Upload PI0.5 UR7e ArrangeBlock fine-tuned policy at step 5550

README.md

@@ -0,0 +1,141 @@
+---
+library_name: lerobot
+pipeline_tag: robotics
+base_model: lerobot/pi05_base
+base_model_relation: finetune
+datasets:
+- CoRL2026-CSI/UR7e-CaP_arrange_block_100epi_10fps
+license: other
+tags:
+- lerobot
+- pi05
+- pi0.5
+- openpi
+- vision-language-action
+- imitation-learning
+- robotics
+- ur7e
+- arrange-block
+- 10fps
+- safetensors
+---
+
+# CoRL2026-CSI/Pi0.5-UR7e-ArrangeBlock_30epoch
+
+This is a LeRobot PI0.5 policy fine-tuned from
+[`lerobot/pi05_base`](https://huggingface.co/lerobot/pi05_base) on
+[`CoRL2026-CSI/UR7e-CaP_arrange_block_100epi_10fps`](https://huggingface.co/datasets/CoRL2026-CSI/UR7e-CaP_arrange_block_100epi_10fps).
+
+The model is intended for UR7e ArrangeBlock manipulation experiments using RGB
+observations, robot proprioception, language instructions, and continuous action chunks.
+It is uploaded as a LeRobot policy checkpoint and should be loaded through the matching
+LeRobot PI0.5 implementation used for training.
+
+## Model Details
+
+- **Policy type:** PI0.5
+- **Base policy:** `lerobot/pi05_base`
+- **PaliGemma variant:** `gemma_2b`
+- **Action expert variant:** `gemma_300m`
+- **Action chunk size:** `16`
+- **Action steps:** `16`
+- **Max state/action dims:** `32` / `32`
+- **Vision encoder frozen:** `false`
+- **Train expert only:** `false`
+- **Gradient checkpointing:** `true`
+- **Training dtype:** `bfloat16`
+
+## Fine-Tuning Setup
+
+- **Dataset:** `CoRL2026-CSI/UR7e-CaP_arrange_block_100epi_10fps`
+- **Training steps:** `5550`
+- **Approx. epochs:** `30.16`
+- **Final training samples:** `1420800`
+- **Final training loss:** `0.009152`
+- **Runtime:** `19.12 hours`
+- **Per-GPU batch size:** `64`
+- **Gradient accumulation steps:** `2`
+- **Number of GPUs:** `2`
+- **Effective batch size:** `256`
+- **Optimizer lr:** `2.5e-05`
+- **Optimizer betas:** `[0.9, 0.95]`
+- **Weight decay:** `0.01`
+- **Scheduler warmup/decay:** `1000` / `30000`
+- **Final decay lr:** `2.5e-06`
+- **DataLoader workers:** `8`
+- **DataLoader prefetch factor:** `1`
+
+## Camera Mapping
+
+- no explicit rename map
+
+## Image Augmentation
+
+- disabled
+
+## Inputs
+
+- `observation.images.base_0_rgb`: `STATE`, shape `[1]`
+- `observation.images.left_wrist_0_rgb`: `STATE`, shape `[1]`
+- `observation.images.realsense_topview`: `VISUAL`, shape `[3, 480, 640]`
+- `observation.images.realsense_wrist`: `VISUAL`, shape `[3, 480, 640]`
+- `observation.images.right_wrist_0_rgb`: `STATE`, shape `[1]`
+- `observation.state`: `STATE`, shape `[7]`
+
+## Outputs
+
+- `action`: `ACTION`, shape `[7]`
+
+## Usage
+
+Install and use the same LeRobot checkout/environment that contains the PI0.5 policy
+implementation, then point `policy.path` to this Hub repo.
+
+```bash
+lerobot-record \
+  --robot.type=<your_robot> \
+  --dataset.repo_id=<your_eval_dataset_repo> \
+  --policy.path=CoRL2026-CSI/Pi0.5-UR7e-ArrangeBlock_30epoch \
+  --episodes=10
+```
+
+For local Python usage, load the policy with LeRobot's policy factory from the training
+checkout.
+
+## Evaluation
+
+This upload records the offline training run metrics only. No rollout success rate is
+claimed here unless a separate real or simulated evaluation is added later.
+
+Final logged training metrics:
+
+- loss: `0.009152`
+- grad norm: `0.377971`
+- learning rate: `2.500051366567086e-06`
+- update time: `6.0738 s/step`
+- dataloading time: `0.0210 s/step`
+
+## Limitations and Safety
+
+This model is a robot control policy and can produce unsafe actions if deployed on
+hardware without appropriate validation, workspace limits, emergency stop handling, and
+task-specific safety checks. Test in simulation or a constrained setup before any
+physical deployment.
+
+The model is specialized to the training dataset, camera mapping, calibration, action
+space, and embodiment configuration. It may not transfer reliably to different robots,
+camera placements, object layouts, or tasks without further validation or fine-tuning.
+
+## License and Terms
+
+The training dataset is marked `apache-2.0`. This fine-tuned model is conservatively
+marked as `other`; users are responsible for checking the applicable base model,
+dataset, and deployment terms before use.
+
+## Files
+
+- `model.safetensors`: fine-tuned policy weights
+- `config.json`: LeRobot PI0.5 policy config
+- `train_config.json`: training configuration
+- `policy_preprocessor.json` and `policy_postprocessor.json`: LeRobot processor pipelines
+- `policy_*_processor.safetensors`: normalization/statistics state used by processors

config.json

@@ -0,0 +1,109 @@
+{
+    "type": "pi05",
+    "n_obs_steps": 1,
+    "input_features": {
+        "observation.images.base_0_rgb": {
+            "type": "STATE",
+            "shape": [
+                1
+            ]
+        },
+        "observation.images.left_wrist_0_rgb": {
+            "type": "STATE",
+            "shape": [
+                1
+            ]
+        },
+        "observation.images.right_wrist_0_rgb": {
+            "type": "STATE",
+            "shape": [
+                1
+            ]
+        },
+        "observation.state": {
+            "type": "STATE",
+            "shape": [
+                7
+            ]
+        },
+        "observation.images.realsense_topview": {
+            "type": "VISUAL",
+            "shape": [
+                3,
+                480,
+                640
+            ]
+        },
+        "observation.images.realsense_wrist": {
+            "type": "VISUAL",
+            "shape": [
+                3,
+                480,
+                640
+            ]
+        }
+    },
+    "output_features": {
+        "action": {
+            "type": "ACTION",
+            "shape": [
+                7
+            ]
+        }
+    },
+    "device": "cuda",
+    "use_amp": false,
+    "use_peft": false,
+    "push_to_hub": false,
+    "repo_id": null,
+    "private": null,
+    "tags": null,
+    "license": null,
+    "pretrained_path": "/home/work/hscho/corl_2026/AutoDataCollector/lerobot/outputs/train/pi05_ur7e_arrange_block_100epi_10fps_fft_bs64_acc2_ep30_20260512_220935/checkpoints/last/pretrained_model",
+    "paligemma_variant": "gemma_2b",
+    "action_expert_variant": "gemma_300m",
+    "dtype": "bfloat16",
+    "chunk_size": 16,
+    "n_action_steps": 16,
+    "max_state_dim": 32,
+    "max_action_dim": 32,
+    "num_inference_steps": 10,
+    "time_sampling_beta_alpha": 1.5,
+    "time_sampling_beta_beta": 1.0,
+    "time_sampling_scale": 0.999,
+    "time_sampling_offset": 0.001,
+    "min_period": 0.004,
+    "max_period": 4.0,
+    "rtc_config": null,
+    "use_relative_actions": false,
+    "relative_exclude_joints": [],
+    "action_feature_names": [],
+    "image_resolution": [
+        224,
+        224
+    ],
+    "empty_cameras": 0,
+    "tokenizer_max_length": 200,
+    "normalization_mapping": {
+        "VISUAL": "IDENTITY",
+        "STATE": "QUANTILES",
+        "ACTION": "QUANTILES"
+    },
+    "gradient_checkpointing": true,
+    "compile_model": false,
+    "compile_mode": "default",
+    "profile_forward_timing": false,
+    "freeze_vision_encoder": false,
+    "train_expert_only": false,
+    "optimizer_lr": 2.5e-05,
+    "optimizer_betas": [
+        0.9,
+        0.95
+    ],
+    "optimizer_eps": 1e-08,
+    "optimizer_weight_decay": 0.01,
+    "optimizer_grad_clip_norm": 1.0,
+    "scheduler_warmup_steps": 1000,
+    "scheduler_decay_steps": 30000,
+    "scheduler_decay_lr": 2.5e-06
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e790ab59cbe5ab60b91f69b3c818616aed6352df9373428615462127060e6fd
+size 9354045072

policy_postprocessor.json

@@ -0,0 +1,32 @@
+{
+  "name": "policy_postprocessor",
+  "steps": [
+    {
+      "registry_name": "unnormalizer_processor",
+      "config": {
+        "eps": 1e-08,
+        "features": {
+          "action": {
+            "type": "ACTION",
+            "shape": [
+              7
+            ]
+          }
+        },
+        "norm_map": {
+          "VISUAL": "IDENTITY",
+          "STATE": "QUANTILES",
+          "ACTION": "QUANTILES"
+        }
+      },
+      "state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
+    },
+    {
+      "registry_name": "device_processor",
+      "config": {
+        "device": "cpu",
+        "float_dtype": null
+      }
+    }
+  ]
+}

policy_postprocessor_step_0_unnormalizer_processor.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a76a02d1d02b4a113d87d519281472b2f79fe4bc82ed6067a3aa9598f8bf5d5
+size 15124

policy_preprocessor.json

@@ -0,0 +1,97 @@
+{
+  "name": "policy_preprocessor",
+  "steps": [
+    {
+      "registry_name": "rename_observations_processor",
+      "config": {
+        "rename_map": {}
+      }
+    },
+    {
+      "registry_name": "to_batch_processor",
+      "config": {}
+    },
+    {
+      "registry_name": "normalizer_processor",
+      "config": {
+        "eps": 1e-08,
+        "features": {
+          "observation.images.base_0_rgb": {
+            "type": "STATE",
+            "shape": [
+              1
+            ]
+          },
+          "observation.images.left_wrist_0_rgb": {
+            "type": "STATE",
+            "shape": [
+              1
+            ]
+          },
+          "observation.images.right_wrist_0_rgb": {
+            "type": "STATE",
+            "shape": [
+              1
+            ]
+          },
+          "observation.state": {
+            "type": "STATE",
+            "shape": [
+              7
+            ]
+          },
+          "observation.images.realsense_topview": {
+            "type": "VISUAL",
+            "shape": [
+              3,
+              480,
+              640
+            ]
+          },
+          "observation.images.realsense_wrist": {
+            "type": "VISUAL",
+            "shape": [
+              3,
+              480,
+              640
+            ]
+          },
+          "action": {
+            "type": "ACTION",
+            "shape": [
+              7
+            ]
+          }
+        },
+        "norm_map": {
+          "VISUAL": "IDENTITY",
+          "STATE": "QUANTILES",
+          "ACTION": "QUANTILES"
+        }
+      },
+      "state_file": "policy_preprocessor_step_2_normalizer_processor.safetensors"
+    },
+    {
+      "registry_name": "pi05_prepare_state_tokenizer_processor_step",
+      "config": {}
+    },
+    {
+      "registry_name": "tokenizer_processor",
+      "config": {
+        "max_length": 200,
+        "task_key": "task",
+        "padding_side": "right",
+        "padding": "max_length",
+        "truncation": true,
+        "tokenizer_name": "google/paligemma-3b-pt-224"
+      }
+    },
+    {
+      "registry_name": "device_processor",
+      "config": {
+        "device": "cuda",
+        "float_dtype": null
+      }
+    }
+  ]
+}

policy_preprocessor_step_2_normalizer_processor.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a76a02d1d02b4a113d87d519281472b2f79fe4bc82ed6067a3aa9598f8bf5d5
+size 15124

train_config.json

@@ -0,0 +1,254 @@
+{
+    "dataset": {
+        "repo_id": "CoRL2026-CSI/UR7e-CaP_arrange_block_100epi_10fps",
+        "root": null,
+        "episodes": null,
+        "image_transforms": {
+            "enable": false,
+            "max_num_transforms": 3,
+            "random_order": false,
+            "tfs": {
+                "brightness": {
+                    "weight": 1.0,
+                    "type": "ColorJitter",
+                    "kwargs": {
+                        "brightness": [
+                            0.8,
+                            1.2
+                        ]
+                    }
+                },
+                "contrast": {
+                    "weight": 1.0,
+                    "type": "ColorJitter",
+                    "kwargs": {
+                        "contrast": [
+                            0.8,
+                            1.2
+                        ]
+                    }
+                },
+                "saturation": {
+                    "weight": 1.0,
+                    "type": "ColorJitter",
+                    "kwargs": {
+                        "saturation": [
+                            0.5,
+                            1.5
+                        ]
+                    }
+                },
+                "hue": {
+                    "weight": 1.0,
+                    "type": "ColorJitter",
+                    "kwargs": {
+                        "hue": [
+                            -0.05,
+                            0.05
+                        ]
+                    }
+                },
+                "sharpness": {
+                    "weight": 1.0,
+                    "type": "SharpnessJitter",
+                    "kwargs": {
+                        "sharpness": [
+                            0.5,
+                            1.5
+                        ]
+                    }
+                },
+                "affine": {
+                    "weight": 1.0,
+                    "type": "RandomAffine",
+                    "kwargs": {
+                        "degrees": [
+                            -5.0,
+                            5.0
+                        ],
+                        "translate": [
+                            0.05,
+                            0.05
+                        ]
+                    }
+                }
+            }
+        },
+        "revision": null,
+        "use_imagenet_stats": true,
+        "video_backend": "torchcodec",
+        "streaming": false
+    },
+    "env": null,
+    "policy": {
+        "type": "pi05",
+        "n_obs_steps": 1,
+        "input_features": {
+            "observation.images.base_0_rgb": {
+                "type": "STATE",
+                "shape": [
+                    1
+                ]
+            },
+            "observation.images.left_wrist_0_rgb": {
+                "type": "STATE",
+                "shape": [
+                    1
+                ]
+            },
+            "observation.images.right_wrist_0_rgb": {
+                "type": "STATE",
+                "shape": [
+                    1
+                ]
+            },
+            "observation.state": {
+                "type": "STATE",
+                "shape": [
+                    7
+                ]
+            },
+            "observation.images.realsense_topview": {
+                "type": "VISUAL",
+                "shape": [
+                    3,
+                    480,
+                    640
+                ]
+            },
+            "observation.images.realsense_wrist": {
+                "type": "VISUAL",
+                "shape": [
+                    3,
+                    480,
+                    640
+                ]
+            }
+        },
+        "output_features": {
+            "action": {
+                "type": "ACTION",
+                "shape": [
+                    7
+                ]
+            }
+        },
+        "device": "cuda",
+        "use_amp": false,
+        "use_peft": false,
+        "push_to_hub": false,
+        "repo_id": null,
+        "private": null,
+        "tags": null,
+        "license": null,
+        "pretrained_path": "/home/work/hscho/corl_2026/AutoDataCollector/lerobot/outputs/train/pi05_ur7e_arrange_block_100epi_10fps_fft_bs64_acc2_ep30_20260512_220935/checkpoints/last/pretrained_model",
+        "paligemma_variant": "gemma_2b",
+        "action_expert_variant": "gemma_300m",
+        "dtype": "bfloat16",
+        "chunk_size": 16,
+        "n_action_steps": 16,
+        "max_state_dim": 32,
+        "max_action_dim": 32,
+        "num_inference_steps": 10,
+        "time_sampling_beta_alpha": 1.5,
+        "time_sampling_beta_beta": 1.0,
+        "time_sampling_scale": 0.999,
+        "time_sampling_offset": 0.001,
+        "min_period": 0.004,
+        "max_period": 4.0,
+        "rtc_config": null,
+        "use_relative_actions": false,
+        "relative_exclude_joints": [],
+        "action_feature_names": [],
+        "image_resolution": [
+            224,
+            224
+        ],
+        "empty_cameras": 0,
+        "tokenizer_max_length": 200,
+        "normalization_mapping": {
+            "VISUAL": "IDENTITY",
+            "STATE": "QUANTILES",
+            "ACTION": "QUANTILES"
+        },
+        "gradient_checkpointing": true,
+        "compile_model": false,
+        "compile_mode": "default",
+        "profile_forward_timing": false,
+        "freeze_vision_encoder": false,
+        "train_expert_only": false,
+        "optimizer_lr": 2.5e-05,
+        "optimizer_betas": [
+            0.9,
+            0.95
+        ],
+        "optimizer_eps": 1e-08,
+        "optimizer_weight_decay": 0.01,
+        "optimizer_grad_clip_norm": 1.0,
+        "scheduler_warmup_steps": 1000,
+        "scheduler_decay_steps": 30000,
+        "scheduler_decay_lr": 2.5e-06
+    },
+    "output_dir": "/home/work/hscho/corl_2026/AutoDataCollector/lerobot/outputs/train/pi05_ur7e_arrange_block_100epi_10fps_fft_bs64_acc2_ep30_20260512_220935",
+    "job_name": "pi05_ur7e_arrange_block_100epi_10fps_fft_bs64_acc2_ep30_20260512_220935",
+    "resume": true,
+    "seed": 1000,
+    "cudnn_deterministic": false,
+    "num_workers": 8,
+    "dataloader_prefetch_factor": 1,
+    "dataloader_persistent_workers": true,
+    "dataloader_pin_memory": true,
+    "batch_size": 64,
+    "gradient_accumulation_steps": 2,
+    "ddp_find_unused_parameters": true,
+    "profile_timing": false,
+    "preprocess_in_workers": true,
+    "groot_preprocess_in_workers": false,
+    "steps": 5550,
+    "eval_freq": 0,
+    "log_freq": 10,
+    "tolerance_s": 0.0001,
+    "save_checkpoint": true,
+    "save_freq": 2775,
+    "use_policy_training_preset": true,
+    "optimizer": {
+        "type": "adamw",
+        "lr": 2.5e-05,
+        "weight_decay": 0.01,
+        "grad_clip_norm": 1.0,
+        "betas": [
+            0.9,
+            0.95
+        ],
+        "eps": 1e-08
+    },
+    "scheduler": {
+        "type": "cosine_decay_with_warmup",
+        "num_warmup_steps": 1000,
+        "num_decay_steps": 30000,
+        "peak_lr": 2.5e-05,
+        "decay_lr": 2.5e-06
+    },
+    "eval": {
+        "n_episodes": 50,
+        "batch_size": 50,
+        "use_async_envs": false
+    },
+    "wandb": {
+        "enable": true,
+        "disable_artifact": true,
+        "project": "lerobot-pi05-ur7e",
+        "entity": null,
+        "notes": null,
+        "run_id": "hwx6slss",
+        "mode": null
+    },
+    "peft": null,
+    "use_rabc": false,
+    "rabc_progress_path": null,
+    "rabc_kappa": 0.01,
+    "rabc_epsilon": 1e-06,
+    "rabc_head_mode": "sparse",
+    "rename_map": {},
+    "checkpoint_path": "/home/work/hscho/corl_2026/AutoDataCollector/lerobot/outputs/train/pi05_ur7e_arrange_block_100epi_10fps_fft_bs64_acc2_ep30_20260512_220935/checkpoints/last"
+}