Skip to content

Latest commit

 

History

History
53 lines (30 loc) · 2.21 KB

Experiments.md

File metadata and controls

53 lines (30 loc) · 2.21 KB
Raw

Here we document some experiments that are included in the paper. For most experiments, we only list the configuration files without repeating the training command (see Train.md for detailed commands).

Our training & eval logs can be found at training-logs in the box folder https://rutgers.box.com/s/uzozemx67kje58ycy3lyzf1zgddz8tyq.

Baselines

  • Push-T (Diffusion Policy [T]): pusht/configs/experiments/image_pusht_diffusion_policy_trans.yaml

  • ALOHA (ACT): aloha/configs/experiments/act.yaml

    Note the original ACT mentions 7 decoding layers but only uses 1 due to a code issue. We set the number of decoding layers to be 4.

  • RLBench (RVT): rlb/configs/rvt2.yaml.

    The pretrained model of RVT2 can be found at weights/rlb/rvt/rvt2_without_time_model_70000.pth. You can run the evaluation with the following command (download the model first):

    python3 eval.py config=./configs/rvt2.yaml  model.weights=./weights/rvt/rvt2_without_time_model_70000.pth

    Note you can also run the official rvt/rvt2 models. These models are in the same folder weights/rlb/rvt (in box). Their config files are rlb/configs/rvt1.official.yaml and rlb/configs/rvt2.official.yaml.

    # rvt
python3 eval.py config=./configs/rvt1.official.yaml  model.weights=./weights/rvt/rvt1_official_model_14.pth

# rvt2
python3 eval.py config=./configs/rvt2.official.yaml  model.weights=./weights/rvt/rvt2_official_model_99.pth

    They shall reproduce the same results as in their papers. To change the output folder, set up headless rendering, find more detailed command in Eval.md.

One-step Prediction

  • Push-T: pusht/configs/experiments/one_step_prediction.yaml

  • ALOHA: aloha/configs/experiments/one_step_prediction.yaml

Chunk Sizes

  • Push-T:
    • pusht/configs/experiments/chunk_size/high_level_plan_c{1,2,3,4}.yaml
    • pusht/configs/experiments/chunk_size/low_level_action_c{1,2,4,8}.yaml

Existing Methods on Different Environments

  • Diffusion Policy on ALOHA (does not work well): aloha/configs/experiments/diffusion_policy.yaml

  • ACT on Push-T: pusht/configs/experiments/pusht_act.yaml

  • ACT on RLBench: rlb/configs/act.yaml