{
"cells": [
{
"cell_type": "markdown",
"id": "a799e7e4",
"metadata": {},
"source": [
"# Day 1 โ Meet the Machines ๐ค\n",
"\n",
"> **Mission briefing:** You passed the entrance exam, so here's your reward: a walk through the Lab's equipment room. Four finished AI machines are humming on the benches โ one that **sees**, one that **reads moods**, one that **writes stories**, and a **sorting hat** that files text into any categories you invent. Today you just get to *play* with them. There's a catch, though: by Demo Day you will have **rebuilt every single one yourself**.\n",
"\n",
"**Previously in the Lab:** you were the Data Detective โ you can read data. Now meet the machines that devour it.\n",
"\n",
"**Today you will:**\n",
"- See the **map of AI** โ how machine learning, deep learning, and the \"superpowers\" fit together\n",
"- Run four **pretrained** models: image recognition, sentiment, text generation, zero-shot sorting\n",
"- Try to **fool** them, bend them, and break them (that's how you learn what they can't do)\n",
"- Meet the **AI Studio** โ the app you'll assemble one tab per day\n",
"\n",
"*These models were trained on giant datasets using enormous computers. You won't train them today โ you'll borrow them, the way real researchers stand on each other's shoulders.*\n",
"\n",
"A precise reading
This course focuses on learning from data. AI also includes rule-based and search systems. Tables are a useful starting point, but images, audio and text need representations suited to their structure.
\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "c35de5c7",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:27.659739Z",
"iopub.status.busy": "2026-08-17T12:57:27.659547Z",
"iopub.status.idle": "2026-08-17T12:57:27.768636Z",
"shell.execute_reply": "2026-08-17T12:57:27.767614Z"
}
},
"outputs": [],
"source": [
"# โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ\n",
"# โ SETUP โ run me first! (identical in every Lab notebook) โ\n",
"# โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ\n",
"import os, sys, pathlib\n",
"\n",
"IN_COLAB = \"google.colab\" in sys.modules\n",
"SMOKE = os.environ.get(\"SMOKE_TEST\", \"0\") == \"1\" # tiny fast run for automated tests\n",
"\n",
"REPO_URL = \"https://github.com/A-Halimi/ai-studio-internship.git\" # instructor: set once\n",
"\n",
"if IN_COLAB:\n",
" if not pathlib.Path(\"/content/ai-studio-internship\").exists():\n",
" !git clone -q {REPO_URL} /content/ai-studio-internship\n",
" %cd /content/ai-studio-internship/notebooks/day01\n",
" %pip -q install transformers==5.13.0\n",
"\n",
"HERE = pathlib.Path.cwd()\n",
"REPO = HERE.parents[1] # .../notebooks/dayNN -> repo root\n",
"DATA_DIR = pathlib.Path(os.environ.get(\"COURSE_DATA_DIR\", REPO / \"data\"))\n",
"SAMPLES_DIR = REPO / \"data_samples\" # small datasets shipped with the repo\n",
"MODELS_DIR = REPO / \"ai_studio\" / \"models\" # where your Studio modules unlock\n",
"for p in (REPO / \"data\", MODELS_DIR):\n",
" p.mkdir(parents=True, exist_ok=True)\n",
"\n",
"import random\n",
"import numpy as np\n",
"SEED = 42\n",
"random.seed(SEED); np.random.seed(SEED)\n",
"print(f\"โ Setup done | Colab: {IN_COLAB} | Smoke test: {SMOKE}\")\n",
"print(f\" data: {DATA_DIR}\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "73d011c6",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:27.770417Z",
"iopub.status.busy": "2026-08-17T12:57:27.770224Z",
"iopub.status.idle": "2026-08-17T12:57:29.046964Z",
"shell.execute_reply": "2026-08-17T12:57:29.045840Z"
}
},
"outputs": [],
"source": [
"import torch\n",
"torch.manual_seed(SEED)\n",
"DEVICE = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
"if DEVICE.type == \"cuda\":\n",
" print(f\"โก Running on GPU: {torch.cuda.get_device_name(0)}\")\n",
"else:\n",
" print(\"๐ข No GPU found โ running on CPU. Everything still works, just slower.\")"
]
},
{
"cell_type": "markdown",
"id": "fb3ce425",
"metadata": {},
"source": [
"## ๐บ๏ธ The map of AI\n",
"\n",
"People throw around \"AI\", \"machine learning\", and \"deep learning\" as if they mean the same thing. They don't โ they **nest**, like Russian dolls:\n",
"\n",
"- **Artificial Intelligence** โ any program that does something we'd call \"smart.\"\n",
"- **Machine Learning** โ a *kind* of AI that isn't hand-coded with rules; it **learns patterns from data**. (Yesterday you saw the patterns; ML is teaching a machine to find them.)\n",
"- **Deep Learning** โ machine learning with **neural networks** stacked many layers deep. It powers almost every jaw-dropping AI you've heard of.\n",
"\n",
"And there are **three ways** a machine can learn:\n",
"\n",
"- **Supervised** โ learn from *labeled* examples (\"this penguin is a Gentoo\"). โ Day 2\n",
"- **Unsupervised** โ find structure with *no* labels (\"these things clump together\"). โ Day 3\n",
"- **Reinforcement** โ learn by *trial, reward, and error*, like training a dog. โ Day 9\n",
"\n",
"Run the next cell for the map."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "20a8c861",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:29.048767Z",
"iopub.status.busy": "2026-08-17T12:57:29.048553Z",
"iopub.status.idle": "2026-08-17T12:57:29.458476Z",
"shell.execute_reply": "2026-08-17T12:57:29.457443Z"
}
},
"outputs": [],
"source": [
"import matplotlib.pyplot as plt\n",
"from matplotlib.patches import FancyBboxPatch\n",
"\n",
"fig, (axL, axR) = plt.subplots(1, 2, figsize=(11, 5))\n",
"\n",
"# Left: AI contains ML contains Deep Learning (nested boxes)\n",
"axL.set_xlim(0, 10); axL.set_ylim(0, 10); axL.axis(\"off\")\n",
"for x, y, w, h, label, color, fs in [\n",
" (0.5, 0.5, 9.0, 9.0, \"Artificial Intelligence\", \"#e8f0fe\", 13),\n",
" (1.6, 1.6, 6.8, 6.3, \"Machine Learning\", \"#c6dafc\", 12),\n",
" (2.8, 2.6, 4.4, 3.9, \"Deep Learning\", \"#8ab4f8\", 12),\n",
"]:\n",
" axL.add_patch(FancyBboxPatch((x, y), w, h, boxstyle=\"round,pad=0.02\",\n",
" fc=color, ec=\"#222\", lw=1.5))\n",
" axL.text(x + w / 2, y + h - 0.5, label, ha=\"center\", va=\"top\",\n",
" fontsize=fs, fontweight=\"bold\")\n",
"axL.text(5.0, 3.9, \"neural networks\\nwith many layers\", ha=\"center\", va=\"center\",\n",
" fontsize=10, style=\"italic\")\n",
"axL.set_title(\"AI โ Machine Learning โ Deep Learning\", fontweight=\"bold\")\n",
"\n",
"# Right: the three ways to learn + two superpowers\n",
"axR.set_xlim(0, 10); axR.set_ylim(0, 10); axR.axis(\"off\")\n",
"axR.set_title(\"How machines learn\", fontweight=\"bold\")\n",
"for i, (name, desc, color) in enumerate([\n",
" (\"Supervised\", \"learn from labeled examples\", \"#a5d6a7\"),\n",
" (\"Unsupervised\", \"find structure, no labels\", \"#fff59d\"),\n",
" (\"Reinforcement\", \"learn by trial and reward\", \"#ef9a9a\"),\n",
"]):\n",
" y = 8.2 - i * 2.3\n",
" axR.add_patch(FancyBboxPatch((0.5, y - 0.9), 9.0, 1.6, boxstyle=\"round,pad=0.02\",\n",
" fc=color, ec=\"#222\"))\n",
" axR.text(5.0, y + 0.15, name, ha=\"center\", fontsize=12, fontweight=\"bold\")\n",
" axR.text(5.0, y - 0.5, desc, ha=\"center\", fontsize=9)\n",
"axR.text(5.0, 0.6, \"Two superpowers you'll build: Vision (images) + Language (text)\",\n",
" ha=\"center\", fontsize=10, fontweight=\"bold\")\n",
"\n",
"plt.tight_layout()\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"id": "d529be71",
"metadata": {},
"source": [
"## ๐๏ธ Superpower 1 โ Vision\n",
"\n",
"First up: a machine that **looks at a photo and names what's in it**. Ours is **ResNet-18**, a neural network trained on **ImageNet** โ 1.2 million photos across 1000 categories (dogs, castles, teapots, you name it).\n",
"\n",
"We don't have to train it โ we just **download the finished brain** and use it. Two details matter:\n",
"\n",
"- `.eval()` puts the network in \"just answer, don't learn\" mode.\n",
"- `weights.transforms()` is the exact **preprocessing** (resize, crop, normalize) the model expects. Feed it anything else and it gets confused โ models are picky eaters."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "1e950c16",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:29.460203Z",
"iopub.status.busy": "2026-08-17T12:57:29.460003Z",
"iopub.status.idle": "2026-08-17T12:57:33.899411Z",
"shell.execute_reply": "2026-08-17T12:57:33.898273Z"
}
},
"outputs": [],
"source": [
"from torchvision.models import resnet18, ResNet18_Weights\n",
"from sklearn.datasets import load_sample_image\n",
"from PIL import Image\n",
"\n",
"weights = ResNet18_Weights.DEFAULT\n",
"model = resnet18(weights=weights).eval().to(DEVICE)\n",
"preprocess = weights.transforms() # the model's required input pipeline\n",
"categories = weights.meta[\"categories\"] # the 1000 label names\n",
"print(f\"ResNet-18 ready โ it knows {len(categories)} categories.\")\n",
"\n",
"\n",
"def classify_top5(img_array):\n",
" \"\"\"Return the model's 5 most confident labels for one image (a NumPy array).\"\"\"\n",
" img = Image.fromarray(img_array)\n",
" batch = preprocess(img).unsqueeze(0).to(DEVICE) # 1 image -> a batch of 1\n",
" with torch.no_grad():\n",
" logits = model(batch)\n",
" probs = torch.softmax(logits, dim=1)[0] # turn scores into probabilities\n",
" top_p, top_i = probs.topk(5)\n",
" return [(categories[i], p.item()) for i, p in zip(top_i, top_p)]\n",
"\n",
"\n",
"def show_top5(img_array, title):\n",
" preds = classify_top5(img_array)\n",
" labels = [name for name, _ in preds][::-1]\n",
" scores = [p for _, p in preds][::-1]\n",
" fig, (ax_img, ax_bar) = plt.subplots(1, 2, figsize=(10, 4))\n",
" ax_img.imshow(img_array); ax_img.axis(\"off\"); ax_img.set_title(title)\n",
" ax_bar.barh(labels, scores)\n",
" ax_bar.set_xlim(0, 1)\n",
" ax_bar.set_xlabel(\"Confidence\")\n",
" ax_bar.set_title(\"Top-5 guesses\")\n",
" plt.tight_layout(); plt.show()\n",
" return preds"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "7fb049df",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:33.901661Z",
"iopub.status.busy": "2026-08-17T12:57:33.901352Z",
"iopub.status.idle": "2026-08-17T12:57:34.482535Z",
"shell.execute_reply": "2026-08-17T12:57:34.481386Z"
}
},
"outputs": [],
"source": [
"china = load_sample_image(\"china.jpg\") # a photo that ships with scikit-learn\n",
"_ = show_top5(china, \"china.jpg\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8d24c421",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:34.484664Z",
"iopub.status.busy": "2026-08-17T12:57:34.484434Z",
"iopub.status.idle": "2026-08-17T12:57:34.571197Z",
"shell.execute_reply": "2026-08-17T12:57:34.570182Z"
}
},
"outputs": [],
"source": [
"flower = load_sample_image(\"flower.jpg\")\n",
"_ = show_top5(flower, \"flower.jpg\")"
]
},
{
"cell_type": "markdown",
"id": "c8fa4311",
"metadata": {},
"source": [
"### ๐ฌ Your turn (mini-task)\n",
"\n",
"Look at the two bar charts. **In a sentence or two, jot down what the model got right and what it got wrong.** Was it confidently correct? Confidently *wrong*? Notice that it never says \"I don't know\" โ it always spreads its bet across the labels it was trained on, even for a photo that matches none of them well. Keeping models honest about their own uncertainty is a real research problem."
]
},
{
"cell_type": "markdown",
"id": "b8a7f2bc",
"metadata": {},
"source": [
"## ๐ฌ Superpower 2 โ Reading the mood\n",
"\n",
"Next machine: **sentiment analysis**. It reads a sentence and decides whether the feeling behind it is **positive** or **negative**, with a confidence score.\n",
"\n",
"The `transformers` library gives us a one-liner called a **pipeline**: name the task, name the model, and it downloads and wires everything up. Ours is a **DistilBERT** model fine-tuned on movie reviews."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "d038c90f",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:34.573468Z",
"iopub.status.busy": "2026-08-17T12:57:34.573320Z",
"iopub.status.idle": "2026-08-17T12:57:38.494008Z",
"shell.execute_reply": "2026-08-17T12:57:38.492840Z"
}
},
"outputs": [],
"source": [
"from transformers import pipeline\n",
"\n",
"mood = pipeline(\"text-classification\",\n",
" model=\"distilbert-base-uncased-finetuned-sst-2-english\",\n",
" device=0 if DEVICE.type == \"cuda\" else -1)\n",
"\n",
"sentences = [\n",
" \"I absolutely loved this movie, what a ride!\",\n",
" \"This is the worst day I have ever had.\",\n",
" \"The food was okay, nothing special.\",\n",
"]\n",
"for s in sentences:\n",
" r = mood(s)[0]\n",
" print(f\"{r['label']:8} ({r['score']:.2f}) {s}\")"
]
},
{
"cell_type": "markdown",
"id": "81fe51d9",
"metadata": {},
"source": [
"### ๐งช Exercise 1 โ Fool the mood reader\n",
"\n",
"These models learn from patterns of words, so they struggle with things humans read instantly โ like **sarcasm**.\n",
"\n",
"- Write a sentence that *sounds* positive word-for-word but really means the opposite, and store it in `sarcastic`.\n",
"- **Goal:** get the model to answer **POSITIVE** for something you clearly meant as a complaint."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "d49c47ec",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:38.496085Z",
"iopub.status.busy": "2026-08-17T12:57:38.495783Z",
"iopub.status.idle": "2026-08-17T12:57:38.504865Z",
"shell.execute_reply": "2026-08-17T12:57:38.503895Z"
}
},
"outputs": [],
"source": [
"# ==================== YOUR CODE HERE ====================\n",
"### HINT: pile on happy words for an unhappy situation\n",
"### HINT: e.g. \"Oh great, my flight got cancelled again. I just LOVE airports.\"\n",
"...\n",
"result = mood(sarcastic)[0]\n",
"print(f\"{result['label']} ({result['score']:.2f}) {sarcastic}\")"
]
},
{
"cell_type": "markdown",
"id": "9ff3c081",
"metadata": {},
"source": [
"## โ๏ธ Superpower 3 โ The story writer\n",
"\n",
"This machine **generates text**. Give **GPT-2** the start of a sentence and it keeps going, one word at a time, each word guessed from everything before it. (Yes โ it's a distant, much smaller ancestor of the models you talk to today.)\n",
"\n",
"A couple of dials:\n",
"\n",
"- `max_new_tokens` โ how many words to add.\n",
"- `temperature` โ the \"creativity\" knob. Low = safe and repetitive, high = wild. We use `0.9` for some spice."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "5334de6a",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:38.506265Z",
"iopub.status.busy": "2026-08-17T12:57:38.506132Z",
"iopub.status.idle": "2026-08-17T12:57:41.245698Z",
"shell.execute_reply": "2026-08-17T12:57:41.244653Z"
}
},
"outputs": [],
"source": [
"storyteller = pipeline(\"text-generation\", model=\"gpt2\",\n",
" device=0 if DEVICE.type == \"cuda\" else -1)\n",
"\n",
"MAX_NEW = 20 if SMOKE else 40\n",
"prompt = \"In the year 3000, the last robot on Earth discovered\"\n",
"torch.manual_seed(SEED) # reproducible demo\n",
"out = storyteller(prompt, max_new_tokens=MAX_NEW, do_sample=True,\n",
" temperature=0.9, pad_token_id=50256)\n",
"print(out[0][\"generated_text\"])"
]
},
{
"cell_type": "markdown",
"id": "e9912811",
"metadata": {},
"source": [
"### ๐งช Exercise 2 โ Be the author\n",
"\n",
"Swap in **your own** prompt and see what the machine dreams up.\n",
"\n",
"- Set `my_prompt` to any opening you like โ a fairy tale, a news headline, a recipe.\n",
"- Run it a few times: because `do_sample=True`, you get a different story every time."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "ec73cea9",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:41.247466Z",
"iopub.status.busy": "2026-08-17T12:57:41.247313Z",
"iopub.status.idle": "2026-08-17T12:57:41.618800Z",
"shell.execute_reply": "2026-08-17T12:57:41.617667Z"
}
},
"outputs": [],
"source": [
"# ==================== YOUR CODE HERE ====================\n",
"### HINT: anything goes โ the funnier the opener, the funnier the result\n",
"...\n",
"torch.manual_seed(SEED) # reproducible demo\n",
"out = storyteller(my_prompt, max_new_tokens=MAX_NEW, do_sample=True,\n",
" temperature=0.9, pad_token_id=50256)\n",
"print(out[0][\"generated_text\"])"
]
},
{
"cell_type": "markdown",
"id": "d9ed7e19",
"metadata": {},
"source": [
"## ๐ฉ Superpower 4 โ The sorting hat\n",
"\n",
"The last machine is the cleverest trick. **Zero-shot classification** sorts text into categories **it was never trained on**. You invent the labels on the spot, and it figures out which one fits best.\n",
"\n",
"\"Zero-shot\" means zero training examples for your specific categories โ it reasons about the *meaning* of the words instead."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "21a987e9",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:41.620589Z",
"iopub.status.busy": "2026-08-17T12:57:41.620445Z",
"iopub.status.idle": "2026-08-17T12:57:42.974349Z",
"shell.execute_reply": "2026-08-17T12:57:42.973259Z"
}
},
"outputs": [],
"source": [
"sorter = pipeline(\"zero-shot-classification\",\n",
" model=\"typeform/distilbert-base-uncased-mnli\",\n",
" device=0 if DEVICE.type == \"cuda\" else -1)\n",
"\n",
"text = \"I spent all weekend building a robot that sorts my socks by color.\"\n",
"labels = [\"technology\", \"sports\", \"cooking\", \"art\"]\n",
"res = sorter(text, candidate_labels=labels)\n",
"for lbl, score in zip(res[\"labels\"], res[\"scores\"]):\n",
" print(f\"{lbl:12} {score:.2f}\")"
]
},
{
"cell_type": "markdown",
"id": "aa7a3fb9",
"metadata": {},
"source": [
"### ๐งช Exercise 3 โ Invent the categories\n",
"\n",
"Give the sorting hat your own text and your own labels.\n",
"\n",
"- Set `my_text` to any sentence, and `my_labels` to 3โ5 categories it might fall into.\n",
"- **Try to trip it up:** a sentence that could belong to two categories at once is the interesting case."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "f50de769",
"metadata": {
"execution": {
"iopub.execute_input": "2026-08-17T12:57:42.975976Z",
"iopub.status.busy": "2026-08-17T12:57:42.975837Z",
"iopub.status.idle": "2026-08-17T12:57:42.998482Z",
"shell.execute_reply": "2026-08-17T12:57:42.997190Z"
}
},
"outputs": [],
"source": [
"my_text = \"The championship match went to overtime and the crowd went wild.\"\n",
"# ==================== YOUR CODE HERE ====================\n",
"### HINT: pick categories the text might plausibly belong to\n",
"...\n",
"res = sorter(my_text, candidate_labels=my_labels)\n",
"print(f\"Text: {my_text}\\n\")\n",
"for lbl, score in zip(res[\"labels\"], res[\"scores\"]):\n",
" print(f\"{lbl:12} {score:.2f}\")"
]
},
{
"cell_type": "markdown",
"id": "07903df6",
"metadata": {},
"source": [
"## ๐งช This is the mission: your AI Studio\n",
"\n",
"Everything you just played with was built by *other* people. Over the next two weeks you'll build your **own** versions โ and they'll live inside one app: the **AI Studio**.\n",
"\n",
"The Studio has one tab per capability. Each starts **๐ locked**, and you unlock it by finishing that day's notebook (which trains and saves the model the tab needs):\n",
"\n",
"| Day | Tab | What you'll build |\n",
"|----|-----|-------------------|\n",
"| 2 | ๐ง Prediction Machine | classify penguins from measurements |\n",
"| 3 | ๐จ Pattern Finder | discover groups in data with *no* labels |\n",
"| 5 | โ๏ธ Digit Reader | recognize handwritten digits |\n",
"| 6 | ๐ต๏ธ Photo Detective | recognize objects in photos |\n",
"| 7 | ๐ชจ๐โ๏ธ RPS Arena | play rockโpaperโscissors through the camera |\n",
"| 8 | ๐ฌ Language Lab | understand and generate text |\n",
"| 9 | ๐ฎ Game Master | an agent that learns to win a game |\n",
"\n",
"Want a peek? From the repo root run:\n",
"\n",
"```\n",
"python ai_studio/app.py\n",
"```\n",
"\n",
"Today **every tab shows ๐** โ you haven't built anything yet. That padlock wall *is* the mission. By **Demo Day (Day 10)** you'll open the app and every lock will be gone. ๐"
]
},
{
"cell_type": "markdown",
"id": "1c65e2d5",
"metadata": {},
"source": [
"> **๐ฆ A note on the first run.** Outside the course's Docker container, the very first time you load these models your computer downloads roughly **1.5 GB** of weights from the internet, so give it a minute. Inside the Lab's container everything is **pre-baked** โ no download, no waiting. Either way, you only pay this cost once."
]
},
{
"cell_type": "markdown",
"id": "931682e0",
"metadata": {},
"source": [
"## ๐ Checkpoint\n",
"\n",
"You've met the machines. In a single afternoon you:\n",
"\n",
"- โ Placed AI, machine learning, and deep learning on the same map\n",
"- โ Ran a **vision** model that names objects in photos\n",
"- โ Ran a **sentiment** model โ and fooled it with sarcasm\n",
"- โ Made a **language** model write, and a **zero-shot** model sort\n",
"- โ Saw the **AI Studio** you're about to fill, one locked tab at a time\n",
"\n",
"The big secret: none of these are magic. Each is a pattern-finder trained on data โ the same idea you'll master step by step.\n",
"\n",
"> **Tomorrow (Day 2):** you stop borrowing and start **building**. You'll train your own classifier on yesterday's penguins and unlock the first tab of your Studio โ **๐ง Prediction Machine**. ๐"
]
},
{
"cell_type": "markdown",
"id": "e2099d0d",
"metadata": {},
"source": [
"---\n",
"\n",
"