Upload ITPS Maze2D pretrained checkpoint

README.md

@@ -0,0 +1,56 @@
+---
+library_name: lerobot
+license: mit
+tags:
+  - lerobot
+  - dp
+  - robotics
+  - maze2d
+  - itps
+  - pytorch_model_hub_mixin
+pipeline_tag: robotics
+---
+
+# ITPS Maze2D — Diffusion Policy (DP)
+
+Pre-trained Diffusion Policy checkpoint used in
+**Inference-Time Policy Steering through Human Interactions**
+([paper](https://huggingface.co/papers/2411.16627), [project page](https://yanweiw.github.io/itps/), [code](https://github.com/yanweiw/itps)).
+
+The model was trained on the [D4RL Maze2D](https://github.com/Farama-Foundation/D4RL)
+dataset and is intended to be loaded with the
+[LeRobot](https://github.com/huggingface/lerobot) policy classes.
+
+## Usage
+
+Clone the inference repo, then load this checkpoint directly from the Hub:
+
+```bash
+git clone https://github.com/yanweiw/itps.git && cd itps
+pip install -e .
+python interact_maze2d.py -p dp --hf
+```
+
+Or load it programmatically:
+
+```python
+from itps.common.policies.diffusion.modeling_diffusion import DiffusionPolicy
+
+policy = DiffusionPolicy.from_pretrained("felixw/itps-dp")
+policy.eval()
+```
+
+## Citation
+
+```bibtex
+@article{wang2024itps,
+  title={Inference-Time Policy Steering through Human Interactions},
+  author={Wang, Yanwei and others},
+  journal={arXiv preprint arXiv:2411.16627},
+  year={2024}
+}
+```
+
+## License
+
+MIT — see [LICENSE](https://github.com/yanweiw/itps/blob/main/LICENSE).

config.json

@@ -0,0 +1,53 @@
+{
+  "beta_end": 0.02,
+  "beta_schedule": "squaredcos_cap_v2",
+  "beta_start": 0.0001,
+  "clip_sample": true,
+  "clip_sample_range": 1.0,
+  "crop_is_random": true,
+  "crop_shape": [
+    84,
+    84
+  ],
+  "diffusion_step_embed_dim": 128,
+  "do_mask_loss_for_padding": false,
+  "down_dims": [
+    128,
+    256,
+    512
+  ],
+  "horizon": 64,
+  "input_normalization_modes": {
+    "observation.environment_state": "min_max",
+    "observation.state": "min_max"
+  },
+  "input_shapes": {
+    "observation.environment_state": [
+      2
+    ],
+    "observation.state": [
+      2
+    ]
+  },
+  "kernel_size": 5,
+  "n_action_steps": 8,
+  "n_groups": 8,
+  "n_obs_steps": 2,
+  "noise_scheduler_type": "DDIM",
+  "num_inference_steps": 10,
+  "num_train_timesteps": 100,
+  "output_normalization_modes": {
+    "action": "min_max"
+  },
+  "output_shapes": {
+    "action": [
+      2
+    ]
+  },
+  "prediction_type": "epsilon",
+  "pretrained_backbone_weights": null,
+  "spatial_softmax_num_keypoints": 32,
+  "use_film_scale_modulation": true,
+  "use_group_norm": true,
+  "vision_backbone": "resnet18"
+}

config.yaml

@@ -0,0 +1,195 @@
+resume: false
+device: cuda:1
+use_amp: false
+seed: 100000
+dataset_repo_id: maze2d
+dataset_root: /afs/csail.mit.edu/u/f/felixw/lerobot/data/maze2d-large-sparse-v1.hdf5
+video_backend: pyav
+training:
+  offline_steps: 100000
+  num_workers: 4
+  batch_size: 256
+  eval_freq: 0
+  log_freq: 250
+  save_checkpoint: true
+  save_freq: 5000
+  online_steps: 0
+  online_rollout_n_episodes: 1
+  online_rollout_batch_size: 1
+  online_steps_between_rollouts: 1
+  online_sampling_ratio: 0.5
+  online_env_seed: null
+  online_buffer_capacity: null
+  online_buffer_seed_size: 0
+  do_online_rollout_async: false
+  image_transforms:
+    enable: false
+    max_num_transforms: 3
+    random_order: false
+    brightness:
+      weight: 1
+      min_max:
+      - 0.8
+      - 1.2
+    contrast:
+      weight: 1
+      min_max:
+      - 0.8
+      - 1.2
+    saturation:
+      weight: 1
+      min_max:
+      - 0.5
+      - 1.5
+    hue:
+      weight: 1
+      min_max:
+      - -0.05
+      - 0.05
+    sharpness:
+      weight: 1
+      min_max:
+      - 0.8
+      - 1.2
+  grad_clip_norm: 10
+  lr: 0.0001
+  lr_scheduler: cosine
+  lr_warmup_steps: 500
+  adam_betas:
+  - 0.95
+  - 0.999
+  adam_eps: 1.0e-08
+  adam_weight_decay: 1.0e-06
+  delta_timestamps:
+    observation.environment_state:
+    - -0.1
+    - 0.0
+    observation.state:
+    - -0.1
+    - 0.0
+    action:
+    - -0.1
+    - 0.0
+    - 0.1
+    - 0.2
+    - 0.3
+    - 0.4
+    - 0.5
+    - 0.6
+    - 0.7
+    - 0.8
+    - 0.9
+    - 1.0
+    - 1.1
+    - 1.2
+    - 1.3
+    - 1.4
+    - 1.5
+    - 1.6
+    - 1.7
+    - 1.8
+    - 1.9
+    - 2.0
+    - 2.1
+    - 2.2
+    - 2.3
+    - 2.4
+    - 2.5
+    - 2.6
+    - 2.7
+    - 2.8
+    - 2.9
+    - 3.0
+    - 3.1
+    - 3.2
+    - 3.3
+    - 3.4
+    - 3.5
+    - 3.6
+    - 3.7
+    - 3.8
+    - 3.9
+    - 4.0
+    - 4.1
+    - 4.2
+    - 4.3
+    - 4.4
+    - 4.5
+    - 4.6
+    - 4.7
+    - 4.8
+    - 4.9
+    - 5.0
+    - 5.1
+    - 5.2
+    - 5.3
+    - 5.4
+    - 5.5
+    - 5.6
+    - 5.7
+    - 5.8
+    - 5.9
+    - 6.0
+    - 6.1
+    - 6.2
+  drop_n_last_frames: 7
+eval:
+  n_episodes: 50
+  batch_size: 50
+  use_async_envs: false
+wandb:
+  enable: true
+  disable_artifact: false
+  project: lerobot
+  notes: ''
+fps: 10
+env:
+  name: maze2d
+  task: null
+  state_dim: 2
+  action_dim: 2
+  fps: ${fps}
+policy:
+  name: diffusion
+  n_obs_steps: 2
+  horizon: 64
+  n_action_steps: 8
+  input_shapes:
+    observation.environment_state:
+    - 2
+    observation.state:
+    - ${env.state_dim}
+  output_shapes:
+    action:
+    - ${env.action_dim}
+  input_normalization_modes:
+    observation.environment_state: min_max
+    observation.state: min_max
+  output_normalization_modes:
+    action: min_max
+  vision_backbone: resnet18
+  crop_shape:
+  - 84
+  - 84
+  crop_is_random: true
+  pretrained_backbone_weights: null
+  use_group_norm: true
+  spatial_softmax_num_keypoints: 32
+  down_dims:
+  - 128
+  - 256
+  - 512
+  kernel_size: 5
+  n_groups: 8
+  diffusion_step_embed_dim: 128
+  use_film_scale_modulation: true
+  noise_scheduler_type: DDIM
+  num_train_timesteps: 100
+  beta_schedule: squaredcos_cap_v2
+  beta_start: 0.0001
+  beta_end: 0.02
+  prediction_type: epsilon
+  clip_sample: true
+  clip_sample_range: 1.0
+  num_inference_steps: 10
+  do_mask_loss_for_padding: false

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae14bf229318d2bf3f652086ec96336233acd4a575fdae273d1c3c80e64a3d80
+size 65524656