| Safe Haskell | None |
|---|---|
| Language | Haskell2010 |
HyperParameters.Defaults
Description
Twin Delayed Deep Deterministic Policy Gradient Algorithm Defaults
Synopsis
- numEpisodes :: Int
- horizonT :: Int
- numEpochs :: Int
- batchSize :: Int
- rngSeed :: Int
- actionLow :: Float
- actionHigh :: Float
- d :: Int
- c :: Float
- γ :: Tensor
- τ :: Tensor
- decayPeriod :: Int
- σMin :: Float
- σMax :: Float
- σEval :: Tensor
- σAct :: Tensor
- σClip :: Float
- hidDim :: Int
- wInit :: Float
- ηφ :: Tensor
- ηθ :: Tensor
- β1 :: Float
- β2 :: Float
- negativeSlope :: Float
- bufferSize :: Int
- explFreq :: Int
- evalFreq :: Int
- k :: Int
Documentation
numEpisodes :: Int Source #
Number of episodes to play
actionHigh :: Float Source #
Action space upper bound
decayPeriod :: Int Source #
Decay Period
negativeSlope :: Float Source #
Leaky ReLU Slope
bufferSize :: Int Source #
Replay Buffer Size