From 5a6f5e77a63cebefb2087901ecaad22969ff2281 Mon Sep 17 00:00:00 2001
From: saliherdemk <saliherdem_kaymak@hotmail.com>
Date: Mon, 14 Oct 2024 18:09:07 +0300
Subject: [PATCH] shown calculation parameters corrected

---
 js/Draw/MLP/MLPView.js  | 18 ++++++++++++------
 js/Draw/MLP/Playable.js |  4 +++-
 js/MLP/Mlp.js           | 19 +++++++++----------
 3 files changed, 24 insertions(+), 17 deletions(-)

diff --git a/js/Draw/MLP/MLPView.js b/js/Draw/MLP/MLPView.js
index a045e9f..c7792eb 100644
--- a/js/Draw/MLP/MLPView.js
+++ b/js/Draw/MLP/MLPView.js
@@ -179,16 +179,23 @@ class MlpView extends Playable {
     };
   }
 
+  setCalculationData() {
+    const layers = this.origin.layers;
+    const slicedDataAll = [];
+    for (let i = 0; i < layers.length; i++) {
+      const { weights, biases, z, outputs } = layers[i];
+      slicedDataAll.push(this.sliceData(weights, biases, z, outputs));
+    }
+    this.calculationComponent?.setData(slicedDataAll);
+  }
+
   updateParameters() {
     const layersElements = this.getAllParameters();
     const layers = this.origin.layers;
-    const slicedDataAll = [];
 
     for (let i = 0; i < layers.length; i++) {
-      const { weights, biases, z, outputs } = layers[i];
-      if (this.calculationComponent) {
-        slicedDataAll.push(this.sliceData(weights, biases, z, outputs));
-      }
+      const { weights, biases, outputs } = layers[i];
+
       const { lines, neurons } = layersElements[i];
       for (let i = 0; i < neurons.length; i++) {
         neurons[i].setOutput(outputs.data[0][i]);
@@ -203,7 +210,6 @@ class MlpView extends Playable {
         }
       }
     }
-    this.calculationComponent?.setData(slicedDataAll);
   }
 
   setOrigin(obj) {
diff --git a/js/Draw/MLP/Playable.js b/js/Draw/MLP/Playable.js
index d8eff44..eb88bdd 100644
--- a/js/Draw/MLP/Playable.js
+++ b/js/Draw/MLP/Playable.js
@@ -308,7 +308,9 @@ class Playable extends Draggable {
       [parseInt(this.batchSize), this.getInput().shape[1]],
     );
     const outputData = this.getOutput()?.getData() ?? null;
-    await this.origin.trainOneStep(inputData, outputData);
+    const mlp_output = await this.origin.forward(inputData);
+    this.calculationComponent && this.setCalculationData();
+    await this.origin.backward(mlp_output, outputData);
     this.updateParameters();
     this.graphComponent?.setData(this.origin.getLossData());
     this.setMsPerStepText(performance.now() - startTime + "ms / step");
diff --git a/js/MLP/Mlp.js b/js/MLP/Mlp.js
index 6a01b78..57fba82 100644
--- a/js/MLP/Mlp.js
+++ b/js/MLP/Mlp.js
@@ -14,13 +14,6 @@ class MLP extends MlpParams {
     }
   }
 
-  forward(inputs) {
-    for (let i = 0; i < this.layers.length; i++) {
-      inputs = this.layers[i].forward(inputs);
-    }
-    return inputs;
-  }
-
   getParameters() {
     const allWeights = [];
     const allBiases = [];
@@ -50,10 +43,16 @@ class MLP extends MlpParams {
     });
   }
 
-  trainOneStep(x_batch, y_batch) {
-    const mlp_output = this.forward(new Tensor(x_batch));
-    if (this.mode == "eval") return;
+  forward(x_batch) {
+    let inputs = new Tensor(x_batch);
+    for (let i = 0; i < this.layers.length; i++) {
+      inputs = this.layers[i].forward(inputs);
+    }
+    return inputs;
+  }
 
+  backward(mlp_output, y_batch) {
+    if (this.mode == "eval") return;
     const loss = errFuncManager.getFunction(this.errFunc)(
       mlp_output,
       new Tensor(y_batch),