DecoderFamiliesScenario
Robotics scenarios · Implementation: mlsysim/engine/scenario_definitions/robotics/proposers.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
One fresh machine.chunk_horizon-step chunk from backbone, per decoder family.
Notes
Equation: result = DecoderFamiliesModel.solve(…) with the declared backbone, fast_tokens, expert_params inputs.
Assumptions: The inputs instantiate the assumptions documented by mlsysim.solvers.DecoderFamiliesModel.solve; accessing result evaluates that solver again.
Excludes: Any extra mechanism beyond DecoderFamiliesModel; this record supplies configuration and does not collect performance measurements.
References
Derivation: This configuration forwards its stored inputs to mlsysim.solvers.DecoderFamiliesModel.solve. The solver contract records the equations and their provenance.
Examples
The named scenario supplies illustrative inputs; the output is calculated.
>>> from mlsysim import Scenarios
>>> case = Scenarios.IntentBackboneDecoderFamilies
>>> round(float(case.result.binned_tokens), 3)
112.0Attributes
| Name | Description |
|---|---|
| backbone | |
| expert_bytes_per_param | |
| expert_params | |
| expert_steps | |
| fast_tokens | |
| kv_bytes_per_value | |
| result | Modeled result from mlsysim.solvers.DecoderFamiliesModel.solve using this |
backbone
Type: VlaEvaluationScenario.
Declared default: required.
fast_tokens
Type: int.
Declared default: required.
expert_params
Type: Quantity.
Declared default: required.
expert_steps
Type: int.
Declared default: required.
expert_bytes_per_param
Type: Quantity.
Declared default: required.
kv_bytes_per_value
Type: Quantity.
Declared default: required.
result
engine.scenario_types.DecoderFamiliesScenario.result
Modeled result from mlsysim.solvers.DecoderFamiliesModel.solve using this scenario’s declared inputs; see that method for units, limits, and feasibility.
Equation: Delegates the declared fields to mlsysim.solvers.DecoderFamiliesModel.solve; returned quantities and feasibility follow that contract.