vitals (version 0.1.0)

vitals_view: Interactively view local evaluation logs

Description

vitals bundles the Inspect log viewer, an interactive app for exploring evaluation logs. Supply a path to a directory of tasks written to json. For individual Task objects, use the $view() method instead.

Usage

vitals_view(dir = vitals_log_dir(), host = "127.0.0.1", port = 7576)

Value

The server object (invisibly)

Arguments

dir

Path to a directory containing task eval logs.

host

Host to serve on. Defaults to "127.0.0.1".

port

Port to serve on. Defaults to 7576, one greater than the Python implementation.

Examples

if (!identical(Sys.getenv("ANTHROPIC_API_KEY"), "")) {
  # set the log directory to a temporary directory
  withr::local_envvar(VITALS_LOG_DIR = withr::local_tempdir())

  library(ellmer)
  library(tibble)

  simple_addition <- tibble(
    input = c("What's 2+2?", "What's 2+3?"),
    target = c("4", "5")
  )

  # create a new Task
  tsk <- Task$new(
    dataset = simple_addition,
    solver = generate(chat_anthropic(model = "claude-3-7-sonnet-latest")),
    scorer = model_graded_qa()
  )

  # evaluate the task (runs solver and scorer) and opens
  # the results in the Inspect log viewer (if interactive)
  tsk$eval()

  # $eval() is shorthand for:
  tsk$solve()
  tsk$score()
  tsk$measure()
  tsk$log()
  tsk$view()

  # get the evaluation results as a data frame
  tsk$get_samples()

  # view the task directory with $view() or vitals_view()
  vitals_view()
}