Skip to contents

Run stage 2 - LLM adjudication of the MAYBE queue

Source: R/stages.R
np_stage2_run.Rd

Resumable, because an adjudicator sits in the middle:

Usage

np_stage2_run(
  project = np_project_root(),
  shard_size = 250L,
  columns = .np_slim_cols,
  verbose = TRUE
)

Arguments

project

Project root. Defaults to np_project_root().

shard_size

Source ids per adjudication shard. Default 250.

columns

Columns carried into the shard files. Defaults to the adjudication set; pass more to give the agent extra context.

verbose

Print progress. Default TRUE.

Value

Invisibly: the outcome frame once decisions exist, otherwise the manifest of shards written.

Details

  • First call — reads stage1_maybe and stage1_k_candidates, writes shard files to 02_stage2/batches/slim/, and stops. Point an agent at 02_stage2/PROMPT.md; it writes one decision-<shard>.csv per shard to 02_stage2/batches/decisions/.

  • Second call — collects those decisions, collapses duplicate source registrations, and writes stage2_yes / stage2_no.

Re-running after only some shards are adjudicated collects what is there and reports which shards are still outstanding. An unreturned shard is a hole in the results, not a set of NOs, so it is never silently treated as one.

See also

np_stage1_run(), np_merge_decisions(), np_stage_schema().