case class CrfParams(minEpochs: Int = 10, maxEpochs: Int = 1000, l2: Float = 1f, c0: Int = 1500000, lossEps: Float = 1e-4f, randomSeed: Option[Int] = None, verbose: nlp.annotators.ner.Verbose.Value = Verbose.Silent) extends Product with Serializable
Hyper Parameters and Setting for LinearChainCrf training
- minEpochs
\- Minimum number of epochs to train
- maxEpochs
\- Maximum number of epochs to train
- l2
\- l2 regularization coefficient
- c0
\- Initial number of steps in decay strategy
- lossEps
\- If loss after a SGD epochs haven't improved (absolutely) more than lossEps, then training is stopped
- randomSeed
\- Seed for random
- verbose
\- Level of verbosity during training procedure
Linear Supertypes
Ordering
- Alphabetic
- By Inheritance
Inherited
- CrfParams
- Serializable
- Serializable
- Product
- Equals
- AnyRef
- Any
- Hide All
- Show All
Visibility
- Public
- All
Instance Constructors
-
new
CrfParams(minEpochs: Int = 10, maxEpochs: Int = 1000, l2: Float = 1f, c0: Int = 1500000, lossEps: Float = 1e-4f, randomSeed: Option[Int] = None, verbose: nlp.annotators.ner.Verbose.Value = Verbose.Silent)
- minEpochs
\- Minimum number of epochs to train
- maxEpochs
\- Maximum number of epochs to train
- l2
\- l2 regularization coefficient
- c0
\- Initial number of steps in decay strategy
- lossEps
\- If loss after a SGD epochs haven't improved (absolutely) more than lossEps, then training is stopped
- randomSeed
\- Seed for random
- verbose
\- Level of verbosity during training procedure