book

Machine Learning Q and AI

Name: Machine Learning Q and AI
Author: Sebastian Raschka
ISBN: 9781718503762

by Sebastian Raschka

April 2024

Intermediate to advanced

264 pages

6h 10m

English

No Starch Press

Read now

Unlock full access

Cover Page
Title Page
Copyright Page
Dedication Page
About the Author
About the Technical Reviewer
BRIEF CONTENTS
CONTENTS IN DETAIL
FOREWORD
ACKNOWLEDGMENTS

INTRODUCTION
Who Is This Book For?What Will You Get Out of This Book?How to Read This BookOnline Resources
PART I: NEURAL NETWORKS AND DEEP LEARNING
1. EMBEDDINGS, LATENT SPACE, AND REPRESENTATIONS
EmbeddingsLatent SpaceRepresentationExercisesReferences
2. SELF-SUPERVISED LEARNING
Self-Supervised Learning vs. Transfer LearningLeveraging Unlabeled DataSelf-Prediction and Contrastive Self-Supervised LearningExercisesReferences
3. FEW-SHOT LEARNING
Datasets and TerminologyExercises
4. THE LOTTERY TICKET HYPOTHESIS
The Lottery Ticket Training ProcedurePractical Implications and LimitationsExercisesReferences
5. REDUCING OVERFITTING WITH DATA
Common MethodsCollecting More DataData AugmentationPretrainingOther MethodsExercisesReferences
6. REDUCING OVERFITTING WITH MODEL MODIFICATIONS
Common MethodsRegularizationSmaller ModelsCaveats with Smaller ModelsEnsemble MethodsOther MethodsChoosing a Regularization TechniqueExercisesReferences
7. MULTI-GPU TRAINING PARADIGMS
The Training ParadigmsModel ParallelismData ParallelismTensor ParallelismPipeline ParallelismSequence ParallelismRecommendationsExercisesReferences
8. THE SUCCESS OF TRANSFORMERS
The Attention MechanismPretraining via Self-Supervised LearningLarge Numbers of ParametersEasy ParallelizationExercisesReferences
9. GENERATIVE AI MODELS
Generative vs. Discriminative ModelingTypes of Deep Generative ModelsEnergy-Based ModelsVariational AutoencodersGenerative Adversarial NetworksFlow-Based ModelsAutoregressive ModelsDiffusion ModelsConsistency ModelsRecommendationsExercisesReferences
10. SOURCES OF RANDOMNESS
Model Weight InitializationDataset Sampling and ShufflingNondeterministic AlgorithmsDifferent Runtime AlgorithmsHardware and DriversRandomness and Generative AIExercisesReferences
PART II: COMPUTER VISION
11. CALCULATING THE NUMBER OF PARAMETERS
How to Find Parameter CountsConvolutional LayersFully Connected LayersPractical ApplicationsExercises
12. FULLY CONNECTED AND CONVOLUTIONAL LAYERS
When the Kernel and Input Sizes Are EqualWhen the Kernel Size Is 1RecommendationsExercises
13. LARGE TRAINING SETS FOR VISION TRANSFORMERS
Inductive Biases in CNNsViTs Can Outperform CNNsInductive Biases in ViTsRecommendationsExercisesReferences
PART III: NATURAL LANGUAGE PROCESSING
14. THE DISTRIBUTIONAL HYPOTHESIS
Word2vec, BERT, and GPTDoes the Hypothesis Hold?ExercisesReferences
15. DATA AUGMENTATION FOR TEXT
Synonym ReplacementWord DeletionWord Position SwappingSentence ShufflingNoise InjectionBack TranslationSynthetic DataRecommendationsExercisesReferences
16. SELF-ATTENTION
Attention in RNNsThe Self-Attention MechanismExercisesReferences
17. ENCODER- AND DECODER-STYLE TRANSFORMERS
The Original TransformerEncodersDecodersEncoder-Decoder HybridsTerminologyContemporary Transformer ModelsExercisesReferences
18. USING AND FINE-TUNING PRETRAINED TRANSFORMERS
Using Transformers for Classification TasksIn-Context Learning, Indexing, and Prompt TuningParameter-Efficient Fine-TuningReinforcement Learning with Human FeedbackAdapting Pretrained Language ModelsExercisesReferences
19. EVALUATING GENERATIVE LARGE LANGUAGE MODELS
Evaluation Metrics for LLMsPerplexityBLEU ScoreROUGE ScoreBERTScoreSurrogate MetricsExercisesReferences
PART IV: PRODUCTION AND DEPLOYMENT
20. STATELESS AND STATEFUL TRAINING
Stateless (Re)trainingStateful TrainingExercises
21. DATA-CENTRIC AI
Data-Centric vs. Model-Centric AIRecommendationsExercisesReferences
22. SPEEDING UP INFERENCE
ParallelizationVectorizationLoop TilingOperator FusionQuantizationExercisesReferences
23. DATA DISTRIBUTION SHIFTS
Covariate ShiftLabel ShiftConcept DriftDomain ShiftTypes of Data Distribution ShiftsExercisesReferences
PART V: PREDICTIVE PERFORMANCE AND MODEL EVALUATION
24. POISSON AND ORDINAL REGRESSION
Exercises
25. CONFIDENCE INTERVALS
Defining Confidence IntervalsThe MethodsMethod 1: Normal Approximation IntervalsMethod 2: Bootstrapping Training SetsMethod 3: Bootstrapping Test Set PredictionsMethod 4: Retraining Models with Different Random SeedsRecommendationsExercisesReferences
26. CONFIDENCE INTERVALS VS. CONFORMAL PREDICTIONS
Confidence Intervals and Prediction IntervalsPrediction Intervals and Conformal PredictionsPrediction Regions, Intervals, and SetsComputing Conformal PredictionsA Conformal Prediction ExampleThe Benefits of Conformal PredictionsRecommendationsExercisesReferences
27. PROPER METRICS
The CriteriaThe Mean Squared ErrorThe Cross-Entropy LossExercises
28. THE K IN K-FOLD CROSS-VALIDATION
Trade-offs in Selecting Values for kDetermining Appropriate Values for kExercisesReferences
29. TRAINING AND TEST SET DISCORDANCE
Exercises
30. LIMITED LABELED DATA
Improving Model Performance with Limited Labeled DataLabeling More DataBootstrapping the DataTransfer LearningSelf-Supervised LearningActive LearningFew-Shot LearningMeta-LearningWeakly Supervised LearningSemi-Supervised LearningSelf-TrainingMulti-Task LearningMultimodal LearningInductive BiasesRecommendationsExercisesReferences
AFTERWORD
APPENDIX: ANSWERS TO THE EXERCISES
Chapter 1Chapter 2Chapter 3Chapter 4Chapter 5Chapter 6Chapter 7Chapter 8Chapter 9Chapter 10Chapter 11Chapter 12Chapter 13Chapter 14Chapter 15Chapter 16Chapter 17Chapter 18Chapter 19Chapter 20Chapter 21Chapter 22Chapter 23Chapter 24Chapter 25Chapter 26Chapter 27Chapter 28Chapter 29Chapter 30
INDEX

Content preview from Machine Learning Q and AI

19EVALUATING GENERATIVE LARGE LANGUAGE MODELS

What are the standard metrics for evaluating the quality of text generated by large language models, and why are these metrics useful?

Perplexity, BLEU, ROUGE, and BERTScore are some of the most common evaluation metrics used in natural language processing to assess the performance of LLMs across various tasks. Although there is ultimately no way around human quality judgments, human evaluations are tedious, expensive, hard to automate, and subjective. Hence, we develop metrics to provide objective summary scores to measure progress and compare different approaches.

This chapter discusses the difference ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Publisher Resources

ISBN: 9781098168827Errata Page

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Machine Learning Q and AI

by Sebastian Raschka

19EVALUATING GENERATIVE LARGE LANGUAGE MODELS

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.