T-WLA-FD

The final model.

One file, T-WLA-FD.pt: a PyTorch checkpoint {"model": state_dict} with 32-bit weights.

huggingface-cli download anon-submission-0518/T-WLA-FD T-WLA-FD.pt --local-dir checkpoints

Place it at checkpoints/T-WLA-FD.pt in the code repository; that is the default location of the inference code (vla_hand_pose/inference/policy.py).

The world model used in training is in anon-submission-0518/T-WLA-WM.