The AI terms that
actually matter.
Clear, technical definitions of 148 key concepts — written by AI engineers, not marketers.
Large Language Model (LLM)
Large Language Models (LLMs) are advanced AI systems designed to understand, generate, and interact with human language at scale, leveraging vast amounts of text data to perform a wide range of linguistic tasks.
Read definitionLarge Vision Model (LVM)
Large Vision Models (LVMs) are advanced artificial intelligence systems designed to process, analyze, and interpret large volumes of visual data, enabling sophisticated image recognition, object detection, and segmentation tasks.
Read definitionLLaDA-Image (InclusionAI)
LLaDA-Image is a 6B-parameter open-source diffusion model from InclusionAI (Ant Group) that unifies text-to-image generation and instruction-guided image editing in a single checkpoint. Released September 4, 2026 under Apache 2.0, it ships alongside LLaDA-Image-Turbo, a 4-step distilled variant built with Twin-DMD distillation for near-real-time generation.
Read definitionLlama 4
Llama 4 is Meta's first natively multimodal, mixture-of-experts model family, released April 2025 under an open-weight community license. It ships as Llama 4 Scout (17B active, 109B total, 16 experts, 10M-token context) and Llama 4 Maverick (17B active, 400B total, 128 experts), both codistilled from a roughly 2-trillion-parameter teacher, Llama 4 Behemoth.
Read definitionLLM Weight Quantization (AWQ, GPTQ, GGUF)
LLM weight quantization is the family of post-training compression methods that store a model's weights at 3-8 bits instead of the 16 bits they were trained in, shrinking memory footprint and speeding up inference with a controlled, measurable accuracy cost. AWQ, GPTQ, and GGUF are the three most widely deployed implementations, each trading off differently between compression ratio, hardware target, and how the bit-width budget gets allocated.
Read definitionllms.txt
llms.txt is a plain-Markdown file published at a website's root that gives large language models and AI agents a short, curated map of a site's most important pages, proposed by Jeremy Howard of Answer.AI as an open convention for making websites legible to LLMs.
Read definitionLogit Lens
The Logit Lens is a mechanistic interpretability technique that reads a transformer's mind mid-computation. By projecting the hidden state at each intermediate layer through the model's final layer-norm and unembedding matrix, it reveals the model's evolving 'best guess' for the next token — showing how a prediction is refined layer by layer through the residual stream.
Read definitionLoRA Fine-Tuning
LoRA (Low-Rank Adaptation) is a parameter-efficient fine-tuning technique that freezes a pre-trained model's weights and injects small trainable low-rank matrices into key layers, enabling effective specialisation at a fraction of the compute and memory cost of full fine-tuning.
Read definitionMamba Model
The Mamba model is a groundbreaking AI architecture that achieves state-of-the-art performance across various modalities by utilizing selective state spaces for efficient, linear-time sequence modeling
Read definitionMatryoshka Embeddings
Matryoshka embeddings are vector embeddings trained so that a short prefix of the full vector (say, the first 128 of 1536 dimensions) is itself a usable, accurate embedding, letting one model serve many storage and latency budgets instead of needing a separate model per size.
Read definitionMicrosoft Agent Framework
Microsoft Agent Framework is Microsoft's unified SDK for building, orchestrating, and deploying AI agents, formed by merging AutoGen's lightweight multi-agent abstractions with Semantic Kernel's enterprise features (type safety, telemetry, middleware). It reached 1.0 in April 2026 for both Python and .NET, ships native MCP and A2A support, and adds a graph-based Workflows engine for deterministic, repeatable multi-agent orchestration.
Read definitionMiniMax M2.7
MiniMax M2.7 is an open-source, self-evolving Mixture-of-Experts AI model that autonomously participates in 30–50% of its own training workflow, featuring 230 billion total parameters, a 200K-token context window, and top-tier performance on agentic and software engineering benchmarks.
Read definitionStay ahead of the curve
Weekly newsletter on agentic AI, LLMs, and what we're building at Superteams — straight to your inbox.
Ready to ship AI in production?
We deploy fractional AI teams that deliver production-grade systems in 30–90 days. No fluff, no obligation.