OpenVLA: REST policy inference server

Pinned server entrypoint that loads an OpenVLA policy and serves action predictions over its upstream FastAPI REST interface. Statically reviewed source only; not run and no Hub preparation adapter.

openvla-vla-scripts-deploy-py | Record version: source-14d74c56a437058eb473f41f93a0a82ec892e16e55ed91236c5d30f4fd8adc65

Record JSON | Markdown | Pinned upstream source

Access

Fetch the pinned repository, verify this source checksum, and use upstream installation and invocation. This record does not install or execute code.

Hub preparation adapter: false. Follow the upstream setup and invocation contract.

Source and license

Repository: https://github.com/openvla/openvla
Revision: c8f03f48af692657d3060c19588038c7220e9af9
Path: vla-scripts/deploy.py
Source SHA256: 14d74c56a437058eb473f41f93a0a82ec892e16e55ed91236c5d30f4fd8adc65
Original authors: Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti
License: MIT
License scope: Selected first-party Python workflow source only. Preserve upstream copyright and per-file notices. Model weights, datasets, third-party code, robot drivers/firmware and services retain separate terms; no rights for those assets are inferred.
License SHA256: c4ee2ba5958af03d74b1d3dfa174e3749171d689fb4c23cc170749c05ae3eeb5
Hub commit: 1d6d9eb134e397d910265a8d6e7d8f257f9e8d7f

Artifact

{
  "kind": "workflow",
  "format": "python",
  "workflow_role": "policy-serving"
}

Requirements

{
  "python": ">=3.8 in project metadata; upstream setup uses Python 3.10.",
  "packages": [
    "Pinned OpenVLA checkout",
    "Upstream PyTorch 2.2.0, transformers 4.40.1, timm 0.9.10 and tokenizers 0.19.1 pins",
    "Workflow-specific dependencies in pinned README and source"
  ],
  "display": "Hardware: Docstring describes a GPU server. Source selects cuda:0 if available and otherwise CPU, while requesting flash_attention_2 and bfloat16. CPU success and minimum VRAM are not established. Weights: Loads AutoProcessor and AutoModelForVision2Seq from the configured model path; default is openvla/openvla-7b. from_pretrained is called with trust_remote_code=True. No weights or remote model code were fetched. Datasets: Accepts caller-provided image/instruction observations; no training dataset is required by this server entrypoint. Services/auth: Starts a local HTTP API, default host 0.0.0.0/port 8000. No provider API key or authentication requirement is declared by the selected entrypoint; remote model retrieval may occur when a model ID is used.",
  "execution_domain": "computation",
  "source_imports": [
    "PIL",
    "dataclasses",
    "draccus",
    "fastapi",
    "fastapi.responses",
    "json",
    "json_numpy",
    "logging",
    "os.path",
    "pathlib",
    "torch",
    "traceback",
    "transformers",
    "typing",
    "uvicorn"
  ],
  "notes": "The source starts a network listener and permits model-provided remote code when actually run. This record is inert provenance, not permission to execute it. Code licensing does not establish model, checkpoint, dataset or service permission.",
  "declared_requirements": {
    "python": ">=3.8 in project metadata; upstream setup uses Python 3.10.",
    "weights": "Loads AutoProcessor and AutoModelForVision2Seq from the configured model path; default is openvla/openvla-7b. from_pretrained is called with trust_remote_code=True. No weights or remote model code were fetched.",
    "datasets": "Accepts caller-provided image/instruction observations; no training dataset is required by this server entrypoint.",
    "hardware": "Docstring describes a GPU server. Source selects cuda:0 if available and otherwise CPU, while requesting flash_attention_2 and bfloat16. CPU success and minimum VRAM are not established.",
    "external_services_and_auth": "Starts a local HTTP API, default host 0.0.0.0/port 8000. No provider API key or authentication requirement is declared by the selected entrypoint; remote model retrieval may occur when a model ID is used.",
    "evidence": [
      {
        "anchor": "self.device = torch.device",
        "line": 74
      },
      {
        "anchor": "trust_remote_code=True",
        "line": 77
      },
      {
        "anchor": "port: int = 8000",
        "line": 120
      }
    ],
    "license_scopes": {
      "code": "MIT for selected first-party workflow source.",
      "weights": "README explicitly says the released OpenVLA models derive from Llama-2 and are subject to the Llama Community License. No checkpoint/model card/base-model license was fetched or evaluated here; MIT is asserted only for the selected first-party Python source.",
      "datasets": "README names BridgeData V2, Open X-Embodiment and LIBERO datasets. Individual dataset terms were not reviewed, and repository MIT is not applied to them.",
      "third_party_code": "External dependencies retain their own licenses."
    },
    "interpretation": "Pinned upstream declarations only; no dependency resolution, model/data access, compatibility inference or runtime validation was performed."
  }
}

Invocation

{
  "entrypoint": "python vla-scripts/deploy.py --openvla_path <approved-local-checkpoint-or-reviewed-model-id>",
  "notes": "The source starts a network listener and permits model-provided remote code when actually run. This record is inert provenance, not permission to execute it."
}

Explicit compatibility

[]

Missing environment fields remain unknown. Discovery labels do not establish robot compatibility.

Evidence

Level: source_reviewed; runtime verification: not_run.

Verified exact source, applicable license and author/setup document bytes at a full Git commit; parsed the Python AST and reviewed this guarded operational entrypoint and its explicit license notices. No upstream code, installation, robot command, dataset, mesh or checkpoint was executed or loaded.

Source review and contributor reports do not establish task completion. Preserve upstream attribution and licenses.