads.5.0.weight", + "decoder.mode_residual_heads.5.0.bias", + "decoder.mode_residual_heads.5.1.weight", + "decoder.mode_residual_heads.5.1.bias", + "decoder.mode_residual_heads.5.3.weight", + "decoder.mode_residual_heads.5.3.bias", + "decoder.slot_delta.0.weight", + "decoder.slot_delta.0.bias", + "decoder.slot_delta.1.weight", + "decoder.slot_delta.1.bias", + "decoder.slot_delta.3.weight", + "decoder.slot_delta.3.bias", + "decoder.proposal_score.0.weight", + "decoder.proposal_score.0.bias", + "decoder.proposal_score.1.weight", + "decoder.proposal_score.1.bias", + "decoder.proposal_score.3.weight", + "decoder.proposal_score.3.bias", + "elastic_state_head.interaction_queries", + "elastic_state_head.interaction_attention.in_proj_weight", + "elastic_state_head.interaction_attention.in_proj_bias", + "elastic_state_head.interaction_attention.out_proj.weight", + "elastic_state_head.interaction_attention.out_proj.bias", + "elastic_state_head.interaction_mlp.0.weight", + "elastic_state_head.interaction_mlp.0.bias", + "elastic_state_head.interaction_mlp.1.weight", + "elastic_state_head.interaction_mlp.1.bias", + "elastic_state_head.interaction_mlp.3.weight", + "elastic_state_head.interaction_mlp.3.bias", + "elastic_state_head.decoder.field_queries", + "elastic_state_head.decoder.field_attention.in_proj_weight", + "elastic_state_head.decoder.field_attention.in_proj_bias", + "elastic_state_head.decoder.field_attention.out_proj.weight", + "elastic_state_head.decoder.field_attention.out_proj.bias", + "elastic_state_head.decoder.field_mlp.0.weight", + "elastic_state_head.decoder.field_mlp.0.bias", + "elastic_state_head.decoder.field_mlp.1.weight", + "elastic_state_head.decoder.field_mlp.1.bias", + "elastic_state_head.decoder.field_mlp.3.weight", + "elastic_state_head.decoder.field_mlp.3.bias", + "elastic_state_head.decoder.summary_proj.0.weight", + "elastic_state_head.decoder.summary_proj.0.bias", + "elastic_state_head.decoder.summary_proj.1.weight", + "elastic_state_head.decoder.summary_proj.1.bias", + "elastic_state_head.decoder.phase_head.0.weight", + "elastic_state_head.decoder.phase_head.0.bias", + "elastic_state_head.decoder.phase_head.1.weight", + "elastic_state_head.decoder.phase_head.1.bias", + "elastic_state_head.decoder.phase_head.3.weight", + "elastic_state_head.decoder.phase_head.3.bias", + "elastic_state_head.decoder.arm_role_head.0.weight", + "elastic_state_head.decoder.arm_role_head.0.bias", + "elastic_state_head.decoder.arm_role_head.1.weight", + "elastic_state_head.decoder.arm_role_head.1.bias", + "elastic_state_head.decoder.arm_role_head.3.weight", + "elastic_state_head.decoder.arm_role_head.3.bias", + "elastic_state_head.decoder.arm_identity.weight", + "elastic_state_head.decoder.support_mode.0.weight", + "elastic_state_head.decoder.support_mode.0.bias", + "elastic_state_head.decoder.support_mode.1.weight", + "elastic_state_head.decoder.support_mode.1.bias", + "elastic_state_head.decoder.support_mode.3.weight", + "elastic_state_head.decoder.support_mode.3.bias", + "elastic_state_head.decoder.access_field.weight", + "elastic_state_head.decoder.access_field.bias", + "elastic_state_head.decoder.target_belief_field.weight", + "elastic_state_head.decoder.target_belief_field.bias", + "elastic_state_head.decoder.visibility_field.weight", + "elastic_state_head.decoder.visibility_field.bias", + "elastic_state_head.decoder.clearance_field.weight", + "elastic_state_head.decoder.clearance_field.bias", + "elastic_state_head.decoder.occluder_contact_field.weight", + "elastic_state_head.decoder.occluder_contact_field.bias", + "elastic_state_head.decoder.grasp_affordance_field.weight", + "elastic_state_head.decoder.grasp_affordance_field.bias", + "elastic_state_head.decoder.support_stability_field.weight", + "elastic_state_head.decoder.support_stability_field.bias", + "elastic_state_head.decoder.persistence_field.weight", + "elastic_state_head.decoder.persistence_field.bias", + "elastic_state_head.decoder.reocclusion_field.weight", + "elastic_state_head.decoder.reocclusion_field.bias", + "elastic_state_head.decoder.disturbance_field.weight", + "elastic_state_head.decoder.disturbance_field.bias", + "elastic_state_head.decoder.uncertainty_field.weight", + "elastic_state_head.decoder.uncertainty_field.bias", + "elastic_state_head.decoder.reocclusion_head.0.weight", + "elastic_state_head.decoder.reocclusion_head.0.bias", + "elastic_state_head.decoder.reocclusion_head.1.weight", + "elastic_state_head.decoder.reocclusion_head.1.bias", + "elastic_state_head.decoder.reocclusion_head.3.weight", + "elastic_state_head.decoder.reocclusion_head.3.bias", + "world_model.state_encoder.0.weight", + "world_model.state_encoder.0.bias", + "world_model.state_encoder.1.weight", + "world_model.state_encoder.1.bias", + "world_model.scene_memory_proj.0.weight", + "world_model.scene_memory_proj.0.bias", + "world_model.scene_memory_proj.1.weight", + "world_model.scene_memory_proj.1.bias", + "world_model.belief_memory_proj.0.weight", + "world_model.belief_memory_proj.0.bias", + "world_model.belief_memory_proj.1.weight", + "world_model.belief_memory_proj.1.bias", + "world_model.action_encoder.0.weight", + "world_model.action_encoder.0.bias", + "world_model.action_encoder.1.weight", + "world_model.action_encoder.1.bias", + "world_model.transition.weight_ih", + "world_model.transition.weight_hh", + "world_model.transition.bias_ih", + "world_model.transition.bias_hh", + "world_model.scene_memory_update.weight", + "world_model.scene_memory_update.bias", + "world_model.belief_memory_update.weight", + "world_model.belief_memory_update.bias", + "world_model.compact_decoder.weight", + "world_model.compact_decoder.bias", + "world_model.target_belief_head.weight", + "world_model.target_belief_head.bias", + "world_model.visibility_head.weight", + "world_model.visibility_head.bias", + "world_model.clearance_head.weight", + "world_model.clearance_head.bias", + "world_model.occluder_contact_head.weight", + "world_model.occluder_contact_head.bias", + "world_model.grasp_affordance_head.weight", + "world_model.grasp_affordance_head.bias", + "world_model.support_stability_head.weight", + "world_model.support_stability_head.bias", + "world_model.persistence_head.weight", + "world_model.persistence_head.bias", + "world_model.reocclusion_head.weight", + "world_model.reocclusion_head.bias", + "world_model.disturbance_head.weight", + "world_model.disturbance_head.bias", + "world_model.uncertainty_head.weight", + "world_model.uncertainty_head.bias", + "world_model.access_head.weight", + "world_model.access_head.bias", + "planner.residual.trunk.0.weight", + "planner.residual.trunk.0.bias", + "planner.residual.trunk.1.weight", + "planner.residual.trunk.1.bias", + "planner.residual.trunk.3.weight", + "planner.residual.trunk.3.bias", + "planner.residual.success_head.weight", + "planner.residual.success_head.bias", + "planner.residual.risk_head.weight", + "planner.residual.risk_head.bias", + "planner.residual.residual_head.weight", + "planner.residual.residual_head.bias" + ], + "unexpected_keys": [] + } +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_full/reveal_benchmark.json b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_full/reveal_benchmark.json new file mode 100644 index 0000000000000000000000000000000000000000..fee80de26bb0a101ae9b1934035a195ffa8592ae --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_full/reveal_benchmark.json @@ -0,0 +1,15 @@ +{ + "full": { + "per_task_success": { + "foliage_proxy": 0.4583333333333333, + "bag_proxy": 0.5, + "cloth_proxy": 0.6666666666666666 + }, + "mean_success": 0.5416666666666666, + "visibility_integral": 34.84331858654817, + "corridor_availability": 0.8431700393557549, + "reocclusion_rate": 0.023209473466003317, + "persistence_horizon_mae": 0.7628522404578011, + "disturbance_cost": 0.21749323016653457 + } +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_full/reveal_benchmark.md b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_full/reveal_benchmark.md new file mode 100644 index 0000000000000000000000000000000000000000..6ab5b46eee7645483079394f29a1d3bebde1e86f --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_full/reveal_benchmark.md @@ -0,0 +1,13 @@ +# Reveal Proxy Benchmark + +## full +- checkpoint: /workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/checkpoint_best.pt +- mean_success: 0.542 +- visibility_integral: 34.843 +- corridor_availability: 0.843 +- reocclusion_rate: 0.023 +- persistence_horizon_mae: 0.763 +- disturbance_cost: 0.217 +- foliage_proxy_success: 0.458 +- bag_proxy_success: 0.500 +- cloth_proxy_success: 0.667 diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_no_depth/reveal_benchmark.json b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_no_depth/reveal_benchmark.json new file mode 100644 index 0000000000000000000000000000000000000000..dcf0a3cc1deb388def5c75e2bfcc2c9078cc381c --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_no_depth/reveal_benchmark.json @@ -0,0 +1,15 @@ +{ + "full": { + "per_task_success": { + "foliage_proxy": 0.375, + "bag_proxy": 0.4166666666666667, + "cloth_proxy": 0.6666666666666666 + }, + "mean_success": 0.48611111111111116, + "visibility_integral": 40.40390850769149, + "corridor_availability": 0.8780669139491187, + "reocclusion_rate": 0.020889146439519574, + "persistence_horizon_mae": 0.529351558783127, + "disturbance_cost": 0.27779827603242463 + } +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_no_depth/reveal_benchmark.md b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_no_depth/reveal_benchmark.md new file mode 100644 index 0000000000000000000000000000000000000000..df89043065dbd90de6930408a6e955c0e714bc93 --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/benchmark_no_depth/reveal_benchmark.md @@ -0,0 +1,13 @@ +# Reveal Proxy Benchmark + +## full +- checkpoint: /workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/checkpoint_best.pt +- mean_success: 0.486 +- visibility_integral: 40.404 +- corridor_availability: 0.878 +- reocclusion_rate: 0.021 +- persistence_horizon_mae: 0.529 +- disturbance_cost: 0.278 +- foliage_proxy_success: 0.375 +- bag_proxy_success: 0.417 +- cloth_proxy_success: 0.667 diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/config_resolved.yaml b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/config_resolved.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a02b2cf736be139e064cb0834562eee7b9210c49 --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/config_resolved.yaml @@ -0,0 +1,149 @@ +experiment_name: proxy_interaction_r3d_stage3_clip_rgbd_seed19 +output_dir: /workspace/VLAarchtests/artifacts/outputs/r3d +device: cuda +seed: 19 +init_checkpoint: /workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt +init_strict: false +data: + proxies: + - foliage_proxy + - bag_proxy + - cloth_proxy + resolution: 224 + dataset_version: reveal_proxy_v6_rgbd_elastic_state + train_episodes_per_proxy: 48 + val_episodes_per_proxy: 16 + train_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_train_clip224_v6_rgbd_stage3_seed19.pt + val_dataset_path: /workspace/VLAarchtests/artifacts/data/reveal_proxy/proxy_val_clip224_v6_rgbd_stage3_seed19.pt + rebuild_dataset: false + chunk_horizon: 8 + rollout_horizon: 5 + history_steps: 6 + planner_candidates: 8 + seed: 19 +optim: + epochs: 4 + batch_size: 2 + num_workers: 4 + lr: 0.0003 + weight_decay: 0.0001 +trainer: + policy_type: elastic_reveal + use_bf16: true + grad_clip_norm: 1.0 + freeze_backbone: true + gradient_checkpointing: false + plan_during_train: true + plan_during_eval: true + support_mode_conditioning: true + planner_mode: trainable + use_depth: true + use_world_model: true + use_role_tokens: true + compute_equivariance_probe: true +policy: + backbone: + model_name: openai/clip-vit-base-patch32 + hidden_dim: 512 + max_text_tokens: 32 + freeze_backbone: true + gradient_checkpointing: false + use_dummy_backbone: false + fusion: + hidden_dim: 512 + num_cameras: 3 + num_layers: 4 + num_heads: 8 + ff_dim: 2048 + dropout: 0.1 + proprio_dim: 32 + proprio_tokens: 1 + memory: + hidden_dim: 512 + action_dim: 14 + history_steps: 6 + scene_history_steps: 3 + belief_history_steps: 8 + num_layers: 2 + dropout: 0.1 + memory_bank_size: 4 + scene_bank_size: 2 + belief_bank_size: 2 + num_heads: 8 + max_history_steps: 8 + decoder: + hidden_dim: 512 + num_heads: 8 + num_layers: 4 + ff_dim: 2048 + dropout: 0.1 + chunk_size: 8 + action_dim: 14 + arm_action_dim: 7 + num_candidates: 8 + num_phases: 5 + num_arm_roles: 4 + num_proposal_modes: 6 + planner_top_k: 4 + reveal_head: + hidden_dim: 512 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + belief_map_size: 32 + field_size: 16 + num_heads: 8 + predict_belief_map: true + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + world_model: + hidden_dim: 512 + action_dim: 14 + num_support_modes: 3 + num_approach_templates: 32 + rollout_horizon: 5 + field_size: 16 + num_heads: 8 + num_phases: 5 + num_arm_roles: 4 + num_interaction_tokens: 8 + belief_map_size: 32 + predict_belief_map: true + scene_bank_size: 2 + belief_bank_size: 2 + planner: + hidden_dim: 512 + num_candidates: 8 + action_dim: 14 + num_support_modes: 3 + utility_margin: 0.1 + num_heads: 8 + num_layers: 2 + num_phases: 5 + num_arm_roles: 4 + top_k: 4 +loss_weights: + action: 1.0 + phase: 0.1 + arm_role: 0.15 + support_mode: 0.1 + corridor: 0.15 + persistence: 0.05 + disturbance: 0.05 + world_model: 0.25 + belief: 0.05 + visibility: 0.05 + clearance: 0.05 + support_stability: 0.05 + reocclusion: 0.05 + occluder_contact: 0.05 + grasp_affordance: 0.05 + planner_success: 0.25 + planner_risk: 0.1 + planner_ranking: 0.2 + proposal_reconstruction: 0.1 + proposal_success: 0.15 + proposal_ranking: 0.2 + proposal_diversity: 0.05 + role_swap_consistency: 0.05 diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/diagnostics_full/proxy_diagnostics.json b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/diagnostics_full/proxy_diagnostics.json new file mode 100644 index 0000000000000000000000000000000000000000..de31173b7b32be78da2a447dbba0efc26d023869 --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/diagnostics_full/proxy_diagnostics.json @@ -0,0 +1,16 @@ +{ + "planner_top1_accuracy": 0.3560606060606061, + "planner_regret": 0.01967032626271248, + "planner_score_utility_spearman": 0.27878788113594055, + "risk_calibration_mse": 0.012325962074100971, + "role_collapse_rate": 0.0, + "proposal_diversity": 0.02638203836977482, + "left_right_equivariance_error": 0.00017077481739265337, + "belief_calibration_brier": 0.003630289575085044, + "reocclusion_calibration_brier": 0.280130535364151, + "support_stability_mae": 0.02492167428135872, + "clearance_auc": 0.9725896896303129, + "memory_write_rate": 0.0, + "memory_saturation": 0.3340458869934082, + "num_samples": 132 +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/metrics.json b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/metrics.json new file mode 100644 index 0000000000000000000000000000000000000000..ab3d3904c65a77be6021bbfe8925849d7846b1dc --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/metrics.json @@ -0,0 +1,230 @@ +[ + { + "epoch": 0, + "train": { + "action": 0.026936710870971804, + "arm_role": 0.03045270662558706, + "belief": 0.11832109189739352, + "clearance": 0.09567467718336144, + "corridor": 0.29363441146714125, + "disturbance": 0.004809834103662069, + "grasp_affordance": 0.020342445986247376, + "occluder_contact": 0.2152292126887723, + "persistence": 5.77494593756169, + "phase": 0.7668704384251644, + "planner_ranking": 0.15500873614793434, + "planner_risk": 0.01612282053492412, + "planner_success": 0.6117850280906024, + "proposal_diversity": 0.0, + "proposal_ranking": 1.2648823336551065, + "proposal_reconstruction": 0.06841283142566681, + "proposal_success": 0.6819258235002819, + "reocclusion": 0.7116702893846913, + "role_swap_consistency": 0.0005399687355843764, + "support_mode": 0.7442666304738899, + "support_stability": 0.15860560094251444, + "total": 1.8048903320965013, + "uncertainty": 0.020460822854472913, + "visibility": 0.1093550581681101, + "world_model": 2.6795224861094824 + }, + "val": { + "action": 0.02333411630807501, + "arm_role": 8.984091993735285e-05, + "belief": 0.09130037987999844, + "clearance": 0.07433769293129444, + "corridor": 0.2433638901760181, + "disturbance": 0.0018495344263624936, + "grasp_affordance": 0.007560361268450365, + "occluder_contact": 0.1924103627150709, + "persistence": 7.665606745309767, + "phase": 0.6567134970065319, + "planner_ranking": 0.0353548031636654, + "planner_risk": 0.011407233709398883, + "planner_success": 0.5408762232823805, + "proposal_diversity": 0.0, + "proposal_ranking": 1.1750569343566895, + "proposal_reconstruction": 0.06537803668867458, + "proposal_success": 0.669422728545738, + "reocclusion": 0.6932079782991698, + "role_swap_consistency": 0.0, + "support_mode": 0.6559179944522453, + "support_stability": 0.16087975142749422, + "total": 1.7218121037338718, + "uncertainty": 0.0011911591864190996, + "visibility": 0.08501114086671309, + "world_model": 2.386733156262022 + } + }, + { + "epoch": 1, + "train": { + "action": 0.022530869659232466, + "arm_role": 6.122400886134098e-05, + "belief": 0.09997689045573535, + "clearance": 0.0787699304128948, + "corridor": 0.2502592285604853, + "disturbance": 0.003108920491142367, + "grasp_affordance": 0.009875784261691336, + "occluder_contact": 0.19081604229776483, + "persistence": 4.516820472091508, + "phase": 0.6890445106907894, + "planner_ranking": 0.04759743569987942, + "planner_risk": 0.010974906991376836, + "planner_success": 0.5852321183210925, + "proposal_diversity": 0.0, + "proposal_ranking": 1.1509872969828154, + "proposal_reconstruction": 0.06461040112925204, + "proposal_success": 0.6747973046804729, + "reocclusion": 0.6984834761996018, + "role_swap_consistency": 0.0005843641104691319, + "support_mode": 0.6843904194078947, + "support_stability": 0.14238819966190738, + "total": 1.5274493932723998, + "uncertainty": 0.004225814943569467, + "visibility": 0.09563646316528321, + "world_model": 2.1734342518605683 + }, + "val": { + "action": 0.023799537031939537, + "arm_role": 3.402016784203818e-05, + "belief": 0.10602470622821288, + "clearance": 0.07505167337755363, + "corridor": 0.2335988457004229, + "disturbance": 0.004122847163561153, + "grasp_affordance": 0.008018463842965888, + "occluder_contact": 0.18937847763299942, + "persistence": 4.54103897376494, + "phase": 0.657283675941554, + "planner_ranking": 0.033948301364034705, + "planner_risk": 0.01159033985723826, + "planner_success": 0.5285389811703654, + "proposal_diversity": 0.0, + "proposal_ranking": 1.1224055741772507, + "proposal_reconstruction": 0.06609346220890681, + "proposal_success": 0.6804587317235542, + "reocclusion": 0.6936469610893365, + "role_swap_consistency": 0.0, + "support_mode": 0.7072254699287992, + "support_stability": 0.1417240992640004, + "total": 1.518586855946165, + "uncertainty": 0.0017800588499416006, + "visibility": 0.10126593957344691, + "world_model": 2.228069293679613 + } + }, + { + "epoch": 2, + "train": { + "action": 0.02169373267771382, + "arm_role": 1.632916299920333e-05, + "belief": 0.10102678033474245, + "clearance": 0.08151812365180568, + "corridor": 0.2404590189064804, + "disturbance": 0.0036459605661512367, + "grasp_affordance": 0.009210268979107863, + "occluder_contact": 0.19557440986758784, + "persistence": 4.1599240601062775, + "phase": 0.7287764699835526, + "planner_ranking": 0.03455519499108636, + "planner_risk": 0.010664231259541242, + "planner_success": 0.5189147469244505, + "proposal_diversity": 0.0, + "proposal_ranking": 1.135868281439731, + "proposal_reconstruction": 0.06376664577738235, + "proposal_success": 0.6781208530852669, + "reocclusion": 0.6952229437075164, + "role_swap_consistency": 0.00026992922978120315, + "support_mode": 0.6853207236842105, + "support_stability": 0.14708005885936712, + "total": 1.4576912641525268, + "uncertainty": 0.0022136893562243417, + "visibility": 0.09373094598321538, + "world_model": 2.0446383184508274 + }, + "val": { + "action": 0.024303340324849793, + "arm_role": 2.400410440631359e-05, + "belief": 0.09132557988844135, + "clearance": 0.0747725333679806, + "corridor": 0.22321437017032594, + "disturbance": 0.0018154052629666473, + "grasp_affordance": 0.008783090055327524, + "occluder_contact": 0.19523962925780902, + "persistence": 4.016559170954155, + "phase": 0.655967711499243, + "planner_ranking": 0.029446791824973458, + "planner_risk": 0.011213887666323872, + "planner_success": 0.48344053057107056, + "proposal_diversity": 0.0, + "proposal_ranking": 1.123392726435806, + "proposal_reconstruction": 0.0669570304453373, + "proposal_success": 0.6750701024676814, + "reocclusion": 0.6929633816083273, + "role_swap_consistency": 0.0, + "support_mode": 0.6728262007236481, + "support_stability": 0.16180097749174543, + "total": 1.44674982807853, + "uncertainty": 0.0029518954829792633, + "visibility": 0.08370742495312836, + "world_model": 2.116791232065721 + } + }, + { + "epoch": 3, + "train": { + "action": 0.020692493430779953, + "arm_role": 0.0010310888290405273, + "belief": 0.09604211818230779, + "clearance": 0.07567685139610579, + "corridor": 0.23569351255795673, + "disturbance": 0.002073299908635674, + "grasp_affordance": 0.009045833137229477, + "occluder_contact": 0.1890327513217926, + "persistence": 1.4987196951724686, + "phase": 0.6707082648026316, + "planner_ranking": 0.029076422584116222, + "planner_risk": 0.010615614641114678, + "planner_success": 0.5114178424603061, + "proposal_diversity": 0.0, + "proposal_ranking": 1.1341317979913008, + "proposal_reconstruction": 0.06304936632514, + "proposal_success": 0.6777858323172519, + "reocclusion": 0.6921605395643334, + "role_swap_consistency": 0.0008031590498974669, + "support_mode": 0.6671489514802632, + "support_stability": 0.13422315941241228, + "total": 1.2908767502558858, + "uncertainty": 0.0018180830574860201, + "visibility": 0.08983320496780307, + "world_model": 1.96784260210238 + }, + "val": { + "action": 0.029847019775347275, + "arm_role": 0.0005091276556909593, + "belief": 0.0935861895030195, + "clearance": 0.06930065837999184, + "corridor": 0.1947595030633789, + "disturbance": 0.0010440175451824898, + "grasp_affordance": 0.008902763256408049, + "occluder_contact": 0.18518025540944302, + "persistence": 0.968033335425637, + "phase": 0.6549982862039045, + "planner_ranking": 0.029935677410652384, + "planner_risk": 0.012325948773650452, + "planner_success": 0.5715719493049564, + "proposal_diversity": 0.0, + "proposal_ranking": 1.151603403416547, + "proposal_reconstruction": 0.07453863706552621, + "proposal_success": 0.666525122794238, + "reocclusion": 0.6971786527922659, + "role_swap_consistency": 0.0, + "support_mode": 0.8148123341979403, + "support_stability": 0.1445934563931642, + "total": 1.2090845984039884, + "uncertainty": 0.0011929568548415166, + "visibility": 0.08995764076032421, + "world_model": 1.6087976585734973 + } + } +] \ No newline at end of file diff --git a/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/summary.json b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..8b6ad77baee8c958086316a5cb7c4a21d2213943 --- /dev/null +++ b/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/summary.json @@ -0,0 +1,557 @@ +{ + "experiment_name": "proxy_interaction_r3d_stage3_clip_rgbd_seed19", + "device": "cuda", + "best_checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage3_clip_rgbd_seed19/checkpoint_best.pt", + "final_train_total": 1.2908767502558858, + "final_val_total": 1.2090845984039884, + "train_time_sec": 148.42910027503967, + "peak_gpu_memory_mb": 1960.35546875, + "num_train_samples": 380, + "num_val_samples": 132, + "planner_mode": "trainable", + "frozen_modules": [], + "init_info": { + "path": "/workspace/VLAarchtests/artifacts/outputs/reveal_runs/proxy_backbone_only_clip/checkpoint_best.pt", + "loaded_keys": 461, + "skipped_shape_mismatch_keys": [ + "memory.gru.weight_ih_l0", + "memory.gru.weight_hh_l0", + "memory.gru.bias_ih_l0", + "memory.gru.bias_hh_l0", + "memory.token_proj.0.weight", + "memory.token_proj.0.bias", + "memory.token_proj.1.weight", + "memory.token_proj.1.bias", + "decoder.actor_role_bias", + "decoder.revealer_decoder.layers.0.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.0.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.0.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.0.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.0.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.0.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.0.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.0.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.0.linear1.weight", + "decoder.revealer_decoder.layers.0.linear1.bias", + "decoder.revealer_decoder.layers.0.linear2.weight", + "decoder.revealer_decoder.layers.0.linear2.bias", + "decoder.revealer_decoder.layers.0.norm1.weight", + "decoder.revealer_decoder.layers.0.norm1.bias", + "decoder.revealer_decoder.layers.0.norm2.weight", + "decoder.revealer_decoder.layers.0.norm2.bias", + "decoder.revealer_decoder.layers.0.norm3.weight", + "decoder.revealer_decoder.layers.0.norm3.bias", + "decoder.revealer_decoder.layers.1.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.1.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.1.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.1.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.1.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.1.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.1.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.1.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.1.linear1.weight", + "decoder.revealer_decoder.layers.1.linear1.bias", + "decoder.revealer_decoder.layers.1.linear2.weight", + "decoder.revealer_decoder.layers.1.linear2.bias", + "decoder.revealer_decoder.layers.1.norm1.weight", + "decoder.revealer_decoder.layers.1.norm1.bias", + "decoder.revealer_decoder.layers.1.norm2.weight", + "decoder.revealer_decoder.layers.1.norm2.bias", + "decoder.revealer_decoder.layers.1.norm3.weight", + "decoder.revealer_decoder.layers.1.norm3.bias", + "decoder.revealer_decoder.layers.2.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.2.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.2.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.2.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.2.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.2.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.2.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.2.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.2.linear1.weight", + "decoder.revealer_decoder.layers.2.linear1.bias", + "decoder.revealer_decoder.layers.2.linear2.weight", + "decoder.revealer_decoder.layers.2.linear2.bias", + "decoder.revealer_decoder.layers.2.norm1.weight", + "decoder.revealer_decoder.layers.2.norm1.bias", + "decoder.revealer_decoder.layers.2.norm2.weight", + "decoder.revealer_decoder.layers.2.norm2.bias", + "decoder.revealer_decoder.layers.2.norm3.weight", + "decoder.revealer_decoder.layers.2.norm3.bias", + "decoder.revealer_decoder.layers.3.self_attn.in_proj_weight", + "decoder.revealer_decoder.layers.3.self_attn.in_proj_bias", + "decoder.revealer_decoder.layers.3.self_attn.out_proj.weight", + "decoder.revealer_decoder.layers.3.self_attn.out_proj.bias", + "decoder.revealer_decoder.layers.3.multihead_attn.in_proj_weight", + "decoder.revealer_decoder.layers.3.multihead_attn.in_proj_bias", + "decoder.revealer_decoder.layers.3.multihead_attn.out_proj.weight", + "decoder.revealer_decoder.layers.3.multihead_attn.out_proj.bias", + "decoder.revealer_decoder.layers.3.linear1.weight", + "decoder.revealer_decoder.layers.3.linear1.bias", + "decoder.revealer_decoder.layers.3.linear2.weight", + "decoder.revealer_decoder.layers.3.linear2.bias", + "decoder.revealer_decoder.layers.3.norm1.weight", + "decoder.revealer_decoder.layers.3.norm1.bias", + "decoder.revealer_decoder.layers.3.norm2.weight", + "decoder.revealer_decoder.layers.3.norm2.bias", + "decoder.revealer_decoder.layers.3.norm3.weight", + "decoder.revealer_decoder.layers.3.norm3.bias", + "decoder.actor_decoder.layers.0.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.0.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.0.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.0.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.0.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.0.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.0.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.0.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.0.linear1.weight", + "decoder.actor_decoder.layers.0.linear1.bias", + "decoder.actor_decoder.layers.0.linear2.weight", + "decoder.actor_decoder.layers.0.linear2.bias", + "decoder.actor_decoder.layers.0.norm1.weight", + "decoder.actor_decoder.layers.0.norm1.bias", + "decoder.actor_decoder.layers.0.norm2.weight", + "decoder.actor_decoder.layers.0.norm2.bias", + "decoder.actor_decoder.layers.0.norm3.weight", + "decoder.actor_decoder.layers.0.norm3.bias", + "decoder.actor_decoder.layers.1.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.1.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.1.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.1.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.1.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.1.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.1.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.1.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.1.linear1.weight", + "decoder.actor_decoder.layers.1.linear1.bias", + "decoder.actor_decoder.layers.1.linear2.weight", + "decoder.actor_decoder.layers.1.linear2.bias", + "decoder.actor_decoder.layers.1.norm1.weight", + "decoder.actor_decoder.layers.1.norm1.bias", + "decoder.actor_decoder.layers.1.norm2.weight", + "decoder.actor_decoder.layers.1.norm2.bias", + "decoder.actor_decoder.layers.1.norm3.weight", + "decoder.actor_decoder.layers.1.norm3.bias", + "decoder.actor_decoder.layers.2.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.2.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.2.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.2.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.2.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.2.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.2.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.2.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.2.linear1.weight", + "decoder.actor_decoder.layers.2.linear1.bias", + "decoder.actor_decoder.layers.2.linear2.weight", + "decoder.actor_decoder.layers.2.linear2.bias", + "decoder.actor_decoder.layers.2.norm1.weight", + "decoder.actor_decoder.layers.2.norm1.bias", + "decoder.actor_decoder.layers.2.norm2.weight", + "decoder.actor_decoder.layers.2.norm2.bias", + "decoder.actor_decoder.layers.2.norm3.weight", + "decoder.actor_decoder.layers.2.norm3.bias", + "decoder.actor_decoder.layers.3.self_attn.in_proj_weight", + "decoder.actor_decoder.layers.3.self_attn.in_proj_bias", + "decoder.actor_decoder.layers.3.self_attn.out_proj.weight", + "decoder.actor_decoder.layers.3.self_attn.out_proj.bias", + "decoder.actor_decoder.layers.3.multihead_attn.in_proj_weight", + "decoder.actor_decoder.layers.3.multihead_attn.in_proj_bias", + "decoder.actor_decoder.layers.3.multihead_attn.out_proj.weight", + "decoder.actor_decoder.layers.3.multihead_attn.out_proj.bias", + "decoder.actor_decoder.layers.3.linear1.weight", + "decoder.actor_decoder.layers.3.linear1.bias", + "decoder.actor_decoder.layers.3.linear2.weight", + "decoder.actor_decoder.layers.3.linear2.bias", + "decoder.actor_decoder.layers.3.norm1.weight", + "decoder.actor_decoder.layers.3.norm1.bias", + "decoder.actor_decoder.layers.3.norm2.weight", + "decoder.actor_decoder.layers.3.norm2.bias", + "decoder.actor_decoder.layers.3.norm3.weight", + "decoder.actor_decoder.layers.3.norm3.bias", + "decoder.revealer_mean.weight", + "decoder.revealer_mean.bias", + "decoder.revealer_log_std.weight", + "decoder.revealer_log_std.bias", + "decoder.actor_mean.weight", + "decoder.actor_mean.bias", + "decoder.actor_log_std.weight", + "decoder.actor_log_std.bias", + "decoder.proposal_score.0.weight", + "decoder.proposal_score.0.bias", + "decoder.proposal_score.1.weight", + "decoder.proposal_score.1.bias" + ], + "missing_keys": [ + "backbone.depth_adapter.depth_proj.0.weight", + "backbone.depth_adapter.depth_proj.0.bias", + "backbone.depth_adapter.depth_proj.1.weight", + "backbone.depth_adapter.depth_proj.1.bias", + "backbone.depth_adapter.depth_proj.3.weight", + "backbone.depth_adapter.depth_proj.3.bias", + "backbone.depth_adapter.geometry_proj.0.weight", + "backbone.depth_adapter.geometry_proj.0.bias", + "backbone.depth_adapter.geometry_proj.1.weight", + "backbone.depth_adapter.geometry_proj.1.bias", + "backbone.depth_adapter.camera_proj.0.weight", + "backbone.depth_adapter.camera_proj.0.bias", + "backbone.depth_adapter.camera_proj.1.weight", + "backbone.depth_adapter.camera_proj.1.bias", + "fusion.geometry_fusion.attn.in_proj_weight", + "fusion.geometry_fusion.attn.in_proj_bias", + "fusion.geometry_fusion.attn.out_proj.weight", + "fusion.geometry_fusion.attn.out_proj.bias", + "fusion.geometry_fusion.gate.0.weight", + "fusion.geometry_fusion.gate.0.bias", + "fusion.geometry_fusion.gate.1.weight", + "fusion.geometry_fusion.gate.1.bias", + "fusion.geometry_fusion.gate.3.weight", + "fusion.geometry_fusion.gate.3.bias", + "fusion.geometry_fusion.out.0.weight", + "fusion.geometry_fusion.out.0.bias", + "fusion.geometry_fusion.out.1.weight", + "fusion.geometry_fusion.out.1.bias", + "memory.scene_memory.position_embedding", + "memory.scene_memory.bank_queries", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.in_proj_weight", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.in_proj_bias", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.out_proj.weight", + "memory.scene_memory.sequence_encoder.layers.0.self_attn.out_proj.bias", + "memory.scene_memory.sequence_encoder.layers.0.linear1.weight", + "memory.scene_memory.sequence_encoder.layers.0.linear1.bias", + "memory.scene_memory.sequence_encoder.layers.0.linear2.weight", + "memory.scene_memory.sequence_encoder.layers.0.linear2.bias", + "memory.scene_memory.sequence_encoder.layers.0.norm1.weight", + "memory.scene_memory.sequence_encoder.layers.0.norm1.bias", + "memory.scene_memory.sequence_encoder.layers.0.norm2.weight", + "memory.scene_memory.sequence_encoder.layers.0.norm2.bias", + "memory.scene_memory.bank_attention.in_proj_weight", + "memory.scene_memory.bank_attention.in_proj_bias", + "memory.scene_memory.bank_attention.out_proj.weight", + "memory.scene_memory.bank_attention.out_proj.bias", + "memory.scene_memory.action_proj.0.weight", + "memory.scene_memory.action_proj.0.bias", + "memory.scene_memory.action_proj.1.weight", + "memory.scene_memory.action_proj.1.bias", + "memory.scene_memory.write_gate.0.weight", + "memory.scene_memory.write_gate.0.bias", + "memory.scene_memory.write_gate.1.weight", + "memory.scene_memory.write_gate.1.bias", + "memory.scene_memory.write_gate.3.weight", + "memory.scene_memory.write_gate.3.bias", + "memory.scene_memory.token_proj.0.weight", + "memory.scene_memory.token_proj.0.bias", + "memory.scene_memory.token_proj.1.weight", + "memory.scene_memory.token_proj.1.bias", + "memory.belief_memory.position_embedding", + "memory.belief_memory.bank_queries", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.in_proj_weight", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.in_proj_bias", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.out_proj.weight", + "memory.belief_memory.sequence_encoder.layers.0.self_attn.out_proj.bias", + "memory.belief_memory.sequence_encoder.layers.0.linear1.weight", + "memory.belief_memory.sequence_encoder.layers.0.linear1.bias", + "memory.belief_memory.sequence_encoder.layers.0.linear2.weight", + "memory.belief_memory.sequence_encoder.layers.0.linear2.bias", + "memory.belief_memory.sequence_encoder.layers.0.norm1.weight", + "memory.belief_memory.sequence_encoder.layers.0.norm1.bias", + "memory.belief_memory.sequence_encoder.layers.0.norm2.weight", + "memory.belief_memory.sequence_encoder.layers.0.norm2.bias", + "memory.belief_memory.bank_attention.in_proj_weight", + "memory.belief_memory.bank_attention.in_proj_bias", + "memory.belief_memory.bank_attention.out_proj.weight", + "memory.belief_memory.bank_attention.out_proj.bias", + "memory.belief_memory.action_proj.0.weight", + "memory.belief_memory.action_proj.0.bias", + "memory.belief_memory.action_proj.1.weight", + "memory.belief_memory.action_proj.1.bias", + "memory.belief_memory.write_gate.0.weight", + "memory.belief_memory.write_gate.0.bias", + "memory.belief_memory.write_gate.1.weight", + "memory.belief_memory.write_gate.1.bias", + "memory.belief_memory.write_gate.3.weight", + "memory.belief_memory.write_gate.3.bias", + "memory.belief_memory.token_proj.0.weight", + "memory.belief_memory.token_proj.0.bias", + "memory.belief_memory.token_proj.1.weight", + "memory.belief_memory.token_proj.1.bias", + "decoder.arm_decoder.layers.0.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.0.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.0.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.0.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.0.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.0.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.0.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.0.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.0.linear1.weight", + "decoder.arm_decoder.layers.0.linear1.bias", + "decoder.arm_decoder.layers.0.linear2.weight", + "decoder.arm_decoder.layers.0.linear2.bias", + "decoder.arm_decoder.layers.0.norm1.weight", + "decoder.arm_decoder.layers.0.norm1.bias", + "decoder.arm_decoder.layers.0.norm2.weight", + "decoder.arm_decoder.layers.0.norm2.bias", + "decoder.arm_decoder.layers.0.norm3.weight", + "decoder.arm_decoder.layers.0.norm3.bias", + "decoder.arm_decoder.layers.1.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.1.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.1.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.1.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.1.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.1.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.1.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.1.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.1.linear1.weight", + "decoder.arm_decoder.layers.1.linear1.bias", + "decoder.arm_decoder.layers.1.linear2.weight", + "decoder.arm_decoder.layers.1.linear2.bias", + "decoder.arm_decoder.layers.1.norm1.weight", + "decoder.arm_decoder.layers.1.norm1.bias", + "decoder.arm_decoder.layers.1.norm2.weight", + "decoder.arm_decoder.layers.1.norm2.bias", + "decoder.arm_decoder.layers.1.norm3.weight", + "decoder.arm_decoder.layers.1.norm3.bias", + "decoder.arm_decoder.layers.2.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.2.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.2.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.2.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.2.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.2.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.2.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.2.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.2.linear1.weight", + "decoder.arm_decoder.layers.2.linear1.bias", + "decoder.arm_decoder.layers.2.linear2.weight", + "decoder.arm_decoder.layers.2.linear2.bias", + "decoder.arm_decoder.layers.2.norm1.weight", + "decoder.arm_decoder.layers.2.norm1.bias", + "decoder.arm_decoder.layers.2.norm2.weight", + "decoder.arm_decoder.layers.2.norm2.bias", + "decoder.arm_decoder.layers.2.norm3.weight", + "decoder.arm_decoder.layers.2.norm3.bias", + "decoder.arm_decoder.layers.3.self_attn.in_proj_weight", + "decoder.arm_decoder.layers.3.self_attn.in_proj_bias", + "decoder.arm_decoder.layers.3.self_attn.out_proj.weight", + "decoder.arm_decoder.layers.3.self_attn.out_proj.bias", + "decoder.arm_decoder.layers.3.multihead_attn.in_proj_weight", + "decoder.arm_decoder.layers.3.multihead_attn.in_proj_bias", + "decoder.arm_decoder.layers.3.multihead_attn.out_proj.weight", + "decoder.arm_decoder.layers.3.multihead_attn.out_proj.bias", + "decoder.arm_decoder.layers.3.linear1.weight", + "decoder.arm_decoder.layers.3.linear1.bias", + "decoder.arm_decoder.layers.3.linear2.weight", + "decoder.arm_decoder.layers.3.linear2.bias", + "decoder.arm_decoder.layers.3.norm1.weight", + "decoder.arm_decoder.layers.3.norm1.bias", + "decoder.arm_decoder.layers.3.norm2.weight", + "decoder.arm_decoder.layers.3.norm2.bias", + "decoder.arm_decoder.layers.3.norm3.weight", + "decoder.arm_decoder.layers.3.norm3.bias", + "decoder.arm_identity.weight", + "decoder.phase_adapter.weight", + "decoder.phase_adapter.bias", + "decoder.role_adapter.weight", + "decoder.role_adapter.bias", + "decoder.context_proj.0.weight", + "decoder.context_proj.0.bias", + "decoder.context_proj.1.weight", + "decoder.context_proj.1.bias", + "decoder.arm_head.0.weight", + "decoder.arm_head.0.bias", + "decoder.arm_head.1.weight", + "decoder.arm_head.1.bias", + "decoder.arm_mean.weight", + "decoder.arm_mean.bias", + "decoder.arm_log_std.weight", + "decoder.arm_log_std.bias", + "decoder.proposal_mode_head.0.weight", + "decoder.proposal_mode_head.0.bias", + "decoder.proposal_mode_head.1.weight", + "decoder.proposal_mode_head.1.bias", + "decoder.proposal_mode_head.3.weight", + "decoder.proposal_mode_head.3.bias", + "decoder.proposal_mode_embeddings.weight", + "decoder.proposal_slot_embeddings.weight", + "decoder.mode_residual_heads.0.0.weight", + "decoder.mode_residual_heads.0.0.bias", + "decoder.mode_residual_heads.0.1.weight", + "decoder.mode_residual_heads.0.1.bias", + "decoder.mode_residual_heads.0.3.weight", + "decoder.mode_residual_heads.0.3.bias", + "decoder.mode_residual_heads.1.0.weight", + "decoder.mode_residual_heads.1.0.bias", + "decoder.mode_residual_heads.1.1.weight", + "decoder.mode_residual_heads.1.1.bias", + "decoder.mode_residual_heads.1.3.weight", + "decoder.mode_residual_heads.1.3.bias", + "decoder.mode_residual_heads.2.0.weight", + "decoder.mode_residual_heads.2.0.bias", + "decoder.mode_residual_heads.2.1.weight", + "decoder.mode_residual_heads.2.1.bias", + "decoder.mode_residual_heads.2.3.weight", + "decoder.mode_residual_heads.2.3.bias", + "decoder.mode_residual_heads.3.0.weight", + "decoder.mode_residual_heads.3.0.bias", + "decoder.mode_residual_heads.3.1.weight", + "decoder.mode_residual_heads.3.1.bias", + "decoder.mode_residual_heads.3.3.weight", + "decoder.mode_residual_heads.3.3.bias", + "decoder.mode_residual_heads.4.0.weight", + "decoder.mode_residual_heads.4.0.bias", + "decoder.mode_residual_heads.4.1.weight", + "decoder.mode_residual_heads.4.1.bias", + "decoder.mode_residual_heads.4.3.weight", + "decoder.mode_residual_heads.4.3.bias", + "decoder.mode_residual_heads.5.0.weight", + "decoder.mode_residual_heads.5.0.bias", + "decoder.mode_residual_heads.5.1.weight", + "decoder.mode_residual_heads.5.1.bias", + "decoder.mode_residual_heads.5.3.weight", + "decoder.mode_residual_heads.5.3.bias", + "decoder.slot_delta.0.weight", + "decoder.slot_delta.0.bias", + "decoder.slot_delta.1.weight", + "decoder.slot_delta.1.bias", + "decoder.slot_delta.3.weight", + "decoder.slot_delta.3.bias", + "decoder.proposal_score.0.weight", + "decoder.proposal_score.0.bias", + "decoder.proposal_score.1.weight", + "decoder.proposal_score.1.bias", + "decoder.proposal_score.3.weight", + "decoder.proposal_score.3.bias", + "elastic_state_head.interaction_queries", + "elastic_state_head.interaction_attention.in_proj_weight", + "elastic_state_head.interaction_attention.in_proj_bias", + "elastic_state_head.interaction_attention.out_proj.weight", + "elastic_state_head.interaction_attention.out_proj.bias", + "elastic_state_head.interaction_mlp.0.weight", + "elastic_state_head.interaction_mlp.0.bias", + "elastic_state_head.interaction_mlp.1.weight", + "elastic_state_head.interaction_mlp.1.bias", + "elastic_state_head.interaction_mlp.3.weight", + "elastic_state_head.interaction_mlp.3.bias", + "elastic_state_head.decoder.field_queries", + "elastic_state_head.decoder.field_attention.in_proj_weight", + "elastic_state_head.decoder.field_attention.in_proj_bias", + "elastic_state_head.decoder.field_attention.out_proj.weight", + "elastic_state_head.decoder.field_attention.out_proj.bias", + "elastic_state_head.decoder.field_mlp.0.weight", + "elastic_state_head.decoder.field_mlp.0.bias", + "elastic_state_head.decoder.field_mlp.1.weight", + "elastic_state_head.decoder.field_mlp.1.bias", + "elastic_state_head.decoder.field_mlp.3.weight", + "elastic_state_head.decoder.field_mlp.3.bias", + "elastic_state_head.decoder.summary_proj.0.weight", + "elastic_state_head.decoder.summary_proj.0.bias", + "elastic_state_head.decoder.summary_proj.1.weight", + "elastic_state_head.decoder.summary_proj.1.bias", + "elastic_state_head.decoder.phase_head.0.weight", + "elastic_state_head.decoder.phase_head.0.bias", + "elastic_state_head.decoder.phase_head.1.weight", + "elastic_state_head.decoder.phase_head.1.bias", + "elastic_state_head.decoder.phase_head.3.weight", + "elastic_state_head.decoder.phase_head.3.bias", + "elastic_state_head.decoder.arm_role_head.0.weight", + "elastic_state_head.decoder.arm_role_head.0.bias", + "elastic_state_head.decoder.arm_role_head.1.weight", + "elastic_state_head.decoder.arm_role_head.1.bias", + "elastic_state_head.decoder.arm_role_head.3.weight", + "elastic_state_head.decoder.arm_role_head.3.bias", + "elastic_state_head.decoder.arm_identity.weight", + "elastic_state_head.decoder.support_mode.0.weight", + "elastic_state_head.decoder.support_mode.0.bias", + "elastic_state_head.decoder.support_mode.1.weight", + "elastic_state_head.decoder.support_mode.1.bias", + "elastic_state_head.decoder.support_mode.3.weight", + "elastic_state_head.decoder.support_mode.3.bias", + "elastic_state_head.decoder.access_field.weight", + "elastic_state_head.decoder.access_field.bias", + "elastic_state_head.decoder.target_belief_field.weight", + "elastic_state_head.decoder.target_belief_field.bias", + "elastic_state_head.decoder.visibility_field.weight", + "elastic_state_head.decoder.visibility_field.bias", + "elastic_state_head.decoder.clearance_field.weight", + "elastic_state_head.decoder.clearance_field.bias", + "elastic_state_head.decoder.occluder_contact_field.weight", + "elastic_state_head.decoder.occluder_contact_field.bias", + "elastic_state_head.decoder.grasp_affordance_field.weight", + "elastic_state_head.decoder.grasp_affordance_field.bias", + "elastic_state_head.decoder.support_stability_field.weight", + "elastic_state_head.decoder.support_stability_field.bias", + "elastic_state_head.decoder.persistence_field.weight", + "elastic_state_head.decoder.persistence_field.bias", + "elastic_state_head.decoder.reocclusion_field.weight", + "elastic_state_head.decoder.reocclusion_field.bias", + "elastic_state_head.decoder.disturbance_field.weight", + "elastic_state_head.decoder.disturbance_field.bias", + "elastic_state_head.decoder.uncertainty_field.weight", + "elastic_state_head.decoder.uncertainty_field.bias", + "elastic_state_head.decoder.reocclusion_head.0.weight", + "elastic_state_head.decoder.reocclusion_head.0.bias", + "elastic_state_head.decoder.reocclusion_head.1.weight", + "elastic_state_head.decoder.reocclusion_head.1.bias", + "elastic_state_head.decoder.reocclusion_head.3.weight", + "elastic_state_head.decoder.reocclusion_head.3.bias", + "world_model.state_encoder.0.weight", + "world_model.state_encoder.0.bias", + "world_model.state_encoder.1.weight", + "world_model.state_encoder.1.bias", + "world_model.scene_memory_proj.0.weight", + "world_model.scene_memory_proj.0.bias", + "world_model.scene_memory_proj.1.weight", + "world_model.scene_memory_proj.1.bias", + "world_model.belief_memory_proj.0.weight", + "world_model.belief_memory_proj.0.bias", + "world_model.belief_memory_proj.1.weight", + "world_model.belief_memory_proj.1.bias", + "world_model.action_encoder.0.weight", + "world_model.action_encoder.0.bias", + "world_model.action_encoder.1.weight", + "world_model.action_encoder.1.bias", + "world_model.transition.weight_ih", + "world_model.transition.weight_hh", + "world_model.transition.bias_ih", + "world_model.transition.bias_hh", + "world_model.scene_memory_update.weight", + "world_model.scene_memory_update.bias", + "world_model.belief_memory_update.weight", + "world_model.belief_memory_update.bias", + "world_model.compact_decoder.weight", + "world_model.compact_decoder.bias", + "world_model.target_belief_head.weight", + "world_model.target_belief_head.bias", + "world_model.visibility_head.weight", + "world_model.visibility_head.bias", + "world_model.clearance_head.weight", + "world_model.clearance_head.bias", + "world_model.occluder_contact_head.weight", + "world_model.occluder_contact_head.bias", + "world_model.grasp_affordance_head.weight", + "world_model.grasp_affordance_head.bias", + "world_model.support_stability_head.weight", + "world_model.support_stability_head.bias", + "world_model.persistence_head.weight", + "world_model.persistence_head.bias", + "world_model.reocclusion_head.weight", + "world_model.reocclusion_head.bias", + "world_model.disturbance_head.weight", + "world_model.disturbance_head.bias", + "world_model.uncertainty_head.weight", + "world_model.uncertainty_head.bias", + "world_model.access_head.weight", + "world_model.access_head.bias", + "planner.residual.trunk.0.weight", + "planner.residual.trunk.0.bias", + "planner.residual.trunk.1.weight", + "planner.residual.trunk.1.bias", + "planner.residual.trunk.3.weight", + "planner.residual.trunk.3.bias", + "planner.residual.success_head.weight", + "planner.residual.success_head.bias", + "planner.residual.risk_head.weight", + "planner.residual.risk_head.bias", + "planner.residual.residual_head.weight", + "planner.residual.residual_head.bias" + ], + "unexpected_keys": [] + } +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/rlbench_open_drawer_r3d_rollout/rollout_eval.json b/artifacts/outputs/r3d/rlbench_open_drawer_r3d_rollout/rollout_eval.json new file mode 100644 index 0000000000000000000000000000000000000000..7588f89495345effea836af44a342b0e8faaa4cc --- /dev/null +++ b/artifacts/outputs/r3d/rlbench_open_drawer_r3d_rollout/rollout_eval.json @@ -0,0 +1,35 @@ +{ + "checkpoint": "/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt", + "plan_requested": true, + "plan_applied": true, + "planner_mode": "trainable", + "support_mode_conditioning": true, + "episodes_per_task": 1, + "episode_length": 5, + "resolution": 224, + "reset_retries": 20, + "cameras": [ + "front", + "wrist_left", + "wrist_right" + ], + "tasks": { + "open_drawer": { + "task_class": "RightOpenDrawer", + "successes": [ + 0.0 + ], + "returns": [ + 0.0 + ], + "path_recoveries": 0, + "noop_fallbacks": 0, + "reset_retries": [ + 0 + ], + "mean_success": 0.0, + "mean_return": 0.0 + } + }, + "mean_success": 0.0 +} \ No newline at end of file diff --git a/artifacts/outputs/r3d/rlbench_open_drawer_r3d_rollout/rollout_eval.md b/artifacts/outputs/r3d/rlbench_open_drawer_r3d_rollout/rollout_eval.md new file mode 100644 index 0000000000000000000000000000000000000000..5c79b18d24b5833a1bd42c0730ed6140a9eaff58 --- /dev/null +++ b/artifacts/outputs/r3d/rlbench_open_drawer_r3d_rollout/rollout_eval.md @@ -0,0 +1,11 @@ +# RLBench Rollout Eval + +- Checkpoint: `/workspace/VLAarchtests/artifacts/outputs/r3d/proxy_interaction_r3d_stage2_clip_seed13/checkpoint_best.pt` +- Plan requested: `True` +- Plan applied: `True` +- Support-mode conditioning: `True` +- Mean success: `0.000` + +## Per-task + +- `open_drawer`: mean_success=0.000, returns=[0.0] diff --git a/artifacts/outputs/r3d/rlbench_smokes/launch_smoke_open_drawer.stderr b/artifacts/outputs/r3d/rlbench_smokes/launch_smoke_open_drawer.stderr new file mode 100644 index 0000000000000000000000000000000000000000..925d7cccb45fcf2ba5aa46463026025532438b3b --- /dev/null +++ b/artifacts/outputs/r3d/rlbench_smokes/launch_smoke_open_drawer.stderr @@ -0,0 +1,5 @@ +error libmamba Could not set lock (Resource temporarily unavailable) +warning libmamba Cannot lock '/workspace/.cache/mamba/proc' + Waiting for other mamba process to finish +qt.qpa.xcb: QXcbConnection: XCB error: 148 (Unknown), sequence: 181, resource id: 0, major code: 140 (Unknown), minor code: 20 +WARNING:root:not sure how _robot_shapes are used is used. diff --git a/artifacts/outputs/r3d/rlbench_smokes/launch_smoke_open_drawer.txt b/artifacts/outputs/r3d/rlbench_smokes/launch_smoke_open_drawer.txt new file mode 100644 index 0000000000000000000000000000000000000000..8d02530bfa93848ca166a97c2fe029e28fee3e9e --- /dev/null +++ b/artifacts/outputs/r3d/rlbench_smokes/launch_smoke_open_drawer.txt @@ -0,0 +1,38 @@ +{ + "display": ":99", + "headless": true, + "task": "RightOpenDrawer", + "description": "open the bottom drawer", + "front_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_left_rgb_shape": [ + 224, + 224, + 3 + ], + "wrist_right_rgb_shape": [ + 224, + 224, + 3 + ], + "right_pose_shape": [ + 7 + ], + "left_pose_shape": [ + 7 + ], + "stepped_mode": "bimanual_noop", + "action_finite": true, + "action_dim": 18, + "reward": 0.0, + "done": false, + "front_rgb_shape_after_step": [ + 224, + 224, + 3 + ] +} +[CoppeliaSim:loadinfo] done. diff --git a/artifacts/outputs/r3d/rlbench_smokes/smoke_test_output.txt b/artifacts/outputs/r3d/rlbench_smokes/smoke_test_output.txt new file mode 100644 index 0000000000000000000000000000000000000000..b378e7ade721587456f1f1e56ef4e18e8e730530 --- /dev/null +++ b/artifacts/outputs/r3d/rlbench_smokes/smoke_test_output.txt @@ -0,0 +1,19 @@ +{ + "camera_names": [ + "front", + "wrist_left", + "wrist_right" + ], + "resolution": [ + 224, + 224 + ], + "global_camera": "front", + "resolved_upstream_root": "/workspace/third_party/peract2_bimanual_probe_20260324", + "imports": { + "rlbench": "ok", + "pyrep": "ok", + "yarr": "ok" + } +} +xvfb-run -a -s '-screen 0 1280x1024x24' /workspace/envs/rlbench/bin/python train.py method=PERACT_BC rlbench.task_name=multi_3cam 'rlbench.tasks=[bimanual_push_box,bimanual_lift_ball,bimanual_dual_push_buttons,bimanual_pick_plate,bimanual_put_item_in_drawer,bimanual_put_bottle_in_fridge,bimanual_handover_item,bimanual_pick_laptop,bimanual_straighten_rope,bimanual_sweep_to_dustpan,bimanual_lift_tray,bimanual_handover_item_easy,bimanual_take_tray_out_of_oven]' rlbench.demos=100 rlbench.demo_path=/workspace/data/rlbench2 replay.path=/workspace/replays/rlbench2 framework.logdir=/workspace/logs/rlbench2 framework.training_iterations=40000 framework.gpu=0 framework.env_gpu=0 framework.start_seed=0 ddp.num_devices=1 'rlbench.cameras=[front,wrist_left,wrist_right]' 'rlbench.camera_resolution=[224,224]' diff --git a/code/reveal_vla_bimanual/eval/__pycache__/__init__.cpython-311.pyc b/code/reveal_vla_bimanual/eval/__pycache__/__init__.cpython-311.pyc index 3688005347278003b796dc68d5935453ef39556a..ca783519390437f6bd17d8118b203e01f3f2b052 100644 Binary files a/code/reveal_vla_bimanual/eval/__pycache__/__init__.cpython-311.pyc and b/code/reveal_vla_bimanual/eval/__pycache__/__init__.cpython-311.pyc differ diff --git a/code/reveal_vla_bimanual/eval/__pycache__/metrics.cpython-311.pyc b/code/reveal_vla_bimanual/eval/__pycache__/metrics.cpython-311.pyc index 85622329991afadac40a90d707d5559565e358b1..d6a94fef601f102a01843bc7161fcc97e7878873 100644 Binary files a/code/reveal_vla_bimanual/eval/__pycache__/metrics.cpython-311.pyc and b/code/reveal_vla_bimanual/eval/__pycache__/metrics.cpython-311.pyc differ diff --git a/code/reveal_vla_bimanual/eval/__pycache__/report.cpython-311.pyc b/code/reveal_vla_bimanual/eval/__pycache__/report.cpython-311.pyc index 758526feaf9c2e1803f5e1738887b6525dfee93d..9b0d32bc34fc756daaf6f83cd31049e428cbf586 100644 Binary files a/code/reveal_vla_bimanual/eval/__pycache__/report.cpython-311.pyc and b/code/reveal_vla_bimanual/eval/__pycache__/report.cpython-311.pyc differ diff --git a/code/reveal_vla_bimanual/eval/__pycache__/run_proxy_diagnostics.cpython-311.pyc b/code/reveal_vla_bimanual/eval/__pycache__/run_proxy_diagnostics.cpython-311.pyc new file mode 100644 index 0000000000000000000000000000000000000000..6db90982dc83da88fb1445353eaa2e29b3b8b96d Binary files /dev/null and b/code/reveal_vla_bimanual/eval/__pycache__/run_proxy_diagnostics.cpython-311.pyc differ diff --git a/code/reveal_vla_bimanual/eval/__pycache__/run_reveal_benchmark.cpython-311.pyc b/code/reveal_vla_bimanual/eval/__pycache__/run_reveal_benchmark.cpython-311.pyc index e9ed64fb9a209d2861c3e62bcbd208fb4436f94e..018668aa99a6f9e6ab33b5a1dedf03d3a42e3dd2 100644 Binary files a/code/reveal_vla_bimanual/eval/__pycache__/run_reveal_benchmark.cpython-311.pyc and b/code/reveal_vla_bimanual/eval/__pycache__/run_reveal_benchmark.cpython-311.pyc differ