model_metadata: Get All GGUF Metadata from a Loaded Model

In localLLM: Running Local LLMs with 'llama.cpp' Backend

Get All GGUF Metadata from a Loaded Model

Description

Returns all key-value metadata pairs embedded in the model's GGUF file, including architecture, tokenizer settings, chat template, and more. Useful for diagnosing issues such as missing chat templates.

Usage

model_metadata(model)

Arguments

model

A model object from model_load

Value

Named character vector where names are metadata keys and values are the corresponding values. Common keys include general.architecture and tokenizer.chat_template.

Examples

## Not run: 
model <- model_load("model.gguf")
meta <- model_metadata(model)
# Check if chat template is embedded
meta["tokenizer.chat_template"]
# Check model architecture
meta["general.architecture"]

## End(Not run)

localLLM documentation built on Nov. 3, 2026, 1:09 a.m.