F

Feature

• Category: Data & representations
• What people say: A column in a dataset.
• What it actually means: An individual measurable property of the data. In classical ML, you engineer features by hand. In deep learning, the network learns features automatically from raw data.
• Common confusion: A stored column can contain several useful features, and a learned representation can contain features with no simple human label.
• Related terms: Embedding, Latent Space, Inductive Bias

Few-Shot

• Category: Prompting & context
• What people say: Give the model a few examples in the prompt.
• What it actually means: In-context learning that includes a small set of demonstrations before the target input so the model can infer the desired task, format, or decision boundary.
• Why it matters: Example quality and coverage matter more than a universal example count. Poor or contradictory demonstrations can reduce reliability.
• Related terms: Zero-Shot, In-Context Learning, Prompt Engineering, Context Window

Fine-tuning

• Category: Math & training
• What people say: Training a model on your data.
• What it actually means: Continuing training from pretrained parameters on a narrower dataset or objective. Depending on the method, you may update all parameters, selected parameters, or added adapter parameters.
• Why it matters: Fine-tuning can adapt behavior, style, format, or task performance, but it is not a dependable replacement for retrieval when facts must stay current or traceable.
• Common confusion: Fine-tuning can influence encoded knowledge, but it does not simply append records to a searchable database inside the model.
• Learn it: Fine-Tuning and LoRA
• Related terms: SFT (Supervised Fine-Tuning), LoRA (Low-Rank Adaptation), QLoRA, RAG (Retrieval-Augmented Generation)

Flaky Test

• Category: AI-native development
• What it actually means: A test that can pass and fail across equivalent runs without a relevant change to the code or intended test environment.
• Why it matters: Flakiness weakens verification gates and can train people or agents to ignore real failures or retry until they obtain a false pass.
• In practice: Preserve the failing seed and environment, quarantine only with an owner and deadline, then fix uncontrolled time, concurrency, network, order, or shared-state dependencies.
• Common confusion: A test that consistently exposes an intermittent product bug is valuable evidence, not necessarily a flaky test.
• Related terms: Regression Test, Test Oracle, Retry with Backoff, Verification Gate
• Sources: De-Flake Your Tests

FlashAttention

• Category: Infrastructure & serving
• What it actually means: An exact attention algorithm that tiles the computation to reduce transfers between accelerator memory levels while avoiding materialization of the full attention matrix in high-bandwidth memory.
• Why it matters: Attention can be limited by memory movement rather than arithmetic, especially for long sequences, so an IO-aware kernel can improve usable speed and memory efficiency.
• In practice: Use a kernel supported by the model's shapes, masks, dtype, and hardware, verify numerical tolerance, and benchmark end-to-end latency rather than quoting a paper result as a fixed multiplier.
• Common confusion: FlashAttention changes how attention is computed, not the mathematical attention result it targets. It is separate from KV caching and quantization.
• Learn it: KV Cache and Flash Attention
• Related terms: Attention, Self-Attention, KV Cache, Mixed Precision
• Sources: FlashAttention

Function Calling

• Category: Agents & tools
• What people say: A model using tools.
• What it actually means: A provider or application interface through which a model emits a structured request naming a tool and its arguments. Application code validates the request, performs the operation, and can return the result for another model step.
• Common confusion: The model requests a function call; your trusted code decides whether and how to execute it. Function calling alone is not a complete agent.
• Learn it: Function Calling
• Related terms: Structured Output, Tool Contract, Agent, MCP (Model Context Protocol)