|
| 1 | +import { |
| 2 | + Tensor, |
| 3 | + Module, Parameter, |
| 4 | + Linear, ReLU, |
| 5 | + Adam, SGD, GradScaler, |
| 6 | + mseLoss, |
| 7 | +} from '../dist/index.js'; |
| 8 | +import { assert, assertClose, section } from './helpers.js'; |
| 9 | + |
| 10 | +// ============================================================ |
| 11 | +// Module system |
| 12 | +// ============================================================ |
| 13 | + |
| 14 | +section('Module system'); |
| 15 | + |
| 16 | +class TestNet extends Module { |
| 17 | + l1: any; |
| 18 | + l2: any; |
| 19 | + relu: any; |
| 20 | + constructor() { |
| 21 | + super(); |
| 22 | + this.l1 = new Linear(3, 4); |
| 23 | + this.l2 = new Linear(4, 2); |
| 24 | + this.relu = new ReLU(); |
| 25 | + } |
| 26 | + forward(x: any) { |
| 27 | + return this.l2.forward(this.relu.forward(this.l1.forward(x))); |
| 28 | + } |
| 29 | +} |
| 30 | + |
| 31 | +const net = new TestNet(); |
| 32 | + |
| 33 | +const params = net.parameters(); |
| 34 | +assert(params.length === 4, 'TestNet has 4 parameters (2 weights + 2 biases)'); |
| 35 | + |
| 36 | +const named = net.namedParameters(); |
| 37 | +assert(named.length === 4, 'namedParameters count'); |
| 38 | +const names = named.map(([n]: [string, any]) => n); |
| 39 | +assert(names.some((n: string) => n.includes('l1')), 'namedParameters includes l1'); |
| 40 | +assert(names.some((n: string) => n.includes('l2')), 'namedParameters includes l2'); |
| 41 | + |
| 42 | +const kids = net.children(); |
| 43 | +assert(kids.length === 3, 'TestNet has 3 children (l1, l2, relu)'); |
| 44 | + |
| 45 | +const allMods = net.modules(); |
| 46 | +assert(allMods.length >= 4, 'modules() includes self + children'); |
| 47 | + |
| 48 | +net.eval(); |
| 49 | +assert(net.training === false, 'eval sets training=false'); |
| 50 | +net.train(); |
| 51 | +assert(net.training === true, 'train sets training=true'); |
| 52 | + |
| 53 | +const netInput = Tensor.rand([2, 3]); |
| 54 | +const netOut = net.forward(netInput); |
| 55 | +assert(netOut.shape[0] === 2 && netOut.shape[1] === 2, 'TestNet output shape'); |
| 56 | + |
| 57 | +// ============================================================ |
| 58 | +// Optimizers |
| 59 | +// ============================================================ |
| 60 | + |
| 61 | +section('Optimizers'); |
| 62 | + |
| 63 | +// SGD |
| 64 | +const sgdParam = Tensor.fromFloat32(new Float32Array([5, 5, 5, 5]), [2, 2]).setRequiresGrad(true); |
| 65 | +const sgdParamObj = new Parameter(sgdParam); |
| 66 | +const sgdTarget = Tensor.zeros([2, 2]); |
| 67 | +const sgdLoss = sgdParamObj.value.sub(sgdTarget).pow(2).mean(); |
| 68 | +sgdLoss.backward(); |
| 69 | +const sgd = new SGD([sgdParamObj], 0.1); |
| 70 | +const sgdBefore = sgdParamObj.value.toFloat32()[0]; |
| 71 | +sgd.step(); |
| 72 | +const sgdAfter = sgdParamObj.value.toFloat32()[0]; |
| 73 | +assert(sgdAfter < sgdBefore, 'SGD step reduces parameter toward target'); |
| 74 | +sgd.zeroGrad(); |
| 75 | + |
| 76 | +// Adam |
| 77 | +const adamParam = Tensor.fromFloat32(new Float32Array([5, 5, 5, 5]), [2, 2]).setRequiresGrad(true); |
| 78 | +const adamParamObj = new Parameter(adamParam); |
| 79 | +const adamTarget = Tensor.zeros([2, 2]); |
| 80 | +const adamLoss = adamParamObj.value.sub(adamTarget).pow(2).mean(); |
| 81 | +adamLoss.backward(); |
| 82 | +const adam = new Adam([adamParamObj], { lr: 0.01 }); |
| 83 | +adam.step(); |
| 84 | +const adamAfter = adamParamObj.value.toFloat32()[0]; |
| 85 | +assert(adamAfter !== 5, 'Adam step changes parameter'); |
| 86 | +adam.zeroGrad(); |
| 87 | + |
| 88 | +// Adam returns grad norm |
| 89 | +const adamParam2 = Tensor.fromFloat32(new Float32Array([3, 3]), [2]).setRequiresGrad(true); |
| 90 | +const adamParamObj2 = new Parameter(adamParam2); |
| 91 | +const adamLoss2 = adamParamObj2.value.pow(2).sum(); |
| 92 | +adamLoss2.backward(); |
| 93 | +const adam2 = new Adam([adamParamObj2], { lr: 0.01 }); |
| 94 | +const gradNorm = adam2.step(); |
| 95 | +assert(typeof gradNorm === 'number', 'Adam.step() returns grad norm'); |
| 96 | + |
| 97 | +// GradScaler |
| 98 | +const scaler = new GradScaler({ initScale: 1024 }); |
| 99 | +assert(scaler.getScale() === 1024, 'GradScaler initial scale'); |
| 100 | + |
| 101 | +const gsLossInput = Tensor.fromFloat32(new Float32Array([2, 3]), [2]); |
| 102 | +const scaledLoss = scaler.scaleLoss(gsLossInput); |
| 103 | +const scaledData = scaledLoss.toFloat32(); |
| 104 | +assertClose(scaledData[0], 2 * 1024, 1e-1, 'scaleLoss scales by initScale'); |
| 105 | +assertClose(scaledData[1], 3 * 1024, 1e-1, 'scaleLoss scales second element'); |
| 106 | + |
| 107 | +// ============================================================ |
| 108 | +// End-to-end training loop |
| 109 | +// ============================================================ |
| 110 | + |
| 111 | +section('End-to-end training'); |
| 112 | + |
| 113 | +const trainX = Tensor.fromFloat32(new Float32Array([0, 1, 2, 3, 4, 5]), [6, 1]); |
| 114 | +const trainY = Tensor.fromFloat32(new Float32Array([1, 3, 5, 7, 9, 11]), [6, 1]); |
| 115 | +const regNet = new Linear(1, 1); |
| 116 | +const regOptim = new Adam(regNet.parameters(), { lr: 0.05 }); |
| 117 | + |
| 118 | +let earlyLoss: number | null = null; |
| 119 | +for (let i = 0; i < 200; i++) { |
| 120 | + regOptim.zeroGrad(); |
| 121 | + const pred = regNet.forward(trainX); |
| 122 | + const loss = mseLoss(pred, trainY); |
| 123 | + if (i === 10) earlyLoss = loss.toFloat32()[0]; |
| 124 | + loss.backward(); |
| 125 | + regOptim.step(); |
| 126 | +} |
| 127 | +const finalPred = regNet.forward(trainX); |
| 128 | +const finalLoss = mseLoss(finalPred, trainY).toFloat32()[0]; |
| 129 | +assert(finalLoss < earlyLoss!, 'training reduces loss'); |
| 130 | +assert(finalLoss < 1.0, 'training converges to low loss'); |
| 131 | + |
0 commit comments