Completed
70. From model logits to prediction probabilities to prediction labels
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
PyTorch for Deep Learning & Machine Learning – Full Course
Automatically move to the next video in the Classroom when playback concludes
- 1 Introduction
- 2 0. Welcome and "what is deep learning?"
- 3 1. Why use machine/deep learning?
- 4 2. The number one rule of ML
- 5 3. Machine learning vs deep learning
- 6 4. Anatomy of neural networks
- 7 5. Different learning paradigms
- 8 6. What can deep learning be used for?
- 9 7. What is/why PyTorch?
- 10 8. What are tensors?
- 11 9. Outline
- 12 10. How to and how not to approach this course
- 13 11. Important resources
- 14 12. Getting setup
- 15 13. Introduction to tensors
- 16 14. Creating tensors
- 17 17. Tensor datatypes
- 18 18. Tensor attributes information about tensors
- 19 19. Manipulating tensors
- 20 20. Matrix multiplication
- 21 23. Finding the min, max, mean & sum
- 22 25. Reshaping, viewing and stacking
- 23 26. Squeezing, unsqueezing and permuting
- 24 27. Selecting data indexing
- 25 28. PyTorch and NumPy
- 26 29. Reproducibility
- 27 30. Accessing a GPU
- 28 31. Setting up device agnostic code
- 29 33. Introduction to PyTorch Workflow
- 30 34. Getting setup
- 31 35. Creating a dataset with linear regression
- 32 36. Creating training and test sets the most important concept in ML
- 33 38. Creating our first PyTorch model
- 34 40. Discussing important model building classes
- 35 41. Checking out the internals of our model
- 36 42. Making predictions with our model
- 37 43. Training a model with PyTorch intuition building
- 38 44. Setting up a loss function and optimizer
- 39 45. PyTorch training loop intuition
- 40 48. Running our training loop epoch by epoch
- 41 49. Writing testing loop code
- 42 51. Saving/loading a model
- 43 54. Putting everything together
- 44 60. Introduction to machine learning classification
- 45 61. Classification input and outputs
- 46 62. Architecture of a classification neural network
- 47 64. Turing our data into tensors
- 48 66. Coding a neural network for classification data
- 49 68. Using torch.nn.Sequential
- 50 69. Loss, optimizer and evaluation functions for classification
- 51 70. From model logits to prediction probabilities to prediction labels
- 52 71. Train and test loops
- 53 73. Discussing options to improve a model
- 54 76. Creating a straight line dataset
- 55 78. Evaluating our model's predictions
- 56 79. The missing piece – non-linearity
- 57 84. Putting it all together with a multiclass problem
- 58 88. Troubleshooting a mutli-class model
- 59 92. Introduction to computer vision
- 60 93. Computer vision input and outputs
- 61 94. What is a convolutional neural network?
- 62 95. TorchVision
- 63 96. Getting a computer vision dataset
- 64 98. Mini-batches
- 65 99. Creating DataLoaders
- 66 103. Training and testing loops for batched data
- 67 105. Running experiments on the GPU
- 68 106. Creating a model with non-linear functions
- 69 108. Creating a train/test loop
- 70 112. Convolutional neural networks overview
- 71 113. Coding a CNN
- 72 114. Breaking down nn.Conv2d/nn.MaxPool2d
- 73 118. Training our first CNN
- 74 120. Making predictions on random test samples
- 75 121. Plotting our best model predictions
- 76 123. Evaluating model predictions with a confusion matrix
- 77 126. Introduction to custom datasets
- 78 128. Downloading a custom dataset of pizza, steak and sushi images
- 79 129. Becoming one with the data
- 80 132. Turning images into tensors
- 81 136. Creating image DataLoaders
- 82 137. Creating a custom dataset class overview
- 83 139. Writing a custom dataset class from scratch
- 84 142. Turning custom datasets into DataLoaders
- 85 143. Data augmentation
- 86 144. Building a baseline model
- 87 147. Getting a summary of our model with torchinfo
- 88 148. Creating training and testing loop functions
- 89 151. Plotting model 0 loss curves
- 90 152. Overfitting and underfitting
- 91 155. Plotting model 1 loss curves
- 92 156. Plotting all the loss curves
- 93 157. Predicting on custom data
