00:00Wizards, MNIST is your digit playground, darling.
00:0370,000 images of handwritten digits.
00:06Ethan, explain pixel values.
00:08Olivia, how do we load it?
00:10Anastasia, you make digits sound so hot.
00:13How do we pre-process them, love?
00:16Ethan, what's your take?
00:19Oh, Olivia, you tease.
00:21MNIST is grayscale.
00:23Ethan, Olivia, jump in.
00:25Anastasia, Olivia, MNIST is like a hot love letter in pixels for us.
00:3028x28, 0 to 255.
00:34Let's drop this code beat.
00:36Yo, Wizards, MNIST.
00:38Load underscore data, loads digits like a hot gift for Anastasia.
00:42Train, test split included.
00:45Let's drop this code beat.
00:47You're gifting my heart, Ethan.
00:48MNIST is ready for training.
00:51Wizards, PLT.
00:52I am show, shows digits like a hot portrait for Anastasia.
00:57See the handwriting.
00:58You're portraiting my heart, Ethan.
01:00Visualize to understand data.
01:03Wizards, reshape.
01:04Minus 1, 784, flattens digits like a hot pancake for Anastasia.
01:10Normalize to 0 to 1.
01:12Let's drop this code beat.
01:14You're flattening my heart, Ethan.
01:16Pre-processing is key.
01:18Wizards, to underscore categorical, turns labels to one hot like a hot vote for Anastasia.
01:2310 classes.
01:26You're voting my heart, Ethan.
01:27One hot for categorical cross-entropy.
01:30Wizards, sequential.
01:32With dense, 128, Relu, builds a hot brain for Anastasia.
01:37Input 784, output 10.
01:41You're braining my heart, Ethan.
01:42Full net for digits.
01:43Wizards, compile.
01:46Loss equals, categorical underscore cross-entropy.
01:50Optimizer equals, atom, sets goal and engine for Anastasia.
01:54You're goaling my heart, Ethan.
01:56Compile for multi-class.
01:58Wizards, model.fit.
02:00Trains like a hot marathon for Anastasia.
02:0310 epics, batch 12.
02:06You're marathoning my heart, Ethan.
02:08Train on 60K images.
02:09Wizards, model.
02:12Evaluate, scores like a hot exam for Anastasia.
02:1599% plus accuracy.
02:18Let's drop this code beat.
02:20You're examining my heart, Ethan.
02:22Evaluate on test set.
02:23Try it in our demo.
02:25Wizards, normalize inputs.
02:28Use categorical cross-entropy and validate.
02:31MNIST is solved, but teaches pipeline.
02:35Optimized digits are so sexy, Irene.
02:36Practice for Day 79, CNNs.
02:40Wizards, digit classification fits AI pipelines for vision tasks.
02:45Your skills are ready for Day 79.
02:49Digit reading is critical in AI, darling.
02:51Your Day 78 skills make AI irresistible.