AllocationGrowthScenario
Data scenarios · Implementation: mlsysim/engine/scenario_definitions/data.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
A compute-optimal allocation X_opt ~ C^exponent when compute grows by compute_multiple.
Notes
Equation: growth is the worked quantity; each property below states its equation and units.
Assumptions: A positive compute multiple scales an allocation by the supplied power-law exponent.
Excludes: Fitting the exponent or predicting training quality.
References
Derivation: The scenario binds declared inputs to calc_allocation_growth. Those implementations document their derivations and sources.
Examples
Execute the registered worked inputs; this is a modeled calculation.
>>> from mlsysim import Scenarios
>>> case = Scenarios.ChinchillaComputeDoubling
>>> value = case.growth
>>> round(value, 3)
0.414Attributes
| Name | Description |
|---|---|
| compute_multiple | |
| exponent | |
| growth | Dimensionless allocation growth from the shared power-law helper. |
compute_multiple
Type: float.
Declared default: required.
exponent
Type: float.
Declared default: required.
growth
engine.scenario_types.AllocationGrowthScenario.growth
Dimensionless allocation growth from the shared power-law helper.
Fractional growth of the allocation.
Equation: return calc_allocation_growth(self.compute_multiple, self.exponent)