dragon_status: Inspect a run

In dragonfarm: Fine-Tune Small Language Models with LoRA from R

Inspect a run

Description

dragon_status() reads the run's state. dragon_progress() returns one row per logged step. dragon_logs() returns the tail of the trainer log.

Usage

dragon_status(run)

dragon_progress(run)

dragon_logs(run, n = 50)

Arguments

run	A dragon_run.
n	Number of log lines to return.

Value

dragon_status(): a list with at least state, one of "queued", "running", "succeeded", "failed", "cancelled".

dragon_progress(): a data frame with columns step, epoch, loss, eval_loss, lr, grad_norm, elapsed_s, eta_s, and for preference runs pref_acc, reward_margin, eval_pref_acc, eval_reward_margin.

dragon_logs(): a character vector.

dragonfarm documentation built on Oct. 12, 2026, 5:07 p.m.