The specific string used within the LLM Interface package to identify and access a provider's API (e.g., 'openai', 'cohere', 'anthropic').
An optional object interfaceOptions that can be used to control LLMInterface.
This Node.js module that provides a standardized way to interact with various LLM providers' APIs.
An optional object options that can be used to send parameters to a LLM.
An object that is compatible with OpenAI chat.completion. The object can contain the model, message, and parameters.
Example:
{
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "Say this is a test!"}],
"temperature": 0.7
}The company or organization that offers a language model API (e.g., OpenAI, Cohere, Anthropic).
A set of rules, protocols, and tools for building software applications. In the context of LLMs, an API allows you to interact with a language model hosted by a provider.
Numerical representations of text that capture their meaning and semantic relationships. Embeddings can be used for tasks like semantic search, clustering, and recommendation.
Specific actions or operations that can be performed by an LLM. The LLM Interface may allow you to define custom functions or utilize pre-defined functions offered by providers.
The process of generating a response from an LLM based on a given prompt.
A type of artificial intelligence model trained on a massive dataset of text and code. LLMs can generate text, translate languages, write different kinds of creative content, and answer your questions in an informative way.
The specific string used to reference a LLM offered by a provider (e.g., 'gpt-3.5-turbo', 'command-nightly').
A communication mode where the LLM directly parses and responds with JSON objects, facilitating structured data exchange and function calling.
A value that the model uses during inference, typically referring to the adjustable weights in the model that were learned during training. Supported parameters vary between providers and models, please check the provider documentation for a list of supported parameters.
The input given to an LLM, which can be a question, instruction, or piece of text.
The output generated by an LLM in response to a prompt.
A feature that allows you to receive the LLM's response as it is being generated, rather than waiting for it to be fully completed.
A parameter that controls the randomness of the LLM's outputs. Lower values make the model more deterministic, while higher values increase randomness.
The basic unit of text used by LLMs. Tokens can be words, subwords, or even characters, depending on the specific model. Token usage can affect pricing in some LLM APIs.
External resources or functionalities (e.g., calculators, code interpreters, search engines) that can be integrated with LLMs to enhance their capabilities.
A decoding strategy where the model considers only the top k most probable next tokens when generating text, promoting diversity in the output.
A decoding strategy where the model considers the smallest set of tokens whose cumulative probability exceeds a certain threshold p, balancing between diversity and relevance.