Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.cyberwave.com/llms.txt

Use this file to discover all available pages before exploring further.

AI models are how your twins stop being puppets and start being autonomous. Cyberwave is the substrate that gets them there: a growing catalog of models, the infrastructure to run them on cloud or edge, and the datasets you need to train your own. 
from cyberwave import Cyberwave

cw = Cyberwave()
arm = cw.twin("acme/twins/arm-station-1")
arm.use_controller("acme/models/my-pick-and-place-vla")
That’s it — the model is now driving the robot. Same call works in simulation (cw.affect("simulation")) and against live hardware.

​
Bring your own model — or pick from the catalog

Use the catalog

Pick from open-source and proprietary models — VLAs (SmolVLA, OpenVLA, Pi 0.5), VLMs (Gemini Robotics, GPT-5, Molmo, PaliGemma), detectors (YOLOv8, SAM2), and image-to-3D (Hunyuan3D, TripoSR).

Fine-tune your own

Fine-tune SmolVLA or OpenVLA on the data you collected with your own robots — no separate training stack.

Register a custom model

Bring your own weights or endpoint — Hugging Face, an internal inference server, a custom ONNX file. Cyberwave treats it as a first-class model.
Every registered model gets a Model Playground page where you can try it on an image, see overlays, copy CLI/SDK snippets, and wire it into a workflow — so experimenting with a new model is genuinely one click.

​
Use them anywhere in your stack

Cyberwave models compose with the rest of the platform. Same model, different roles:
Where the model runsWhat it doesReference
As a twin controllerDrive a robot end-to-end (a VLA picks-and-places, an RL policy walks)Controllers
Inside a workflowOne call_model node — runs cloud VLM or edge ML transparentlyWorkflow nodes
From the SDKAsync cloud calls for VLM / LLM tasksvlm_generation / llm_generation
In simulationThe same code drives a MuJoCo twincw.affect("simulation")
The best automations usually combine more than one — an edge YOLO that’s cheap to run on every frame and a cloud VLM that reasons about the rare interesting frame. See the edge-to-cloud VLM tutorial for that pattern end-to-end.

​
Edge + cloud, both first-class

Cyberwave runs models in both places, on purpose.

Edge models

Local inference on your own hardware — fast, private, offline-capable. YOLO, SAM2, ONNX/TensorRT detectors all run inside the edge worker generated from your workflow.

Cloud models

Heavy-weight VLMs and VLAs (Gemini Robotics, GPT-5, OpenVLA) run on a Cloud Node or VLA Cloud Node with a GPU attached — Cyberwave handles the provisioning.
You don’t manage any of the boring parts: weight downloads, GPU provisioning, picking the right edge-compatible variant, network reconnects, version pinning. Cyberwave routes the right model to the right runtime — see the model deployment reference for the wire-level details.

​
Datasets — collect, import, export

Models need data. Cyberwave gives you the loop end-to-end: record on the edge → replay in the browser → slice into episodes → train.

From your own runs

Every recording in Replay can be turned into episodes and a dataset, ready to train on.

Import from anywhere

LeRobot v3 / v2.1, RLDS, HDF5, Zarr, GR00T, MCAP, ROS bag, RoboDM, Hugging Face — see the full import matrix.

Export, no lock-in

Convert any dataset to Cyberwave Parquet, LeRobot v3, RLDS, or OpenVLA TFDS via the export tab or API.
Browse the public dataset catalog, pick one, and feed it straight into SmolVLA or OpenVLA training — the format conversion happens for you.

ML Models reference

Capabilities, providers, registration, inference, and the full VLA stack.

Model Playground

The interactive page behind every model in the catalog.

Sandwich robot (SmolVLA)

A community tutorial: collect data, fine-tune SmolVLA, run it on a real arm — all on Cyberwave.