Struct ChatTemplate 

pub struct ChatTemplate { /* private fields */ }

Renders a conversation to a tokenizable string using the model’s Jinja2 chat template (from tokenizer_config.json).

Implementations

impl ChatTemplate

pub fn from_tokenizer_config(path: &Path) -> Result<Self>

Load from a tokenizer_config.json file.

pub fn from_template(template: impl Into<String>) -> Self

Build with a raw Jinja2 template string.

pub fn render( &self, messages: &[ChatMessage], add_generation_prompt: bool, ) -> Result<String>

Render a list of chat messages to a prompt string.

Set add_generation_prompt = true to append the assistant turn header (so the model knows to generate).