Large language model

<p>A large language model (LLM) is a <a href="/facts/Language_model/FntSpg0j">language model</a> trained with <a href="/facts/Self-supervised_learning/51WjBKrn">self-supervised</a> <a href="/facts/Machine_learning/e0w0XJTu">machine learning</a> on a vast amount of text, designed for <a href="/facts/Natural_language_processing/1hjMKsSN">natural language processing</a> tasks, especially <a href="/facts/Natural_language_generation/qudtLU6Y">language generation</a>.
</p><p>The largest and most capable LLMs are <a href="/facts/Generative_pre-trained_transformer/U6JNeDMe">generative pretrained transformers</a> (GPTs), which are largely used in <a href="/facts/Generative_artificial_intelligence/ykT3GGyT">generative</a> <a href="/facts/Chatbot/kASkMjs7">chatbots</a> such as <a href="/facts/ChatGPT/ptB1y6J5">ChatGPT</a> or <a href="/facts/Gemini_(chatbot)/Kwfa4rMW">Gemini</a>. LLMs can be <a href="/facts/Fine-tuning_(deep_learning)/9eRVyFNy">fine-tuned</a> for specific tasks or guided by <a href="/facts/Prompt_engineering/SVE53ytb">prompt engineering</a>. These models acquire <a href="/facts/Predictive_learning/9ghl8DpD">predictive power</a> regarding <a href="/facts/Syntax/6PxSV98p">syntax</a>, <a href="/facts/Semantics/1L4WwLAP">semantics</a>, and <a href="/facts/Ontology_(information_science)/SQlhl6lp">ontologies</a> inherent in human <a href="/facts/Text_corpus/bYeEQQdh">language corpora</a>, but they also inherit inaccuracies and <a href="/facts/Algorithmic_bias/vHaHsIdV">biases</a> present in the <a href="/facts/Training%2c_validation%2c_and_test_data_sets/V3mMQb2G">data</a> they are trained in.
</p>

Large language model open-in-new

Large language model