Use common layer builder for identity-like layer

lib/model/nns/layer/base.js

@@ -143,12 +143,18 @@ const buildUnaryLayer = (name, calcFunc, gradFunc) => {
 	class TempLayer extends Layer {
 		calc(x) {
 			this._i = x
+			if (!calcFunc) {
+				return (this._o = x)
+			}
 			this._o = x.copy()
 			this._o.map(calcFunc)
 			return this._o
 		}
 
 		grad(bo) {
+			if (!gradFunc) {
+				return bo
+			}
 			const bi = bo.copy()
 			for (let i = 0; i < bi.length; i++) {
 				bi.value[i] *= gradFunc(this._i.value[i], this._o.value[i])
@@ -199,10 +205,7 @@ const unaryLayers = {
 		calc: v => ~v,
 		grad: () => 0,
 	},
-	ceil: {
-		calc: Math.ceil,
-		grad: () => 1,
-	},
+	ceil: { calc: Math.ceil },
 	cloglog: {
 		calc: v => 1 - Math.exp(-Math.exp(v)),
 		grad: (i, o) => Math.exp(i) * (1 - o),
@@ -219,6 +222,7 @@ const unaryLayers = {
 		calc: Math.cosh,
 		grad: i => Math.sinh(i),
 	},
+	detach: { grad: () => 0 },
 	elish: {
 		calc: v => (v >= 0 ? v : Math.exp(v) - 1) / (1 + Math.exp(-v)),
 		grad: (i, o) => ((i >= 0 ? 1 : Math.exp(i)) + o * Math.exp(-i)) / (1 + Math.exp(-i)),
@@ -247,10 +251,7 @@ const unaryLayers = {
 		calc: Math.exp,
 		grad: (i, o) => o,
 	},
-	floor: {
-		calc: Math.floor,
-		grad: () => 1,
-	},
+	floor: { calc: Math.floor },
 	gelu: {
 		calc: v => {
 			const erf = v => {
@@ -278,6 +279,7 @@ const unaryLayers = {
 		calc: v => (v <= -3 ? 0 : 3 <= v ? v : (v * (v + 3)) / 6),
 		grad: i => (i <= -3 ? 0 : 3 <= i ? 1 : (2 * i + 3) / 6),
 	},
+	identity: {},
 	is_inf: {
 		calc: v => v === Infinity || v === -Infinity,
 		grad: () => 0,
@@ -336,14 +338,8 @@ const unaryLayers = {
 		calc: v => v / (1 + Math.sqrt(1 + v ** 2)),
 		grad: i => 1 / (Math.sqrt(i ** 2 + 1) + i ** 2 + 1),
 	},
-	round: {
-		calc: Math.round,
-		grad: () => 1,
-	},
-	sign: {
-		calc: Math.sign,
-		grad: () => 1,
-	},
+	round: { calc: Math.round },
+	sign: { calc: Math.sign },
 	silu: {
 		calc: v => v / (1 + Math.exp(-v)),
 		grad: (i, o) => 1 / (1 + Math.exp(-i)) + o * (1 - o / i),

lib/model/nns/layer/index.js

@@ -17,7 +17,6 @@ export { default as CondLayer } from './cond.js'
 export { default as ConstLayer } from './const.js'
 export { default as ConvLayer } from './conv.js'
 export { default as ConcatenatedReLULayer } from './crelu.js'
-export { default as DetachLayer } from './detach.js'
 export { default as DropoutLayer } from './dropout.js'
 export { default as ElasticELULayer } from './eelu.js'
 export { default as ELULayer } from './elu.js'
@@ -38,7 +37,6 @@ export { default as HardSigmoidLayer } from './hard_sigmoid.js'
 export { default as HardTanhLayer } from './hard_tanh.js'
 export { default as HexpoLayer } from './hexpo.js'
 export { default as HuberLayer } from './huber.js'
-export { default as IdentityLayer } from './identity.js'
 export { default as IncludeLayer } from './include.js'
 export { default as InputLayer } from './input.js'
 export { default as ImprovedSigmoidLayer } from './isigmoid.js'

tests/lib/model/nns/layer/detach.test.js

@@ -2,17 +2,17 @@ import NeuralNetwork from '../../../../../lib/model/neuralnetwork.js'
 import Matrix from '../../../../../lib/util/matrix.js'
 import Tensor from '../../../../../lib/util/tensor.js'
 
-import DetachLayer from '../../../../../lib/model/nns/layer/detach.js'
+import Layer from '../../../../../lib/model/nns/layer/base.js'
 
 describe('layer', () => {
 	test('construct', () => {
-		const layer = new DetachLayer({})
+		const layer = Layer.fromObject({ type: 'detach' })
 		expect(layer).toBeDefined()
 	})
 
 	describe('calc', () => {
 		test('matrix', () => {
-			const layer = new DetachLayer({})
+			const layer = Layer.fromObject({ type: 'detach' })
 
 			const x = Matrix.randn(100, 10)
 			const y = layer.calc(x)
@@ -24,7 +24,7 @@ describe('layer', () => {
 		})
 
 		test('tensor', () => {
-			const layer = new DetachLayer({})
+			const layer = Layer.fromObject({ type: 'detach' })
 
 			const x = Tensor.randn([15, 10, 7])
 			const y = layer.calc(x)
@@ -40,7 +40,7 @@ describe('layer', () => {
 
 	describe('grad', () => {
 		test('matrix', () => {
-			const layer = new DetachLayer({})
+			const layer = Layer.fromObject({ type: 'detach' })
 
 			const x = Matrix.randn(100, 10)
 			layer.calc(x)
@@ -55,7 +55,7 @@ describe('layer', () => {
 		})
 
 		test('tensor', () => {
-			const layer = new DetachLayer({})
+			const layer = Layer.fromObject({ type: 'detach' })
 
 			const x = Tensor.randn([15, 10, 7])
 			layer.calc(x)
@@ -73,15 +73,15 @@ describe('layer', () => {
 	})
 
 	test('toObject', () => {
-		const layer = new DetachLayer({})
+		const layer = Layer.fromObject({ type: 'detach' })
 
 		const obj = layer.toObject()
 		expect(obj).toEqual({ type: 'detach' })
 	})
 
 	test('fromObject', () => {
-		const layer = DetachLayer.fromObject({ type: 'detach' })
-		expect(layer).toBeInstanceOf(DetachLayer)
+		const layer = Layer.fromObject({ type: 'detach' })
+		expect(layer).toBeDefined()
 	})
 })
 

tests/lib/model/nns/layer/identity.test.js

@@ -2,17 +2,17 @@ import NeuralNetwork from '../../../../../lib/model/neuralnetwork.js'
 import Matrix from '../../../../../lib/util/matrix.js'
 import Tensor from '../../../../../lib/util/tensor.js'
 
-import IdentityLayer from '../../../../../lib/model/nns/layer/identity.js'
+import Layer from '../../../../../lib/model/nns/layer/base.js'
 
 describe('layer', () => {
 	test('construct', () => {
-		const layer = new IdentityLayer({})
+		const layer = Layer.fromObject({ type: 'identity' })
 		expect(layer).toBeDefined()
 	})
 
 	describe('calc', () => {
 		test('matrix', () => {
-			const layer = new IdentityLayer({})
+			const layer = Layer.fromObject({ type: 'identity' })
 
 			const x = Matrix.randn(100, 10)
 			const y = layer.calc(x)
@@ -24,7 +24,7 @@ describe('layer', () => {
 		})
 
 		test('tensor', () => {
-			const layer = new IdentityLayer({})
+			const layer = Layer.fromObject({ type: 'identity' })
 
 			const x = Tensor.randn([15, 10, 7])
 			const y = layer.calc(x)
@@ -40,7 +40,7 @@ describe('layer', () => {
 
 	describe('grad', () => {
 		test('matrix', () => {
-			const layer = new IdentityLayer({})
+			const layer = Layer.fromObject({ type: 'identity' })
 
 			const x = Matrix.randn(100, 10)
 			layer.calc(x)
@@ -55,7 +55,7 @@ describe('layer', () => {
 		})
 
 		test('tensor', () => {
-			const layer = new IdentityLayer({})
+			const layer = Layer.fromObject({ type: 'identity' })
 
 			const x = Tensor.randn([15, 10, 7])
 			layer.calc(x)
@@ -73,15 +73,15 @@ describe('layer', () => {
 	})
 
 	test('toObject', () => {
-		const layer = new IdentityLayer({})
+		const layer = Layer.fromObject({ type: 'identity' })
 
 		const obj = layer.toObject()
 		expect(obj).toEqual({ type: 'identity' })
 	})
 
 	test('fromObject', () => {
-		const layer = IdentityLayer.fromObject({ type: 'identity' })
-		expect(layer).toBeInstanceOf(IdentityLayer)
+		const layer = Layer.fromObject({ type: 'identity' })
+		expect(layer).toBeDefined()
 	})
 })