31:T42f,// CELL 5: Showing the AI's Verdict // ============================================================================= // AUDIENCE POINT: "Finally, the program gives us its final grade! It looks at // the context of our words and tells us whether the overall feeling is // POSITIVE or NEGATIVE, along with a percentage showing how sure it is." html`

Sentiment Verdict: ${result[0].label}

Confidence Level: ${(result[0].score * 100).toFixed( 1 )}% Sure

`32:[{"id":"02f911dc105efd55700a63a60cca9600aad2d08df0e2ef8319b6b843c6bcfc19acc6f3ee25131b85557eb24ce787b12be8d01dbfdcba16aaa455b6ef84e9f8a3","href":"https://static.observableusercontent.com/files/02f911dc105efd55700a63a60cca9600aad2d08df0e2ef8319b6b843c6bcfc19acc6f3ee25131b85557eb24ce787b12be8d01dbfdcba16aaa455b6ef84e9f8a3","name":"image.png","type":"image/png","lastModified":1785691377898.207,"size":15074}] 2f:{"title":"Hello Hugging Face ","cells":[{"id":0,"value":"# Hello Hugging Face

","pinned":false,"hidden":false,"mode":"md"},{"id":75,"value":"{\n const t = md`\n## Sentiment Analysis\nMany of you are familiar with autocorrect on your smartphones—how it tries to predict the next word you type. Imagine taking that concept, but instead of teaching it a few words, you gave it every book, article, encyclopedia, and conversation ever written on the internet.\n\nThat is a Transformer model. It’s not a physical robot; it's a very clever computer program that reads human language, understands context, and can tell whether a sentence sounds happy, sad, positive, or negative.\n\nThis notebook leverages a Transformer hosted on the [Hugging Face Cloud](https://huggingface.co/) to demonstrate 100% client-side sentiment analysis.\n\nThe notebook's workflow is as follows: \n\n1. [Fetch Code](https://www.jsdelivr.com/package/npm/@huggingface/transformers) ➔ 2. Download Model ➔ 3. Process Locally ➔ 4. Display Result\n`;\n\n t.querySelectorAll(\"a\").forEach((n) => (n.target = \"_blank\"));\n return t;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":105,"value":"{\n const t = md`\n

1. Fetch Code: (click for details)

\n
\n**Loading the Engine**\n * **What happens:** Your browser connects to the internet to grab [\\`Transformers.js\\`]( https://www.jsdelivr.com/package/npm/@huggingface/transformers)—a small, free software toolkit.\n

\n * **Analogy:** *Think of this step like bringing a small dictionary into the room. We aren't asking a supercomputer in another state to do the work; we are just downloading the instructions so our own laptop can do the thinking.*\n

`;\n t.querySelectorAll(\"a\").forEach((n) => (n.target = \"_blank\"));\n return t;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":7,"value":"// CELL 1: Importing the AI Engine from the Cloud\n// =============================================================================\n// Start by bringing a small, free software toolkit into\n// the web browser over the internet.\n//\n// WHAT THIS DOES:\n// Downloads `Transformers.js`, the open-source JavaScript library built by\n// Hugging Face that lets web browsers run AI models locally.\n// More details at: https://www.jsdelivr.com/package/npm/@huggingface/transformers\ntransformers = import(\n \"https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.0.0\"\n)","pinned":false,"hidden":false,"mode":"ojs"},{"id":110,"value":"{\n const t = md`\n

2. Download Model: (click for details)

\n
\n**Fetching the Model Weights**\n * **What happens:** The computer contacts Hugging Face (the open library where thousands of AI programs live) and downloads a specific model called DistilBERT (~250 megabytes).

\n * **Where it’s hosted:**\n * Originally, it lives on Hugging Face's servers (a massive open digital warehouse).\n * Once loaded, it caches directly into your laptop’s web browser memory.\n

\n * **Analogy:** *Hugging Face is like the central public library for AI models. We are checking out a small, specialized book on 'Sentiment Analysis' and putting it on our laptop's desk. Once it's downloaded, we don't need the internet anymore.*\n

`;\n return t;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":11,"value":"// CELL 2: Checking Out the AI Model from the Hugging Face Public Library\n// =============================================================================\n// Hugging Face is like a giant global library for AI models.\n// We are checking out a small model named 'DistilBERT' and putting it onto\n// our laptop's desk. Once downloaded (~250 MB), it stays stored right here in\n// your browser—no subscriptions or supercomputers required.\nclassifier = {\n // 1. Unpack the pipeline tool from the imported library in Cell 1\n const { pipeline } = await transformers;\n\n // 2. Download and load the pre-trained 'DistilBERT' sentiment model.\n // The browser caches these weights locally so it only downloads once.\n return await pipeline(\n \"sentiment-analysis\",\n \"Xenova/distilbert-base-uncased-finetuned-sst-2-english\"\n );\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":113,"value":"{\n const t = md`\n

3. Process Locally: (click for details)

\n
\n**Local \"On-Device\" Thinking (Cells 3 & 4)**\n * **What happens:** You type a sentence into the input box. The model converts your text into numbers, compares those numbers against what it learned during training, and calculates a statistical probability.\n

\n * **Key Point (Privacy & Speed):** *Notice how fast the answer appears when you type? Your sentence never leaves this laptop. It’s not being sent to Google, Microsoft, or any cloud server. The brain power is happening entirely right here inside this web browser page.*\n

`;\n return t;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":14,"value":"// CELL 3: The Interactive Text Box\n// =============================================================================\nviewof userText = Inputs.textarea({\n label: \"Type a sentence:\",\n value: \"The volunteers at our retirement community are dedicated.\",\n placeholder: \"Type something happy or sad...\",\n width: \"100%\" // Uses Observable's native width option to fill the screen\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":19,"value":"// CELL 4: Local \"On-Device\" Inference (Privacy-First)\n// =============================================================================\n// WHAT THIS DOES:\n// Automatically re-runs the sentiment calculation every single time `userText`\n// changes, converting words into statistical probabilities.\n//\n// Notice how instantly the answer updates when we change a word?\n// That is because your text NEVER leaves this laptop. It isn't being sent\n// to Google or Microsoft. The math is happening 100% privately inside this browser.\nresult = await classifier(userText)","pinned":false,"hidden":false,"mode":"ojs"},{"id":117,"value":"{\n const t = md`\n

4. Dispaly Result: (click for details)

\n
\n**Showing the Answer (Cell 5)**\n * **What happens:** The mathematical score is converted into human language: a colored box (Green for positive, Red for negative) and a percentage score showing how confident the AI is.\n

\n * **Analogy:** *The program gives us its final grade. If I type 'The weather today is lovely,' it reads 'lovely' in context and says, 'I am 99.8% confident this is a POSITIVE statement.'*\n

`;\n return t;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":21,"value":"$31","pinned":false,"hidden":false,"mode":"ojs"}],"files":"$W32","stdlib":"1","titling":"auto"}