Loading published evidence…
Loading published evidence…
EXTERNAL POLICY ASSESSMENT
A non-executing assessment showing why a base VLA checkpoint is not itself deployment-ready evidence.
Known Robot is the metadata assessor. lerobot is the attributed upstream author or organization. Known Robot is not the policy author or evaluator.
d9f33c94a60fb382c90dea2164c96845bd955e28 ↗metadata_only · validator 1.6.0This assessment inspected the listed metadata files only. It did not download weights, import policy code, run inference, control hardware, execute simulation, or create evidence.
{
"method": "metadata_only",
"assessor": "Known Robot",
"validator_version": "1.6.0",
"assessed_at": "2026-09-24T21:19:56.122708Z",
"executed_policy_code": false,
"evaluated_policy": false,
"established_compatibility": false,
"status": "incomplete",
"artifact_intents": [
"base_policy",
"training_checkpoint"
],
"artifact_intent_notice": "Descriptive classification only; it is not a compatibility, performance, safety or deployment conclusion."
}{
"algorithm": "sha256",
"manifest_sha256": "0ce8ab26273642380c8d0b985572c7d0add2101449b813b67fe27c50514806b7",
"inventory_sha256": "3a7c4a539b533e63ca9de0f88070baf1cbbb300497cf67794951d35d70bddca2",
"claims_sha256": "667372a0d2ecda4370ae97d1d4fe0e564d08a1d4b0c72a60bab05573c27fb177",
"source_revision": "d9f33c94a60fb382c90dea2164c96845bd955e28"
}Each statement stays in one evidence lane. Detection is not declaration, an upstream claim is not a Known Robot measurement, and none of these lanes establishes compatibility.
policy.framework — "lerobot"policy.architecture — "smolvla"runtime.observation_shape — {"observation.state":{"type":"STATE","shape":[6]},"observation.images.camera1":{"type":"VISUAL","shape":[3,256,256]},"observation.images.camera2":{"type":"VISUAL","shape":[3,256,256]},"observation.images.camera3":{"type":"VISUAL","shape":[3,256,256]}}runtime.action_shape — {"action":{"type":"ACTION","shape":[6]}}skill.source.type — "huggingface" · Hugging Face model APIskill.source.repository — "lerobot/smolvla_base" · Hugging Face model APIskill.source.revision — "d9f33c94a60fb382c90dea2164c96845bd955e28" · Hugging Face model APIskill.license — "Apache-2.0" · Hugging Face model-card metadataNone. This assessment did not execute the artifact. Any future observed or measured result must be published separately as evidence.
This is the portable metadata output, including nulls. In provenance-bound 1.4.0+ assessments, verified Hub repository and commit declarations are added explicitly and separately identified above.
{
"skill": {
"name": "Smolvla Base",
"source": {
"type": "huggingface",
"revision": "d9f33c94a60fb382c90dea2164c96845bd955e28",
"repository": "lerobot/smolvla_base"
},
"license": "Apache-2.0",
"version": null
},
"policy": {
"framework": "lerobot",
"architecture": "smolvla",
"framework_version": null
},
"dataset": {
"schema": null,
"revision": null,
"repository": null
},
"runtime": {
"action_shape": {
"action": {
"type": "ACTION",
"shape": [
6
]
}
},
"dependencies": [],
"observation_shape": {
"observation.state": {
"type": "STATE",
"shape": [
6
]
},
"observation.images.camera1": {
"type": "VISUAL",
"shape": [
3,
256,
256
]
},
"observation.images.camera2": {
"type": "VISUAL",
"shape": [
3,
256,
256
]
},
"observation.images.camera3": {
"type": "VISUAL",
"shape": [
3,
256,
256
]
}
},
"control_frequency_hz": null
},
"hardware": {
"gripper": null,
"sensors": [],
"robot_family": null
},
"evaluations": [],
"compatibility": [],
"schema_version": "1.0"
}Missing declarations remain visible. Model-card prose does not silently satisfy portable metadata fields.
skill.version — Declare the policy or artifact version.policy.framework_version — Pin the framework version.hardware.robot_family — Declare the robot family used or targeted.hardware.gripper — Declare the end effector or gripper.hardware.sensors — Describe every required sensor and its role.runtime.control_frequency_hz — Declare the required control frequency in hertz.runtime.dependencies — Provide a dependency lockfile or dependency declaration.dataset.repository — Identify the training or evaluation dataset.dataset.revision — Pin the immutable dataset revision.dataset.schema — Describe the dataset feature schema.runtime.dependencies — Record resolved package versions; a filename or file hash alone is not an environment lock.evaluations — No evaluation result is recorded; compatibility remains unverified.compatibility — No known compatible or incompatible configuration is recorded.Hashes make the metadata snapshot auditable without republishing upstream checkpoint weights.
[
{
"path": "README.md",
"sha256": "a484071b8d7d0591452f795813fb54af9199fc1c48ef1ee995358d5ba07ee070",
"bytes": 4625
},
{
"path": "config.json",
"sha256": "650584b56c104720f7a3c91d1ec6bec9e8de8ac11e60c92ba2fa82d93eda147d",
"bytes": 2299
},
{
"path": "policy_postprocessor.json",
"sha256": "2b78bb742065288df2ec63b0ee35f97a1f6950171cf2272f7e34b1fe3873b17b",
"bytes": 660
},
{
"path": "policy_preprocessor.json",
"sha256": "7683d648280a72f0d51e869fb8dd1596beed90b7f84849330cf1bc26634a4bd6",
"bytes": 1872
}
]Future evidence must preserve its execution protocol, trials, outcomes, and sources. It remains separate from this assessment and cannot retroactively strengthen what the metadata established.