Skip to content

class QOTracker(pydantic.BaseModel)#

Parent classes#

class pydantic.BaseModel

Note

class QOTracker is an element of internal API of nneve.quantum_oscillator subpackage was only ment to interact with class QONetwork and class QOConstants.

class QOTracker is responsible for gathering network state metrics during network learning process. The class is also equipped with methods to plot graphs based on the collected data. The class is also equipped with methods to plot graphs based on the collected data. Its additional ability is to create network status messages that are shown on the command line while the network is being learned.

Instance attributes#

Note

  • Attributes are mutable
  • Arbitrary types are allowed to be used as attribute values

c: float#

total_loss: List[float] = Field(default_factory=list)#

Stores history of total_loss values which were presented by the network during training process.

eigenvalue: List[float] = Field(default_factory=list)#

Stores history of eigenvalue values which were presented by the network during training process.

residuum: List[float] = Field(default_factory=list)#

Stores history of residuum values which were presented by the network during training process.

function_loss: List[float] = Field(default_factory=list)#

Stores history of function_loss values which were presented by the network during training process.

eigenvalue_loss: List[float] = Field(default_factory=list)#

Stores history of eigenvalue_loss values which were presented by the network during training process.

drive_loss: List[float] = Field(default_factory=list)#

Stores history of drive_loss values which were presented by the network during training process.

c: List[float] = Field(default_factory=list)#

Stores history of c values which were presented by the network during training process.

Instance methods#

def push_stats(self, ...) -> None#

Parameters#

Appends passed positional arguments to corresponding histories.

name type description
total_loss float value to be appended to total_loss history list.
eigenvalue float value to be appended to eigenvalue history list.
residuum float value to be appended to residuum history list.
function_loss float value to be appended to function_loss history list.
eigenvalue_loss float value to be appended to eigenvalue_loss history list.
drive_loss float value to be appended to drive_loss history list.
c float value to be appended to c history list.
*_ Any extra arguments are ignored.

Returns:#

type description
NoneType no value is returned