@epfml/discojs 3.0.1-p20241107104659.0 → 3.0.1-p20241203151748.0

package/dist/client/client.js

@@ -149,6 +149,8 @@ export class Client extends EventEmitter {
         }
         url.pathname += `tasks/${this.task.id}/model.json`;
         const response = await fetch(url);
+        if (!response.ok)
+            throw new Error(`fetch: HTTP status ${response.status}`);
         const encoded = new Uint8Array(await response.arrayBuffer());
         return await serialization.model.decode(encoded);
     }

package/dist/client/federated/federated_client.js

@@ -2,7 +2,7 @@ import createDebug from "debug";
 import { serialization } from "../../index.js";
 import { Client, shortenId } from "../client.js";
 import { type } from "../messages.js";
-import { waitMessage, waitMessageWithTimeout, WebSocketServer, } from "../event_connection.js";
+import { waitMessage, WebSocketServer, } from "../event_connection.js";
 import * as messages from "./messages.js";
 const debug = createDebug("discojs:client:federated");
 /**
@@ -53,7 +53,7 @@ export class FederatedClient extends Client {
             type: type.ClientConnected,
         };
         this.server.send(msg);
-        const { id, waitForMoreParticipants, payload, round, nbOfParticipants } = await waitMessageWithTimeout(this.server, type.NewFederatedNodeInfo);
+        const { id, waitForMoreParticipants, payload, round, nbOfParticipants } = await waitMessage(this.server, type.NewFederatedNodeInfo);
         // This should come right after receiving the message to make sure
         // we don't miss a subsequent message from the server
         // We check if the server is telling us to wait for more participants

package/dist/default_tasks/index.d.ts

@@ -4,3 +4,4 @@ export { mnist } from './mnist.js';
 export { simpleFace } from './simple_face.js';
 export { titanic } from './titanic.js';
 export { wikitext } from './wikitext.js';
+export { tinderDog } from './tinder_dog.js';

package/dist/default_tasks/index.js

@@ -4,3 +4,4 @@ export { mnist } from './mnist.js';
 export { simpleFace } from './simple_face.js';
 export { titanic } from './titanic.js';
 export { wikitext } from './wikitext.js';
+export { tinderDog } from './tinder_dog.js';

package/dist/default_tasks/tinder_dog.d.ts

@@ -0,0 +1,2 @@
+import type { TaskProvider } from '../index.js';
+export declare const tinderDog: TaskProvider<'image'>;

package/dist/default_tasks/tinder_dog.js

@@ -0,0 +1,72 @@
+import * as tf from '@tensorflow/tfjs';
+import { models } from '../index.js';
+export const tinderDog = {
+    getTask() {
+        return {
+            id: 'tinder_dog',
+            displayInformation: {
+                taskTitle: 'GDHF 2024 | TinderDog',
+                summary: {
+                    preview: 'Which dog is the cutest....or not?',
+                    overview: "Binary classification model for dog cuteness."
+                },
+                model: 'The model is a simple Convolutional Neural Network composed of two convolutional layers with ReLU activations and max pooling layers, followed by a fully connected output layer. The data preprocessing reshapes images into 64x64 pixels and normalizes values between 0 and 1',
+                dataFormatInformation: 'Accepted image formats are .png .jpg and .jpeg.',
+                dataExampleText: '',
+                dataExampleImage: 'https://storage.googleapis.com/deai-313515.appspot.com/tinder_dog_preview.png',
+                sampleDatasetLink: 'https://storage.googleapis.com/deai-313515.appspot.com/tinder_dog.zip',
+                sampleDatasetInstructions: 'Opening the link should start downloading a zip file which you can unzip. To connect the data, pick one of the data splits (the folder 0 for example) and use the CSV option below to select the file named "labels.csv". You can now connect the images located in the same folder.'
+            },
+            trainingInformation: {
+                epochs: 10,
+                roundDuration: 2,
+                validationSplit: 0, // nicer plot for GDHF demo
+                batchSize: 10,
+                dataType: 'image',
+                IMAGE_H: 64,
+                IMAGE_W: 64,
+                LABEL_LIST: ['Cute dogs', 'Less cute dogs'],
+                scheme: 'federated',
+                aggregationStrategy: 'mean',
+                minNbOfParticipants: 3,
+                tensorBackend: 'tfjs'
+            }
+        };
+    },
+    async getModel() {
+        const seed = 42; // set a seed to ensure reproducibility during GDHF demo
+        const imageHeight = this.getTask().trainingInformation.IMAGE_H;
+        const imageWidth = this.getTask().trainingInformation.IMAGE_W;
+        const imageChannels = 3;
+        const model = tf.sequential();
+        model.add(tf.layers.conv2d({
+            inputShape: [imageHeight, imageWidth, imageChannels],
+            kernelSize: 5,
+            filters: 8,
+            activation: 'relu',
+            kernelInitializer: tf.initializers.heNormal({ seed })
+        }));
+        model.add(tf.layers.conv2d({
+            kernelSize: 5, filters: 16, activation: 'relu',
+            kernelInitializer: tf.initializers.heNormal({ seed })
+        }));
+        model.add(tf.layers.maxPooling2d({ poolSize: 2, strides: 2 }));
+        model.add(tf.layers.dropout({ rate: 0.25, seed }));
+        model.add(tf.layers.flatten());
+        model.add(tf.layers.dense({
+            units: 32, activation: 'relu',
+            kernelInitializer: tf.initializers.heNormal({ seed })
+        }));
+        model.add(tf.layers.dropout({ rate: 0.25, seed }));
+        model.add(tf.layers.dense({
+            units: 2, activation: 'softmax',
+            kernelInitializer: tf.initializers.heNormal({ seed })
+        }));
+        model.compile({
+            optimizer: tf.train.adam(0.0005),
+            loss: 'categoricalCrossentropy',
+            metrics: ['accuracy']
+        });
+        return Promise.resolve(new models.TFJS('image', model));
+    }
+};

package/dist/task/task_handler.js

@@ -9,7 +9,7 @@ function urlToTasks(base) {
     return ret;
 }
 export async function pushTask(base, task, model) {
-    await fetch(urlToTasks(base), {
+    const response = await fetch(urlToTasks(base), {
         method: "POST",
         body: JSON.stringify({
             task,
@@ -17,9 +17,13 @@ export async function pushTask(base, task, model) {
             weights: await serialization.weights.encode(model.weights),
         }),
     });
+    if (!response.ok)
+        throw new Error(`fetch: HTTP status ${response.status}`);
 }
 export async function fetchTasks(base) {
     const response = await fetch(urlToTasks(base));
+    if (!response.ok)
+        throw new Error(`fetch: HTTP status ${response.status}`);
     const tasks = await response.json();
     if (!Array.isArray(tasks)) {
         throw new Error("Expected to receive an array of Tasks when fetching tasks");

package/dist/training/disco.js

@@ -134,10 +134,13 @@ export class Disco extends EventEmitter {
     }
     async #preprocessSplitAndBatch(dataset) {
         const { batchSize, validationSplit } = this.#task.trainingInformation;
-        const preprocessed = await processing.preprocess(this.#task, dataset);
-        const [training, validation] = (this.#preprocessOnce
+        let preprocessed = await processing.preprocess(this.#task, dataset);
+        preprocessed = (this.#preprocessOnce
             ? new Dataset(await arrayFromAsync(preprocessed))
-            : preprocessed).split(validationSplit);
+            : preprocessed);
+        if (validationSplit === 0)
+            return [preprocessed.batch(batchSize).cached(), undefined];
+        const [training, validation] = preprocessed.split(validationSplit);
         return [
             training.batch(batchSize).cached(),
             validation.batch(batchSize).cached(),

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@epfml/discojs",
-  "version": "3.0.1-p20241107104659.0",
+  "version": "3.0.1-p20241203151748.0",
   "type": "module",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",