Skip to main content
Back to Tags

LLM

139 items tagged with "llm"

Filter by type:

Best Practices14

Best Practice

OpenAI Safety & Alignment Best Practices

Mitigation strategies (RLHF, red-teaming, tiered access) for large language model deployment.

Best Practice

Retrieval-Augmented Generation (RAG) Best Practices

RAG grounds a large language model in external documents retrieved at query time, reducing hallucination and letting answers reflect current, private data without retraining the model.

Best Practice

Prompt Engineering Best Practices

Prompt engineering is the practice of designing clear instructions, examples, and structure so a large language model returns accurate, consistent, and useful output.

Best Practice

LLM Evaluation and Evals

LLM evaluation measures model and application quality with repeatable tests, scoring accuracy, faithfulness, safety, and cost so teams can ship and improve with evidence.

Best Practice

LLM Guardrails

LLM guardrails are programmatic checks on model inputs and outputs that enforce safety, format, topic, and policy rules, blocking or correcting unsafe or off-policy responses.

Best Practice

AI Red Teaming

AI red teaming is structured adversarial testing of AI systems to find harmful, biased, or insecure behavior before attackers or real users do, using crafted attacks and probes.

Best Practice

Model Context Protocol (MCP)

The Model Context Protocol is an open standard that lets AI applications connect to external tools and data sources through a uniform client-server interface.

Best Practice

AI Agent Design Patterns

AI agent design patterns are reusable structures for LLM systems that plan, use tools, and act over multiple steps, covering reflection, tool use, planning, and multi-agent collaboration.

Best Practice

LLM Observability

LLM observability is the practice of tracing, logging, and measuring LLM applications in production to monitor quality, cost, latency, and safety and to debug failures.

Best Practice

OWASP Top 10 for LLM Applications (2025)

The OWASP Top 10 for LLM Applications lists the most critical security risks for generative AI systems, including prompt injection, sensitive data disclosure, and supply chain risk.

Best Practice

Prompt Injection Defense

Prompt injection defense protects LLM applications from attacks that hide malicious instructions in user input or retrieved content to override the system's intended behavior.

Best Practice

LLM Cost Optimization

LLM cost optimization reduces the spend of running language model applications through model selection, caching, prompt efficiency, and token-aware design without sacrificing quality.

Best Practice

Fine-Tuning vs RAG Decision Framework

A decision framework for choosing between fine-tuning, RAG, or both, based on whether the goal is new knowledge, consistent behavior, freshness, or domain adaptation.

Best Practice

Hallucination Mitigation

Hallucination mitigation reduces confident but false LLM output through grounding, retrieval, citation, verification, and uncertainty handling so answers can be trusted.

Reference Architectures9

Reference Architecture

Vector Search Platform for RAG

A vector search platform on Azure powering retrieval-augmented generation with embeddings, a vector database, and hybrid search.

Reference Architecture

Retrieval-Augmented Generation Platform on AWS

A reference design for a production RAG system on AWS that grounds LLM answers in private documents using a managed vector store and Bedrock models.

Reference Architecture

LLM Gateway and Proxy on Kubernetes

A reference design for a self-hosted LLM gateway on Kubernetes that centralizes routing, rate limiting, cost tracking, and guardrails across multiple model providers.

Reference Architecture

Multi-Agent LLM System on Azure

A reference design for a multi-agent application on Azure where specialized LLM agents coordinate through an orchestrator and shared tools to complete complex tasks.

Reference Architecture

LLM Fine-Tuning Pipeline on GCP

A reference design for fine-tuning open LLMs on GCP using Vertex AI custom training, parameter-efficient methods, and an evaluation gate before deployment.

Reference Architecture

LLM Observability Stack on Kubernetes

A reference design for an LLM observability platform on Kubernetes that captures traces, evaluations, cost, and quality signals across AI applications.

Reference Architecture

LLM Guardrails and Safety Layer on AWS

A reference design for an LLM safety layer on AWS that filters inputs and outputs, blocks prompt injection, and enforces content and PII policies.

Reference Architecture

Self-Hosted Open LLM Inference on Kubernetes

A reference design for serving open-weight LLMs on Kubernetes with GPU autoscaling, continuous batching, and an OpenAI-compatible API.

Reference Architecture

Conversational AI Assistant on Azure

A reference design for an enterprise conversational assistant on Azure that combines RAG, tool calling, and memory behind a governed orchestration layer.

Comparisons10

Comparison

RAG vs Fine-Tuning

Retrieval-augmented generation injects external knowledge at query time; fine-tuning bakes behavior into model weights. They solve different problems and often combine.

Comparison

Fine-Tuning vs Prompt Engineering

Prompt engineering steers a model with instructions and examples in the context; fine-tuning changes the weights. Cost, control, and durability differ sharply.

Comparison

GPT vs Claude

OpenAI's GPT and Anthropic's Claude are leading proprietary LLM families. They differ in design philosophy, context handling, and integration ecosystems rather than raw capability tier.

Comparison

Open-Weight vs Proprietary LLMs

Open-weight models can be downloaded and self-hosted; proprietary models are accessed via API. The split shapes control, cost, privacy, and capability ceilings.

Comparison

Llama vs Mistral

Llama (Meta) and Mistral are two leading open-weight LLM families. They differ in licensing, model sizes, mixture-of-experts use, and ecosystem maturity.

Comparison

vLLM vs TGI

vLLM and Hugging Face Text Generation Inference (TGI) are high-throughput LLM serving engines. Both optimize GPU inference but differ in ecosystem and tuning focus.

Comparison

LangChain vs LlamaIndex

LangChain is a broad framework for LLM application orchestration; LlamaIndex specializes in data ingestion and retrieval for RAG. They overlap but optimize for different centers of gravity.

Comparison

Hugging Face vs OpenAI API

Hugging Face provides open models, tooling, and self-hosting paths; the OpenAI API offers managed access to proprietary models. The choice trades control against convenience.

Comparison

Quantization vs Full Precision

Quantization stores model weights at lower bit-widths to cut memory and speed inference; full precision preserves maximum accuracy. The trade-off is size and speed versus fidelity.

Comparison

AI Agents vs Workflows

Agentic systems let an LLM decide its own steps and tool use; workflows orchestrate LLMs through predefined paths. The choice trades flexibility against predictability.

Benchmarks39

Benchmark

MMLU (Massive Multitask Language Understanding)

A 57-subject multiple-choice benchmark testing broad academic and professional knowledge across STEM, humanities, social sciences, and law.

Benchmark

MMLU-Pro

A harder, reasoning-focused successor to MMLU with ten answer options and tougher questions designed to separate frontier models that saturated the original.

Benchmark

GSM8K (Grade School Math 8K)

A benchmark of ~8,500 grade-school math word problems that test multi-step arithmetic reasoning with a single numeric answer.

Benchmark

MATH (Competition Mathematics)

A benchmark of 12,500 competition-style math problems across algebra, geometry, number theory, and calculus, graded on exact final-answer match.

Benchmark

HumanEval

A code-generation benchmark of 164 Python programming problems graded by executing unit tests, popularizing the pass@k metric.

Benchmark

MBPP (Mostly Basic Python Problems)

A benchmark of ~1,000 entry-level Python programming tasks with test cases, used to evaluate basic code synthesis from short descriptions.

Benchmark

SWE-bench

A benchmark of real GitHub issues from open-source Python repositories where a model must produce a patch that resolves the issue and passes tests.

Benchmark

SWE-bench Verified

A 500-task, human-validated subset of SWE-bench with clear specifications and reliable tests, used as the standard clean measure of agentic coding.

Benchmark

BIG-bench (Beyond the Imitation Game)

A collaborative suite of 200+ diverse tasks probing reasoning, knowledge, and emergent abilities beyond conventional language benchmarks.

Benchmark

BBH (BIG-bench Hard)

A 23-task subset of BIG-bench focused on challenging multi-step reasoning where chain-of-thought prompting yields large gains.

Benchmark

HellaSwag

A commonsense sentence-completion benchmark where models pick the most plausible continuation among adversarially generated distractors.

Benchmark

ARC (AI2 Reasoning Challenge)

A grade-school science question benchmark split into Easy and Challenge sets, the latter built from questions retrieval methods answer incorrectly.

Benchmark

TruthfulQA

A benchmark measuring whether models avoid generating false answers that mimic common human misconceptions and falsehoods.

Benchmark

GPQA (Graduate-Level Google-Proof Q&A)

A benchmark of expert-written, graduate-level science questions designed to be extremely hard even with web access, testing deep domain reasoning.

Benchmark

MMMU (Massive Multi-discipline Multimodal Understanding)

A multimodal benchmark of college-level questions requiring joint reasoning over text and images such as diagrams, charts, and figures.

Benchmark

MT-Bench

A multi-turn conversational benchmark where a strong LLM judge scores model responses across categories on a 1-10 quality scale.

Benchmark

Chatbot Arena

A live, crowdsourced evaluation where users compare two anonymous model responses and votes are aggregated into Elo-style rankings.

Benchmark

HELM (Holistic Evaluation of Language Models)

A standardized framework evaluating language models across many scenarios and multiple metrics including accuracy, robustness, fairness, and efficiency.

Benchmark

DROP (Discrete Reasoning Over Paragraphs)

A reading-comprehension benchmark requiring discrete operations like addition, counting, sorting, and comparison over passage content.

Benchmark

WinoGrande

A large-scale commonsense benchmark of pronoun-resolution sentence pairs designed to require world knowledge rather than lexical cues.

Benchmark

AGIEval

A benchmark built from human standardized exams such as college entrance, law, and civil-service tests to measure human-centric reasoning.

Benchmark

LiveCodeBench

A contamination-resistant coding benchmark that continuously collects new competitive-programming problems and evaluates by execution over time.

Benchmark

AIME (Competition Math Benchmark)

An olympiad-level math benchmark using American Invitational Mathematics Examination problems with integer answers, a key frontier reasoning test.

Benchmark

tau-bench (Tool-Agent-User Benchmark)

An agentic benchmark testing tool-using models in simulated customer-service dialogues that require following domain policies and calling APIs correctly.

Benchmark

Terminal-Bench

An agentic benchmark evaluating models on completing real command-line tasks inside a sandboxed terminal, verified by automated checks.

Benchmark

RULER (Long-Context Benchmark)

A synthetic long-context benchmark with configurable tasks measuring a model's effective context length beyond simple retrieval.

Benchmark

IFEval (Instruction-Following Eval)

A benchmark that measures whether LLMs follow precise, verifiable formatting and content instructions using automatic checks rather than subjective judgment.

Benchmark

BFCL (Berkeley Function-Calling Leaderboard)

A benchmark for evaluating how accurately LLMs select, call, and parameterize functions and tools, including parallel, multiple, and multi-turn calling scenarios.

Benchmark

MGSM (Multilingual Grade School Math)

A multilingual extension of grade-school math word problems that tests whether LLMs can reason through arithmetic in many languages, not just English.

Benchmark

SimpleQA

A factuality benchmark of short, fact-seeking questions with single verifiable answers, designed to measure how often LLMs are correct, wrong, or appropriately abstain.

Benchmark

FRAMES (Factuality, Retrieval, And reasoning MEasurement Set)

A benchmark for retrieval-augmented generation that tests end-to-end factuality, multi-document retrieval, and multi-hop reasoning on questions needing several sources.

Benchmark

MuSR (Multistep Soft Reasoning)

A benchmark of long natural-language narratives requiring multistep commonsense and logical reasoning, such as murder mysteries and object-placement puzzles.

Benchmark

LiveBench

A contamination-resistant benchmark that continuously refreshes questions from recent sources and grades automatically against objective ground truth across many task categories.

Benchmark

BigCodeBench

A code-generation benchmark for realistic programming tasks that require composing many library calls, evaluated with rigorous test suites and high branch coverage.

Benchmark

CRUXEval (Code Reasoning, Understanding, and Execution)

A benchmark that tests whether models can reason about code execution by predicting function inputs from outputs and outputs from inputs.

Benchmark

RewardBench

A benchmark for reward models and LLM judges that measures how well they prefer better responses over worse ones across chat, reasoning, safety, and refusal cases.

Benchmark

AlpacaEval

An automated LLM-as-judge benchmark that estimates a model's win rate against a reference model on open-ended instructions, with a length-controlled variant to reduce verbosity bias.

Benchmark

Arena-Hard

An automatic benchmark of challenging, real-user-derived prompts graded by an LLM judge, built to align closely with human preference rankings and separate strong models.

Benchmark

HarmBench

A standardized red-teaming benchmark that measures how often automated attacks elicit harmful behaviors from LLMs and how well refusal and defenses hold up.

FAQs14

FAQ

What is a large language model (LLM)?

A large language model is a neural network trained on vast amounts of text to predict the next token in a sequence, which lets it generate and underst...

FAQ

What is retrieval-augmented generation (RAG)?

Retrieval-augmented generation is a technique that supplements a language model with relevant documents fetched at query time, rather than relying onl...

FAQ

What is the difference between fine-tuning and RAG?

Fine-tuning updates a model's weights by training it further on domain-specific examples, changing how the model behaves and what style or skills it h...

FAQ

What is fine-tuning a model?

Fine-tuning is the process of continuing to train a pre-trained model on a smaller, task-specific dataset so it adapts to a particular domain, style, ...

FAQ

What is a token in the context of LLMs?

A token is the basic unit of text that a language model reads and produces, typically a word fragment, whole word, or punctuation mark rather than a s...

FAQ

What is a context window?

A context window is the maximum number of tokens a language model can consider at once, covering both the input prompt and the generated output. If a ...

FAQ

What is prompt engineering?

Prompt engineering is the practice of designing the instructions, examples, and context given to a language model to get reliable, accurate outputs. T...

FAQ

What is an AI agent?

An AI agent is a system that uses a language model to decide and take actions toward a goal, rather than producing a single response. It typically ope...

FAQ

What is hallucination in LLMs?

Hallucination is when a language model produces text that sounds confident and plausible but is factually wrong or unsupported by its sources. It happ...

FAQ

What is temperature in LLM generation?

Temperature is a parameter that controls the randomness of a language model's output by scaling the probability distribution over the next token. A lo...

FAQ

What is a transformer architecture?

The transformer is a neural network architecture, introduced in 2017, that processes sequences using a self-attention mechanism instead of recurrence....

FAQ

What is quantization in machine learning?

Quantization reduces the numerical precision of a model's weights and activations, for example from 32-bit floating point to 8-bit or 4-bit integers, ...

FAQ

What is the Model Context Protocol (MCP)?

The Model Context Protocol is an open standard that defines how AI applications connect to external tools, data sources, and services in a consistent ...

FAQ

What is chain-of-thought prompting?

Chain-of-thought prompting asks a language model to work through a problem step by step before giving a final answer, rather than responding immediate...

Glossaries20

Glossary

Large Language Model (LLM)

A large language model is a neural network trained on vast text corpora to predict the next token, enabling it to generate and understand natural language across many tasks.

Glossary

Token

A token is the basic unit of text an LLM processes, typically a word fragment, whole word, or character, produced by a tokenizer and mapped to a numeric ID.

Glossary

Tokenization

Tokenization is the process of splitting raw text into tokens that a model can map to numeric IDs, usually using a subword algorithm such as byte-pair encoding.

Glossary

Context Window

The context window is the maximum number of tokens a language model can consider at once, covering both the input prompt and the generated output.

Glossary

Embedding

An embedding is a dense numeric vector that represents the meaning of text, an image, or other data so that similar items sit close together in vector space.

Glossary

Vector Search

Vector search finds items whose embeddings are closest to a query embedding, enabling semantic retrieval by meaning rather than exact keyword match.

Glossary

Retrieval-Augmented Generation (RAG)

Retrieval-augmented generation is a technique that retrieves relevant documents at query time and supplies them to an LLM as context so its answers are grounded in external data.

Glossary

Fine-Tuning

Fine-tuning is the process of further training a pretrained model on a smaller, task-specific dataset to specialize its behavior, style, or domain knowledge.

Glossary

Prompt Engineering

Prompt engineering is the practice of designing and refining the text instructions given to a language model to steer its output toward accurate, useful results.

Glossary

Inference

Inference is the process of running a trained model on new inputs to produce outputs, as opposed to the training phase that creates the model.

Glossary

Training

Training is the process of adjusting a model's parameters from data so it learns to perform a task, typically by minimizing a loss function with gradient descent.

Glossary

Transformer

A transformer is a neural network architecture built around self-attention that processes sequences in parallel, forming the basis of modern large language models.

Glossary

Attention Mechanism

An attention mechanism lets a model weigh the relevance of different parts of its input when producing each output, focusing on the most pertinent tokens.

Glossary

Hallucination

Hallucination is when a language model generates fluent, confident output that is factually incorrect, fabricated, or unsupported by its sources.

Glossary

Temperature (LLM Sampling)

Temperature is a sampling parameter that scales an LLM's output probabilities, controlling how random or deterministic its token choices are.

Glossary

Top-p Sampling (Nucleus Sampling)

Top-p sampling restricts an LLM's next-token choice to the smallest set of tokens whose cumulative probability exceeds a threshold p, then samples from that set.

Glossary

Quantization

Quantization reduces the numeric precision of a model's weights and activations, shrinking memory use and speeding inference with limited accuracy loss.

Glossary

AI Agent

An AI agent is a system that uses a language model to plan and take actions toward a goal, calling tools, observing results, and iterating with limited human input.

Glossary

Model Context Protocol (MCP)

The Model Context Protocol is an open standard that defines how AI applications connect to external tools, data sources, and prompts through a uniform interface.

Glossary

Foundation Model

A foundation model is a large model pretrained on broad data at scale that can be adapted, through fine-tuning or prompting, to a wide range of downstream tasks.