import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("ewykric/lingbot-va-wamdata-eef-left-wrist-step10000", dtype=torch.bfloat16, device_map="cuda")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
WAMData EEF Left-Wrist Post-Train
Selected checkpoint: checkpoint_step_10000 from
wamdata_eef_left_wrist_gbs32_10k_20260717_104512.
The Hugging Face repository for this post-trained model contains only:
transformer/config.jsontransformer/diffusion_pytorch_model.safetensorsdeploy/lingbot_action_stats_wamdata.jsonSHA256SUMS
Download robbyant/lingbot-va-base separately. At runtime, assemble its vae,
tokenizer, and text_encoder with this repository's transformer.
Robot Contract
- Cameras:
exterior_image_1_left, thenwrist_image_left. - Input resolution in the model server: 224 x 320 per camera.
- Dataset and control rate: 30 Hz.
- Action representation: absolute left EEF
xyz + quaternion_xyzw + gripper. - Returned active action shape:
(8, 4, 20). - The first 20-action slot of the initial chunk is conditioning, not executable.
- Normalize predicted quaternions to unit length before sending them to the robot.
- Inactive right-arm and joint channels are masked during training and sampling.
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support