34:["$","title",null,{"children":"Fashion-mnist (Convolutional Neural Network) | Rithwik Anand | Observable"}] 3c:[{"id":"875cf7ad5ed1d1506ef3a2c1f5f198f0e05efd3de13743556ebc8849ba0c898f003e262d5fe63fdaadeb0f39eacf7971db44c0cfeed48b99d2feca4368ffc3fd","href":"https://static.observableusercontent.com/files/875cf7ad5ed1d1506ef3a2c1f5f198f0e05efd3de13743556ebc8849ba0c898f003e262d5fe63fdaadeb0f39eacf7971db44c0cfeed48b99d2feca4368ffc3fd","name":"model.json","type":"application/json","lastModified":1684377422957.315,"size":3244},{"id":"6d82047114837dac33ddf8b5217c6360a26227a05c890a8386e3e86219aea963e403e30ebaed405350396166ddb7a8506862c69e20853ea435c73d46743055fc","href":"https://static.observableusercontent.com/files/6d82047114837dac33ddf8b5217c6360a26227a05c890a8386e3e86219aea963e403e30ebaed405350396166ddb7a8506862c69e20853ea435c73d46743055fc","name":"weights.bin","type":"application/octet-stream","lastModified":1684377426152.125,"size":827688}] 38:{"id":"6279137529d755ba","slug":"fashion-mnist-convolutional-neural-network","title":"Fashion-mnist (Convolutional Neural Network)","thumbnail":"c77d2490ce6ee235ad23770012617d234fc00b48c509113c45675d91129baed5","access_level":"public","create_time":"$D2023-05-18T02:30:24.801Z","update_time":"$D2023-05-18T02:39:47.261Z","view_time":null,"first_public_time":"$D2023-05-18T02:30:24.801Z","has_unpublished_edits":null,"fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"77b49d5139b68ed8","login":"rithwikanand","name":"Rithwik Anand","avatar_url":"https://avatars-next.observableusercontent.com/2d11b8fc94986ebb4c9233d528cbb27cc8b80f1263a7ad00b479736ba2da942f","tier":"starter_2024","bio":"","followed_by_you":false,"follows_you":false},"version":40,"publish_version":null,"role":null,"authored":false,"authors":[{"type":"individual","id":"2e0717d814773a4a","login":"rithwikanand","name":"Rithwik Anand","avatar_url":"https://avatars-next.observableusercontent.com/2d11b8fc94986ebb4c9233d528cbb27cc8b80f1263a7ad00b479736ba2da942f","tier":null,"bio":"","followed_by_you":false,"follows_you":false}],"members":[{"type":"individual","id":"2e0717d814773a4a","login":"rithwikanand","name":"Rithwik Anand","avatar_url":"https://avatars-next.observableusercontent.com/2d11b8fc94986ebb4c9233d528cbb27cc8b80f1263a7ad00b479736ba2da942f","tier":null,"role":"editor"}],"comments":[],"custom_thumbnail":true,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":null,"type":"notebook","body":{"titling":"auto","title":"Fashion-mnist (Convolutional Neural Network)","cells":[{"id":0,"value":"# Fashion-mnist (Convolutional Neural Network)","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"","database":"$undefined"},{"id":5,"value":"result_div = {\n\twhile(true) {\n\t\tconst ix = ~~(Math.random() * training_data.inputs.length)\n\t\tconst input = training_data.inputs[ix]\n\t\tconst inputT = tf.tidy(() => {\n\t\t\tconst input_ = tf.tensor(input, [28, 28], 'int32')\n\t\t\tconst normalized = normalize(input_, 0, 255)\n\t\t\treturn normalized.reshape([1, 28, 28, 1])\n\t\t})\n\t\tconst prediction = tf.tidy(() => {\n\t\t\tconst prediction = model1.predict(inputT)\n\t\t\treturn prediction.squeeze().argMax()\n\t\t})\n\t\tyield result(ix, (await prediction.data())[0])\n\t\tawait Promises.delay(2000)\n\t}\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":7,"value":"training_data = {\n\tconst { TRAINING_DATA } = await import('https://storage.googleapis.com/jmstore/TensorFlowJS/EdX/TrainingData/fashion-mnist.js')\n\ttf.util.shuffleCombo(TRAINING_DATA.inputs, TRAINING_DATA.outputs)\n\treturn TRAINING_DATA\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":9,"value":"tf = require('@tensorflow/tfjs')","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":11,"value":"classes = ['T-shirt', 'Trouser', 'Pullover', 'Dress', 'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":13,"value":"function normalize(inputTensor, min, max) {\n\treturn tf.tidy(() => {\n\t\tconst range = max - min\n\t\treturn inputTensor.sub(min).div(range)\n\t})\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":15,"value":"tensors = {\n\tconst inputs = tf.tensor(training_data.inputs, [training_data.inputs.length, 28, 28])\n\tconst outputs = tf.tensor(training_data.outputs, [training_data.outputs.length])\n\tconst normalizedInputs = normalize(inputs, 0, 255);\n\tyield { inputs, normalizedInputs, outputs }\t\n\ttry {\n\t\tyield invalidation\n\t} finally {\n\t\tinputs.dispose()\n\t\tnormalizedInputs.dispose()\n\t\toutputs.dispose()\n\t}\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":17,"value":"function result(training_index, prediction) {\n\tconst input = training_data.inputs[training_index]\n\tconst output = training_data.outputs[training_index]\n\n\treturn html`\n\t\t

\n\t\t

${drawImage(input)}

\n\t\t

${classes[output]}

\n\t\t

${output === prediction ? '===' : '!=='}

\n\t\t

${classes[prediction]}

\n\t\t

${output === prediction ? '✅' : '⛔️'}

\n\t\t

\n\t\t\n\t\t`\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":19,"value":"function drawImage(arr) {\n\tconst ctx = DOM.context2d(14, 14);\n\n\tconst imageData = ctx.getImageData(0, 0, 28, 28)\n\tfor(let i in arr) {\n\t\timageData.data[i * 4] = arr[i]\n\t\timageData.data[i * 4 + 1] = arr[i]\n\t\timageData.data[i * 4 + 2] = arr[i]\n\t\timageData.data[i * 4 + 3] = 255\n\t}\n\tctx.putImageData(imageData, 0, 0)\n\treturn ctx.canvas\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":21,"value":"model = {\n\tconst model = tf.sequential({\n\t\tlayers: [\n\t\t\ttf.layers.conv2d({\n\t\t\t\tinputShape: [28, 28, 1],\n\t\t\t\tfilters: 16,\n\t\t\t\tkernelSize: 3,\n\t\t\t\tstrides: 1,\n\t\t\t\tpadding: 'same',\n\t\t\t\tactivation: 'relu',\n\t\t\t}),\n\t\t\ttf.layers.maxPooling2d({\n\t\t\t\tpoolSize: 2,\n\t\t\t\tstride: 2,\n\t\t\t}),\n\t\t\ttf.layers.conv2d({\n\t\t\t\tfilters: 32,\n\t\t\t\tkernelSize: 3,\n\t\t\t\tstrides: 1,\n\t\t\t\tpadding: 'same',\n\t\t\t\tactivation: 'relu',\n\t\t\t}),\n\t\t\ttf.layers.maxPooling2d({\n\t\t\t\tpoolSize: 2,\n\t\t\t\tstride: 2,\n\t\t\t}),\n\t\t\ttf.layers.flatten(),\n\t\t\ttf.layers.dense({\n\t\t\t\tunits: 128,\n\t\t\t\tactivation: 'relu'\n\t\t\t}),\n\t\t\ttf.layers.dense({\n\t\t\t\tunits: 10,\n\t\t\t\tactivation: 'softmax'\n\t\t\t})\n\t\t]\n\t})\n\tmodel.summary()\n\tyield model\n\ttry {\n\t\tyield invalidation\n\t} finally {\n\t\tmodel.dispose()\n\t}\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":23,"value":"files = new Map(await Promise.all([\n\tFileAttachment(\"model.json\"),\n\tFileAttachment(\"weights.bin\")\n].map(async f => [f.name, await f.url()])))","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":25,"value":"model1 = await tf.loadLayersModel(\"/model.json\", {\n\tfetchFunc: (url, ...options) => fetch(files.get(url.slice(1)), ...options)\n})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W3c","stdlib":"1"}} 33:null