output_as.RdConfigure a model to adhere to specific output constraints.
output_as(x, format, ...)The Agent object
A TextFormat object, or something convertible to one, representing the output constraints. For anything more complex than a string, the output will typically need to be JSON, with constraints expressed as JSON schema. While it is possible to pass a list directly encoding a JSON schema, it is usually more convenient to pass a stubbed data.frame (for data.frame outputs) or an S7 class, in which case the output is an instance of the class.
Arguments passed to convert when converting format to
a TextFormat.
The input Agent object except with new output constraints.
instruct, for generally instructing a
model. system_prompt_as for more advanced system prompt
construction.
if (FALSE) { # \dontrun{
agent <- openai_agent("gpt-4o-mini", temperature = 0) |>
instruct("Answer questions about this dataset:", mtcars)
# Return a filtered subset of mtcars, which serves as the prototype
filtered_agent <- agent |>
output_as(mtcars)
filtered_agent |>
predict("Cars with mpg > 20")
# Return a summary data.frame
summary_agent <- agent |>
output_as(data.frame(
cyl = integer(),
avg_mpg = numeric(),
avg_hp = numeric()
))
summary_agent |>
predict("Average mpg and hp by number of cylinders.")
# Using class_data.frame for more general output
agent |>
output_as(S7::class_data.frame) |>
predict("Top 5 most fuel efficient cars including mpg, hp, and wt.")
# Using a custom S7 class for more specialized data
Person <- S7::new_class("Person", properties = list(
first_name = S7::class_character,
last_name = S7::class_character
))
openai_agent(temperature = 0) |>
output_as(Person) |>
predict("Creators of R")
} # }