---
title: "init(inputSize:hiddenSize:layerCount:usesBiases:batchFirst:isBidirectional:dropout:)"
framework: ML Compute
role: symbol
role_heading: Initializer
platforms: [iOS 14.0+, iPadOS 14.0+, Mac Catalyst 14.0+, macOS 11.0+, tvOS 14.0+]
path: "mlcompute/mlclstmdescriptor/init(inputsize:hiddensize:layercount:usesbiases:batchfirst:isbidirectional:dropout:)"
---

# init(inputSize:hiddenSize:layerCount:usesBiases:batchFirst:isBidirectional:dropout:)

Creates a batch first LSTM descriptor that allows you to indicate whether the input and output shape is batch first.

## Declaration

```swift
convenience init(inputSize: Int, hiddenSize: Int, layerCount: Int, usesBiases: Bool, batchFirst: Bool, isBidirectional: Bool, dropout: Float)
```

## Parameters

- `inputSize`: The number of expected features in the input.
- `hiddenSize`: The number of features in the hidden state.
- `layerCount`: The number of recurrent layers.
- `usesBiases`: A Boolean that indicates whether you use bias weights. The default value is true.
- `batchFirst`: A Boolean that indicates whether the first index of the input and output shape contains the batch size. The default value is true.
- `isBidirectional`: A Boolean that indicates whether the layer becomes bidirectional. The default value is false.
- `dropout`: The dropout probability rate.

## See Also

### Creating LSTM Descriptors

- [init(inputSize:hiddenSize:layerCount:)](mlcompute/mlclstmdescriptor/init(inputsize:hiddensize:layercount:).md)
- [init(inputSize:hiddenSize:layerCount:usesBiases:isBidirectional:dropout:)](mlcompute/mlclstmdescriptor/init(inputsize:hiddensize:layercount:usesbiases:isbidirectional:dropout:).md)
- [init(inputSize:hiddenSize:layerCount:usesBiases:batchFirst:isBidirectional:returnsSequences:dropout:)](mlcompute/mlclstmdescriptor/init(inputsize:hiddensize:layercount:usesbiases:batchfirst:isbidirectional:returnssequences:dropout:).md)
- [init(inputSize:hiddenSize:layerCount:usesBiases:batchFirst:isBidirectional:returnsSequences:dropout:resultMode:)](mlcompute/mlclstmdescriptor/init(inputsize:hiddensize:layercount:usesbiases:batchfirst:isbidirectional:returnssequences:dropout:resultmode:).md)