SpeculativeEpisodeScenario
Economics scenarios · Implementation: mlsysim/engine/scenario_definitions/economics.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
Speculative decoding on a serial episode’s model time, at two draft acceptance levels.
Notes
Equation: The result property delegates to the named solver documented below; it converts declared scenario units before passing the inputs.
Assumptions: The nested serial episode supplies model fraction; draft length, cost ratio, and acceptance are fixed.
Excludes: A prediction of acceptance quality, variable sequence length, and non-model speedup.
References
Derivation: the scenario composes mlsysim.solvers.SpeculativeEpisodeModel. Their contracts supply the underlying equations; property equations show the remaining unit conversions and accounting.
Examples
>>> from mlsysim import Scenarios
>>> scenario = Scenarios.SerialRepairEpisodeSpeculation
>>> value = scenario.result
>>> type(value).__name__
'SpeculativeEpisodeResult'Attributes
| Name | Description |
|---|---|
| accepted_high | |
| accepted_low | |
| draft_cost_ratio | |
| draft_length | |
| episode | |
| result | Return the typed accounting result from SpeculativeEpisodeModel using this |
episode
Type: SerialEpisodeScenario.
Declared default: required.
draft_length
Type: int.
Declared default: required.
draft_cost_ratio
Type: float.
Declared default: required.
accepted_high
Type: float.
Declared default: required.
accepted_low
Type: float.
Declared default: required.
result
engine.scenario_types.SpeculativeEpisodeScenario.result
Return the typed accounting result from SpeculativeEpisodeModel using this scenario’s declared inputs; units and exclusions follow that solver contract.
Equation: SpeculativeEpisodeModel().solve(float(e.total_time.to('second').magnitude), e.model_fraction, draft_length, draft_cost_ratio, accepted_high, accepted_low).