Concepts starting with F

213 concepts, each with a plain-language explanation, examples and diagrams. Pick a letter or open a concept.

Failure Mode Analysis for AI Systems — A validation gate is a checkpoint that measures how a system performs before it is deployed, and it comes before a pilot rather…
Fall Through and Case Grouping — The traditional switch uses a colon after each case label, and a case label may hold one constant before Java 17.
FDE Work from Chapter 12 — Production engineering instinct is the understanding of a system's operational state: knowing that a system is on and doing work…
Feature Construction — RBF networks are function approximators whose features are radial basis functions.
Feature Construction for Linear Function Approximation — Each OrderN polynomial basis function multiplies one powered term for every state variable.
Feature Delivery Engineering in AI Products — An AI engineering role can be self-terminating for a single product: when the engineer does the job well, that product no longer…
Feature Engineering — A machine learning problem is defined by its inputs, predicted outputs, and problem type.
Feature extraction — Preprocessing changes raw data into numerical forms that neural networks can use.
Feature Manipulation and Normalization — Feature transformations should be selected according to the property that needs to change.
Feature Mappings for Polynomial Classifiers — Compression keeps a smaller representation from which a zero-training-loss predictor can be reconstructed in the realizable case.
Feature Maps and Filters — Convolution focuses on local patches rather than treating the entire image as one undivided pattern.
Feature Maps in Convolutional Neural Networks — Max pooling replaces each local window with its maximum value.
Feature Normalization — Feature manipulation transforms each original feature to create a resulting feature vector.
Feature Representations in Machine Learning — Separating polynomials are functions of the form sign(p(x)), with p a degree-r polynomial on R^d.
Feature Representations in VGG Networks — VGG19 acts as a pretrained measurement network rather than as the object being generated.
Feature Selection — Feature manipulation transforms each original feature to create a resulting feature vector.
Feature Selection and Model Design — Selecting features determines the dimensionality of the feature space.
Feature Selection Techniques — Filter methods reduce a feature set by scoring each feature independently.
Feature Transformation — Feature learning learns a function ψ that maps instances from X into d-dimensional feature vectors.
Feature Vectors — X names the collection of objects being studied, while Y names the collection of possible labels.
Feature-Based Splitting — Decision tree learning searches for a tree that minimizes the relevant bound, but solving that search exactly is computationally…
Feed-Forward Networks in Transformers — A TransformerBlock class builds its attention module, feed-forward network, two LayerNorm modules, and a dropout module from a…
Feedback Loops from Customer Deployments to Product Platforms — A single do-everything delivery role tends to split into two specialized lanes as organizations grow, and the same pattern has…
Feedforward artificial neural networks — A sufficiently large single hidden layer with sigmoid units can approximate any continuous function on a compact input region to…
Feedforward Neural Network Architecture — Neural networks with an appropriate depth-2 structure can represent every Boolean function.
Feedforward Neural Network Structure — Activation functions determine how a neuron converts its scalar input into an output.
Feedforward Neural Networks — Feedforward networks process inputs independently because they do not maintain state between inputs.
Fetching Data from Web APIs — ET.fromstring converts an XML string into a tree structure, transforming flat text into a navigable, queryable object.
Fetching Web Content with urllib — Parentheses in regex patterns create capturing groups that extract specific substrings from a match.
Fighting Overfitting — A machine learning problem is defined by its inputs, predicted outputs, and problem type.
File and Archive Operations — Verbosity levels (-v and -q flags) let users control how much information a script outputs, making it suitable for different…
File handling basics — The with statement automates resource acquisition and cleanup, eliminating the need for explicit try...finally blocks.
File handling best practices — The finally block executes unconditionally after try and except blocks, making it ideal for resource cleanup.
File Handling in Python — urllib is a standard Python library that simplifies web page retrieval by hiding HTTP protocol and header details behind a…
File I/O and Data Persistence — The pickle module converts Python objects into byte streams for storage and retrieves them back into memory.
File I/O Operations in Python — While this chapter will focus on using Python to work with data in SQLite database files, many operations can be done more…
File Input and Output — Next, we will learn how to deal with input/output and how to access files in Python.
File Input and Output Basics — Try/except structures prevent program crashes by catching exceptions and executing recovery code when operations fail.
File Input and Output Operations — Pythonic means doing something the Python way, combining both engineering and artistic principles in your code.
File Input and the open() Function — find() returns the starting index of a substring if found, or -1 if not found
File Input/Output Streams — C++ file streams are used for reading from and writing to files.
File Management and Directories — The files and directories to be backed up are specified in a list.
File Modes and Their Purposes — Writing to a file requires opening it with mode 'w', calling write() to insert data, and closing the file to flush the buffer to…
File Modes: Reading, Writing, and Appending — Opening a file asks the operating system to locate the file by name and verify it exists and is readable.
File Reading and Loops — File parsing crashes often occur when code assumes data that is not always present, such as non-empty lines or fields at…
File Stream Best Practices — User-defined types can define << and >> operators when their values have meaningful textual representations.
File Transfer and Version Control — Newline characters mark the end of lines in text files, but different systems use different representations: \n, \r, or \r\n.
File Transfer Protocols (FTP) — curl and wget are command-line tools for retrieving files from the web on Unix-like systems using HTTP or FTP protocols.
Filter Methods for Feature Selection — Greedy selection evaluates feature subsets through the learning algorithm rather than separating selection from learning.
Filtering and Mapping Streams — reduce combines the values of a stream into one result using a BinaryOperator, and a sum is one example where the operation is…
Filtering and Processing Data from Files — The continue statement skips the rest of the current loop iteration and jumps back to the loop start, enabling efficient…
Filtering and Transforming Lists — Built-in functions (len, sum, max, min) extract information from lists without requiring custom loops.
Filtering Dictionary Data — Looping through a dictionary directly iterates over its keys, not its values.
Filtering Lists by Condition — A for loop over a dictionary iterates through keys; to filter by value, use the index operator to retrieve each value and test…
Filtering Results with WHERE Clauses — JOIN reconstructs data from multiple normalized tables by matching rows based on key values specified in the ON condition.
Final Evaluation of a Fine-Tuned Model — The number of epochs is the number of full passes through the training dataset, and it is a hyperparameter that depends on the…
Finally Blocks — Suppose you are reading a file in your program. How do you ensure that the file object is closed properly whether or not an…
Finally blocks for cleanup code — The else block in exception handling runs only when the try block succeeds without raising an exception.
Finding and Selecting Tags with BeautifulSoup — Tag inspection in BeautifulSoup means accessing the individual components of a Tag object: its attributes (via get() or attrs)…
Finding Errors With the Eclipse Debugger — A workspace is the directory where your work is stored, and you can use several workspaces and choose one at startup.
Finding extremes in data — Loops scan through data by following a three-part structure: initialize a tracking variable, compute on each item, and inspect…
Finding Extremes: Maximum and Minimum Values in Loops — An accumulator is a variable that starts with an initial value before a loop, gets updated on each iteration, and holds the…
Finding the Maximum Value — Data scanning loops follow a three-step construction: initialize variables before the loop, perform computation on each item…
Fine-Tuning a Classifier with LoRA — Freezing sets requires_grad to False on every original parameter, so none of the pretrained weights are updated during…
Fine-Tuning a GPT Model for Text Classification — A saved model can be resumed: load the model and optimizer in a new Python session or notebook, then continue pretraining for…
Fine-Tuning a Pre-Trained LLM for Spam Classification — train_classifier_simple nests a batch loop inside an epoch loop, and each batch resets gradients, computes loss, backpropagates…
Fine-Tuning a Pretrained Convnet — A pretrained network is a saved network trained previously on a large dataset, such as ImageNet.
Fine-Tuning an LLM for Classification — Pretrained weights are weights trained on a large dataset, and loading them gives an LLM a solid starting point for fine-tuning.
Fine-Tuning an LLM for Spam Classification — LoRA weight matrices are kept separate from the original model weights, so the pretrained weights can remain unchanged.
Fine-Tuning and Evaluating LLMs — Loading a pretrained LLM means bringing an already trained model in as the starting point for further training.
Fine-Tuning and Evaluating the Model — Initial accuracy is measured separately on the training, validation, and test datasets, giving a baseline before fine-tuning…
Fine-Tuning Categories for Large Language Models — Instruction fine-tuning trains a language model on a set of tasks, with each task presented alongside a specific instruction.
Fine-Tuning GPT-2 for Text Classification — Loading pretrained GPT-2 weights means calling load_weights_into_gpt with the gpt instance and the params, then moving the model…
Fine-Tuning Large Language Models — Evaluating a classification model involves calculating the classification accuracy.
Fine-Tuning Large Language Models on Instruction Data — Larger models such as gpt2-medium (355M) are more computationally intensive than gpt2-small (124M), because they have more…
Fine-Tuning Large Language Models with Human Feedback — Experimental LLM architectures are designs that are not based on the transformer, and the source uses them to show that LLMs…
Fine-Tuning LLMs for Classification — Many pretrained LLMs are available as open source models and can be used as general-purpose tools to write, extract, and edit…
Fine-Tuning LLMs on Instruction Data — A baseline evaluation runs a pretrained LLM on an instruction-following task before any fine-tuning, so later improvements have…
Fine-Tuning Pretrained Language Models for Classification — Fine-tuning starts from a pretrained LLM, a model that has already been trained, rather than from scratch.
Fine-Tuning Pretrained Models — The GPT-3 authors did not share their training dataset, so Dolma is offered as a comparable publicly available resource for…
Fine-Tuning to Follow Instructions — Fine-tuning adapts a pretrained LLM to a specific task; for classification, the training data consists of examples that carry…
Finetuning and Model Optimization Techniques — Alignment tax names costs or penalties associated with aligning technologies or models with human preferences or with more…
Finetuning Language Models — Foundation models include both large language models and large multimodal models, and they can be adapted for many purposes.
Finetuning Pre-Trained Language Models — Knowledge distillation, also called model distillation, trains a small student model to mimic a larger teacher model.
Finetuning Techniques — Finetuning is performed by both model developers, who post-train models before release, and application developers, who adapt…
Finetuning Versus Training from Scratch — Data diversity is the variety of training data, covering topics, speaking styles, and formats, and it is needed so the model's…
Finite Classes Are Agnostic PAC Learnable — Uniform convergence is a statement about all hypotheses in H simultaneously.
Finite Hypothesis Classes — i.i.d. sampling means independent examples drawn from the same distribution D.
Finite Markov Decision Processes — Use the Markov-property condition to identify an MDP.
First Normal Form: Eliminating Repeating Groups — Redundancy occurs when the same information is stored multiple times across different rows in a table, such as repeating an…
First Slice of a Project — Writing code is often only about 20 percent of the problem; the other 80 percent is testing, reviewing, deploying, and…
First Version of the Backup Program — The second version works fine when I do many backups, but when there are lots of backups, I am finding it hard to differentiate…
First-Class Functions — A Java lambda expression is written with the arrow operator ->, with a dummy parameter on the left and the computing expression…
First-visit Monte Carlo methods — On-policy first-visit MC control alternates between episode generation and policy improvement.
Fixed Policies and On-line Adaptation — On-line learning adapts during system operation.
Fixed-Parameter Templates — A variadic template, sometimes called a variadic, can accept any number of arguments, so its argument count is set by each call…
Flat Partial Returns — Discounting-aware importance sampling uses the structure of discounted returns instead of treating each return as an indivisible…
Floating-Point Output Precision — A sticky manipulator persists for subsequent operations on the stream rather than applying only to one output statement.
Floating-Point Precision and Rounding — math.pi is a variable from the math module that provides an approximation of π accurate to about 15 digits; use it without…
Floor Division and Integer Division — The modulus operator (%) returns the remainder when one integer is divided by another.
FLOSS (Free/Libré and Open Source Software) — Software that allows users to freely distribute copies, read the source code, make changes, and use pieces of it in new free…
Following PEP 8 Style Guidelines — The def keyword marks the start of a function definition and is required for every function.
Fonts and the Font Class in Java — FontMetrics measures characters and strings in one font and is obtained from g.getFontMetrics(font).
for Loops — A for loop repeats a block of code for each item in a sequence, assigning each item to a loop variable one at a time.
For Loops and Iteration — Write a program that prompts the user for a filename, opens the file, and prints each line in uppercase using the upper() string…
For Loops and Iteration in Python — The .items() method returns tuples of (key, value) pairs from a dictionary.
For Loops and Loop Control — The counter pattern is a fundamental computational technique: initialize a counter to 0, loop through data, check a condition…
For Loops and Range — for i in range(1, 5): print i else : print 'The for loop is over'
For Loops and String Iteration — For loops are the most common and natural way to iterate through list elements, using identical syntax to string iteration.
For Loops and String Traversal — Traversal is the process of handling a string one character at a time, from beginning to end.
For Loops with else Clauses — The if statement is used to check a condition: if the condition is true, we run a block of statements (called the if-block )…
Foreign Keys and Referential Integrity — A one-to-many relationship connects one entity instance to multiple instances of another entity.
Formal and Actual Parameters — Formal parameters are the variables declared in a subroutine's parameter list, and actual parameters are the values or…
Formal Parameters and Passing Values to Subroutines — A subroutine can use local variables declared inside it, formal parameters that receive values from the caller, and variables…
Formatted and Unformatted Output — The I/O stream library provides formatted and unformatted buffered I/O of text and numeric values.
Formatted Input and Output — A stringstream can be used for both reading and writing, and istringstream and ostringstream specialize it for parsing and…
Formatted Input Operations — An istringstream is a class for reading from a string stream, so it is the right choice when a program needs to interpret a…
Formatting Output with Manipulators — An ostringstream collects text and values written to it with the << operator, in the order they are written.
Formatting Strings for Output — The * operator repeats a string by multiplying it by an integer, producing a new string with the original string concatenated…
Formatting Training Examples for Finetuning — A finetuned model works best when the prompts you send match the format of its finetuning data, including small details such as…
Forward Deployed Engineering (FDE) as a Discovery Role — The FDE role is described by practitioners in two ways: as a product strategy that discovers what the product must become, and…
Forward Deployed Engineering and FDE Programs — The high projected failure figure is best read as a forecast of future repair workload, not as proof that FDE programs are a…
Forward Deployed Engineering: Role Overview — The FDE role has historically not been entry-level because its combination of capabilities was assumed to need years of…
Forward Greedy Selection — Greedy selection evaluates feature subsets through the learning algorithm rather than separating selection from learning.
Forward Propagation in Neural Networks — A feedforward neural network is a directed acyclic graph whose nodes are neurons and whose directed edges are weighted…
Forward View of a Learning Algorithm — Learning is postponed until the episode has finished.
Forward-Deployed Engineering and Customer Engagements — The spec-and-test tradition treats test cases as guarantees that always pass, while the eval discipline treats them as samples…
Forward-Deployed Engineering and Deployment Work — A price fixed at contract time is set before the operators' habits are known, so discovery is the main protection against a…
Forward-Deployed Engineering Model — A forward-deployed engineer is assigned to a single customer, embeds with that customer's team, and writes code until the work…
Foundation Model Applications — The course is organized into modules, chapters, and lessons, and each lesson builds on the ones before it.
Foundation Model Design and Usage — Nonlinear functions are needed to break the linearity of feedforward layers, and simpler functions that are faster to compute…
Foundation Model Training Stages — In-context learning teaches a model a behavior through examples or instructions placed in the prompt, without updating the…
Foundation Models — Pretraining trains an LLM on a large, diverse corpus of raw text to develop a broad understanding of language.
Foundation Models and Application Building — Model development is the layer most associated with traditional ML engineering and has three responsibilities: modeling and…
Foundation Models and Data Modalities — Foundation models emerged through a chain: self-supervision enabled the move from language models to large language models, and…
Foundation Models and Demonstration Data — Preference finetuning aims to get AI models to behave according to human preference, which the source calls an ambitious, if not…
Foundation Models and Emerging Applications — A larger model is usually paired with more training data, because the goal is to maximize its performance rather than to match a…
Foundation Models and Finetuning — Memory is a bottleneck for foundation models during both inference and finetuning, and finetuning typically needs much more…
Foundation Models and How They Are Adapted — AI engineering is the process of building applications on top of foundation models, and it has emerged as one of the…
Foundation Models and Language Models — Evaluation receives less investment than other parts of AI engineering, and the gap appears in research attention, tooling, and…
Foundation Models and Model APIs — AI products are harder to maintain because the AI space moves very fast, and building on foundation models means committing to…
Foundation Models and Parameters — A model's parameter count is usually written at the end of its name, as in Llama-13B.
Foundation Models and Their Training Data — The words prompt and context are used inconsistently, so check which meaning an author intends; in this course, prompt is the…
Foundations of Deep Learning — Deep learning matters partly because its recent surge is linked to multiple contributing factors.
Foundations of Machine Learning — Getting Started introduces how to approach the course rather than requiring you to memorize its entire structure.
Foundations of Neuroscience for Reinforcement Learning — Dendrites receive information and the axon carries neuronal output.
Foundations of Reinforcement Learning — Part III is the book's final move beyond standard reinforcement learning ideas.
Founder Experience as Transferable Engineering Skill — The Founder-FDE arbitrage is the gap that appears when the same skills are compensated differently in a founder role than in an…
Founder-Mode Operator Profile — Founder-mode temperament is a mindset that holds a core thesis tightly while staying loose on the decisions downstream of it.
Founder-Mode Temperament — Founder-mode splits a plan into a tightly held thesis and loosely held downstream decisions, while keeping strong opinions at…
Fourier Basis Functions — A d-dimensional unit hypercube contains state vectors s = (s₁, ..., s_d)ᵀ with every component in [0, 1].
Free Store and Object Allocation — A concrete type has its representation as part of its definition, as with complex and Vector, and resembles built-in types.
Freezing Model Layers for Fine-Tuning — Replacing model.out_head changes the final projection from the vocabulary size of 50,257 to a size set by the task, here…
From Token IDs to Embedding Vectors — The tokenizer splits text with re.split at punctuation, the double dash, and whitespace, keeping matched punctuation through a…
Fruitful Functions vs. Void Functions — The return statement sends a value from inside a function back to the calling code, making that value the result of the function…
Function Adaptors and Objects — Some algorithms require an operation to be called as a nonmember function, and this is the major use of mem_fn().
Function Approximation and Reinforcement Learning — Function approximation turns examples of a desired function into a broader estimate.
Function Approximation for Continuous State Spaces — Polynomial basis functions provide a family of candidate functions for approximation.
Function Approximation in Reinforcement Learning — On the 10-armed testbed, UCB generally performs better than epsilon-greedy action selection.
Function approximation with neural networks — A sufficiently large single hidden layer with sigmoid units can approximate any continuous function on a compact input region to…
Function Arguments and Object Modification — Variables are references to objects in memory, not the objects themselves.
Function Arguments and Return Values — Parameter names and argument variable names are independent; the function receives the value, not the name.
Function Basics and Return Statements — The return statement is used to return from a function i.e. break out of the function. We can optionally return a value from the…
Function Calling in Language Models — A single model performs one kind of action, while external tools let an agent perceive its environment and act upon it.
Function Calls and Scope — Python tracebacks report where an error was discovered, not necessarily where it was caused — the actual problem often lies on…
Function Declarations and Definitions — Argument passing follows the semantics of copy initialization: the argument's type is checked against the parameter type, and an…
Function Definition and Parameters — Parameters are specified within the pair of parentheses in the function definition, separated by commas. When we call…
Function Definitions and Calls — def say_hello(): # block belonging to the function print 'hello world' # End of function say_hello() # call the function…
Function Design Best Practices — Default argument values make certain function parameters optional by specifying a value to use if the caller does not provide one.
Function Objects — A lambda expression is a notation that generates a function object, similar in kind to writing Less_than<int>{x}.
Function Objects and Callable Values — bind() creates a new function object from an existing function and the arguments you choose to bind.
Function Objects and Function Adaptors — The function adaptor mem_fn(mf) produces a function object that can be called as a nonmember function.
Function Objects and Lambda Expressions — A function object carries its comparison value inside itself, avoiding separate functions per value and global variables
Function Objects and Operator Overloading — std::function is a type that can hold any callable object, so one variable can refer to a function, a function object, or a…
Function Objects and Standard Algorithms — A function object extends the meaning of a standard algorithm by supplying behavior that the algorithm itself does not know.
Function Objects and the Application Operator — A thread can be given a function and its arguments, and the function then runs in a separate thread with those arguments.
Function Objects in C++ — std::function is a type that can hold any object you can invoke using the call operator ().
Function Overloading — bind() cannot resolve an overloaded function name on its own, because the name alone does not identify one function.
Function Overloading in C++ — Comprehensibility comes first because maintainability depends on being able to understand code.
Function Overriding — A derived function overrides a base virtual function only when name and type match exactly.
Function Parameters — When you use positional arguments, you must remember the exact order in which the function expects them. If a function takes…
Function Parameters and Arguments — How It Works We define a function called say_hello using the syntax as explained above. This function takes no parameters and…
Function Parameters and Mutable Objects — Aliasing occurs when two or more variables refer to the same object in memory. The assignment b = a creates an alias, not a copy.
Function Parameters and Return Values — How It Works We define a function called say_hello using the syntax as explained above. This function takes no parameters and…
Function Parameters and Scope — Reassigning a function parameter (e.g., t = t[1:]) does not modify the original list passed to the function; it only changes…
Function Prototypes and Headers in C++ — C and C++ can use different linkage conventions, mainly because C++ emphasizes type checking and supports overloading.
Function Return Values — The return statement is used to return from a function i.e. break out of the function. We can optionally return a value from the…
Function Scope and Local Variables — When you declare variables inside a function definition, they are not related in any way to other variables with the same names…
Function Templates — Design and debug a non-template version first, then generalize it by adding parameters.
Functional Interfaces — A functional interface is an interface with a single abstract method, and the source uses it to represent a single method.
Functional Interfaces and Lambda Expressions — A default method is declared inside an interface with the modifier default and carries a working implementation.
Functional Interfaces in Java — A Java lambda expression is written with the arrow operator ->, with a dummy parameter on the left and the computing expression…
Functions and Arguments — A type function is evaluated at compile-time, not while the program runs.
Functions and Blocks — How It Works In this program, we accept input from the user, but we process the input string only if it is at least 3 characters…
Functions and Code Reuse — Sequential execution performs statements one after another in the order they appear in the script.
Functions and Function Arguments — Parameters are specified within the pair of parentheses in the function definition, separated by commas. When we call…
Functions and How to Call Them — Python is not intelligent and cannot guess your intent. It is strict about syntax and will reject any code that violates its…
Functions and how to define them — A module is a .py file containing functions and variables designed to be reused by other programs.
Functions and Modularity — Python has no switch statement; this is a deliberate design choice, not an oversight.
Functions and Procedures — Reuse means writing a set of instructions once, naming it, and calling that name multiple times throughout a program
Functions and Return Values — Python provides several methods for common list operations. Three of the most frequently used are append, extend, and sort. Each…
Functions and Routines in Java — An API is a specification of the routines a toolbox or device makes available, the parameters each one takes, and the task each…
Functions and Their Inputs and Outputs — Determinism is the property that identical inputs always produce identical outputs in a computer program—this is expected and…
Functions in C++ — Package meaningful operations as carefully named functions so that a call explains its own purpose.
Functions of Several Variables — A partial derivative measures the rate at which a multivariate function changes with respect to one of its variables.
Functions: input() and print() — Every program takes input, performs processing, and produces output—this is the fundamental structure of all programs.
Functions: Organizing and Reusing Code — Programming languages, like natural languages, have vocabulary (words that must be spelled correctly) and grammar (rules for…
Functions: Organizing Code Into Reusable Blocks — Programming is a problem-solving skill that develops through practice and feedback, not a collection of syntax rules to memorize.
Fundamental Types in C++ — Built-in types can be built from the fundamental types, the const modifier, and the declarator operators.
Fundamentals of machine learning — Deep learning became more popular through the combined influence of hardware, data, algorithms, investment, and democratization.
Future Frontiers in Reinforcement Learning Research — The section broadens reinforcement learning beyond its standard ideas by examining relationships with psychology and neuroscience.
Futures and get() in Concurrent Tasks — A promise is the sending side that matches a future's get(), providing put operations set_value() and set_exception().
Futures and promises — A packaged_task simplifies setting up tasks connected with futures and promises so that they can be run on threads.