You can deploy this model and start earning money today!

antony-zhao/
lehome-dreamer-act

Name: lehome-dreamer-act
Availability: InStock

safetensors

Views

No views yet

LeHome Challenge — Dreamer+ACT Policy

Setup

Place ACT checkpoint files into outputs/train/act_dreamer_300k/checkpoints/300000/pretrained_model/
Place world_model-step_90000.pth into outputs/dreamer_offline/checkpoints/
Place dreamer_act_policy.py into scripts/eval_policy/ and add from .dreamer_act_policy import DreamerACTPolicy to scripts/eval_policy/__init__.py
Place dreamer.py, dreamer_utils.py, models.py, buffers.py into scripts/
Place dataset metadata from Datasets/example/four_types_dreamer/meta into the same path
In scripts/utils/evaluation.py, change if args.policy_type == "lerobot": to if args.policy_type in ("lerobot", "dreamer_act"):

Evaluation

bash

python -m scripts.eval \
    --policy_type dreamer_act \
    --policy_path outputs/train/act_dreamer_300k/checkpoints/300000/pretrained_model \
    --dataset_root Datasets/example/four_types_dreamer \
    --num_episodes 5 \
    --enable_cameras \
    --device cpu \
    --headless

Self-Reported Results

Success Rate: 65.42%
Total Episodes: 240

antony-zhao/lehome-dreamer-act

LeHome Challenge — Dreamer+ACT Policy

Setup

Evaluation

Self-Reported Results

antony-zhao/
lehome-dreamer-act