Language Model
A language model is a model of patterns in language that assigns probabilities or comparable scores to linguistic sequences. Its units may be characters, words, subwords, bytes, or other tokens.
Read the overviewThe encyclopedia of artificial intelligence
Understand AI.
Explore the encyclopedia, compare models, and build practical skills.
CtrlK
Explore all learning paths
Worth exploring
An in-depth explainer and new additions to the encyclopedia.
New to the wiki
KV cache offloading
KV cache offloading is the practice of moving part or all of a Transformer model's KV cache out of accelerator memory (GPU HBM) into a larger, slower tier such as CPU DRAM, local NVMe storage, or remote storage, and then bringing back, or recomputing
Fact-checked Sep 16, 2026Diogo Almeida
Diogo Almeida (full name Diogo Moitinho de Almeida) is a machine learning researcher and the co-founder and chief executive of TypeSafe AI, a San Francisco lab that came out of stealth on 15 September 2026 with about $40 million in seed funding led by DCVC…
Fact-checked Sep 16, 2026Put AI knowledge to work
All toolsCompare models
Which model fits my needs?
Compare capabilities, context limits and deployment options. Build a shortlist to test on your own tasks.
Open tool API cost plannerWhat will my API workload cost?
Estimate costs from your request volume and token usage. Check the pricing assumptions behind each estimate.
Open tool VRAM calculatorCan my hardware run this model?
Estimate memory for weights, context and runtime overhead. Actual memory use depends on your setup.
Open toolRecently revised
Explore an updated article and check its review status.
Calibration (machine learning)
Calibration in machine learning is the property that the probability scores produced by a probabilistic classifier match the empirical frequency of the predicted event: a model that assigns a confidence of 0.8 to a set of inputs should be correct on about 80…
Fact-checked Sep 16, 2026Structured output
Structured output is a set of techniques and API features that constrain a large language model (LLM) to emit responses that exactly conform to a predefined format or schema, such as JSON, XML, or a custom grammar, instead of free-form text.
Fact-checked Sep 16, 2026DeepSeek Sparse Attention (DSA)
DeepSeek Sparse Attention (DSA) is a trainable, fine-grained sparse attention mechanism introduced by the Chinese AI company DeepSeek in its experimental model DeepSeek-V3.2-Exp, released on September 29, 2025.
Fact-checked Sep 16, 2026
Provider status
Checking official OpenAI and Claude status sources.
Checking now
How this encyclopedia is checked
828 of 4,546Independently fact-checked18.2% so far, most-linked articles first. We publish the gap, not just the total.Cite it or cut itSourcing standard4,544 of 4,546 articles carry inline source markers.Found an error?Corrections are public1 resolved report; 114.6h median. Too few reports to establish a typical response time.
Help improve AI WikiSuggest a correction, add a source, or research a missing topic.
Help build AI Wiki
Fix one fact, add a source, or propose a missing article.
Requested by readers
Missing pages that people or automated systems tried to open.
Missing from existing articles
Topics already linked from other AI Wiki articles.
- 1Automation10 linking articles
- 2Open Compute Project9 linking articles
- 3OpenTelemetry9 linking articles
Found something inaccurate? Review corrections or use the report button on any article.
Repositories to research
No matching article was recorded when these repositories were detected. Open the original repository and search the wiki before proposing a page.
nvidia/foundationposenvidia/c-foundationstereo-stencent/Simple-Attention-Sparsificationtencent/EVIE-4.5B