Creates a properly formatted chat template for Gemma models, which use
<start_of_turn> and <end_of_turn> markers instead of ChatML format.
This function addresses compatibility issues with apply_chat_template()
when used with Gemma models.
Usage
apply_gemma_chat_template(messages, add_assistant = TRUE)
Arguments
- messages
A list of message objects, each with 'role' and 'content' fields
- add_assistant
Whether to add the assistant turn prefix (default: TRUE)
Value
A character string with properly formatted Gemma chat template
Examples
if (FALSE) { # \dontrun{
messages <- list(
list(role = "system", content = "You are a helpful assistant."),
list(role = "user", content = "Hello!")
)
formatted <- apply_gemma_chat_template(messages)
} # }