gepa — sample-efficient reflective Pareto search
SkillSearchRuns the GEPA optimization loop (arXiv:2507.19457) — sample-efficient reflective Pareto search. A cheap train-minibatch pre-gate decides whether a proposal is worth an expensive val evaluation, and parents are sampled from a per-instance frontier so specialists survive instead of being averaged away. Use when rollouts are expensive and the scorer returns informative per-task feedback, and you want the most quality per evaluation. Use hill-climb instead for a first baseline run or for feedback-poor binary pass/fail tasks.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the gepa — sample-efficient reflective Pareto search skill
What this skill tells your AI
The instructions your AI receives, as published by skillberry-ai/cap-evolve in skills/algorithms/gepa/SKILL.md and read by ahel’s review.
algorithms/hill-climb owns the mechanics every algorithm shares: parent →
proposal → val gate → commit, specified once in
algorithms/hill-climb/references/run-step.md. Read that first. This page states
only what GEPA (Agrawal et al., 2025) does differently, and why those
differences are the paper's actual contribution rather than decoration. A thin
wrapper over cap_evolve.gepa.gepa_loop.
The two mechanisms, and why removing either turns GEPA back into hill-climb
1. The parent is sampled from per-instance winners, not from the global best. A mean is a lossy summary. A candidate that fixes one genuinely hard task while regressing three easy ones has a worse mean than the incumbent, so a best-parent rule discards it — and with it the only text in the pool that has ever solved that task. GEPA instead scores per val instance and samples frequency-weighted over candidates that (co-)win at least one, so specialists and stepping-stones stay reachable as parents while their mean is still behind. That is the quality-diversity argument (MAP-Elites): keep the set that covers the task distribution, not the single champion. Sampling is stochastic and seeded, so the exploration is reproducible.
2. A cheap train minibatch pre-gates the expensive val evaluation. Rollouts
dominate cost and a full-val eval costs |val| · n_trials of them. Most
proposals are bad; paying full price to find that out is what makes naive
reflective search unaffordable, and GEPA's headline "~35× fewer rollouts" comes
almost entirely from not paying it. So parent and child are evaluated on the
same small train minibatch (2 · minibatch-size rollouts, eval-cached) and
the child is dropped unless sum(child) > sum(parent). The minibatch never
decides acceptance — it decides whether acceptance is worth measuring.
A side benefit of (2): reflection reads train traces, so the proposer never sees the split its gate is computed on.
What differs from hill-climb, step by step
- Parent — frequency-weighted sample over per-instance (co-)winners
(
--selection-strategy, defaultpareto_per_instance), not the current best. - Signal — a minibatch of
--minibatch-size(default 4) train ids, evaluated with traces, instead of the whole train focus set. - Reflective dataset —
REFLECTION.mdin the optimizer workdir, over the parent's failing minibatch tasks (phases/diagnoseowns what one is and what shape it takes). "Failing" is the hard thresholdreward < 1.0, so with a graded scorer that never reaches 1.0 every sampled task is listed and the header always reads0/N pass— read it as "sampled tasks, worst first". Each entry is truncated to ~800 chars and at most 12 tasks are written: a summary, not an archive; untruncated rollouts stay inrollouts/train/. The prompt also carries the run's cross-iteration files (LEDGER.md,JOURNAL.md,PROCESS.md,RUNMAP.md+prior_iterations/) so a proposal builds on prior work. All five are real here since #396:harness.record_iterationwrites thestepevent those files are built from and folds the optimizer's appendedJOURNAL.mdentry back into the run-level handover, so the history accumulates across iterations. - Local gate — child on the same minibatch,
sum(child) > sum(parent), else dropped with no val spend. This is the extra stage; everything after it is hill-climb's. - Merge — every
--merge-cadenceaccepts, find two strict-frontier dominators sharing a common ancestor both beat and recombine them component-by-component (each component from whichever descendant changed it), then minibatch-gate and val-gate the result like any other child.
Note the word "frontier" covers two different sets here: the sampling pool in
step 1 is every candidate with ≥1 instance win, which can include dominated
candidates; the strict per-task Pareto frontier (selection.pareto_frontier) is
a subset of it and is what the merge and the reported frontier_size use.
Component selection
A component is one editable file of the candidate. (Unrelated to
hill-climb's --focus, which selects tasks; this selects files.)
--component-selector round_robin(default): one component per iteration, cycled, written toFOCUS.md. Small attributable changes are exactly the unit the merge can later recombine — a sprawling multi-file rewrite cannot be.--component-selector all: list every component; the optimizer may edit anywhere. Use for monolithic capabilities or genuinely cross-cutting changes.
For a single-file capability the two coincide and the merge skips gracefully
(gepa_merge_skip) rather than emitting a degenerate child.
Key hyperparameters
--max-metric-calls(default 0 = unlimited): PRIMARY budget, checked between iterations. An in-flight iteration runs to completion, so actual spend can exceed it by up to2·minibatch-size + |val|·n-trials— and a merge fires inside an accepting iteration, adding3·minibatch-size(the merge and BOTH parents, on a freshly sampled minibatch) plus a second|val|·n-trials, for a worst case of5·minibatch-size + 2·|val|·n-trials. Set it below your hard ceiling.--max-iterations(default 50): secondary cap on propose→gate iterations.--minibatch-size(default 4): train ids per cheap local gate.--n-trials(default 1): rollouts/task on the full-val eval (raise under noise so the significance gate is trustworthy). Minibatch evals are always 1 trial.--max-merges(default 2): cap on merge attempts that built a candidate — a merge rejected at either gate consumes one. A skip (no eligible pair) is free.--merge-cadence(default 3): accepts between merge attempts.--protected-paths(empty = off;default= the built-in globs): seals the eval surface (scorer/gold/tasks/tests). A child that edits one is INDECISIVE — no reward recorded, not remembered as rejected, stall counter untouched — because scoring a gold-hacking edit at all would teach the optimizer that it worked. It still chargesspent.iterations(record_iteration(..., indecisive=True)): the rollouts and the optimizer call were really spent, so the spend meter counts it and only the evidence meter does not.--workers(default 1): pools the minibatch rollouts. Only safe when the adapter'srun_targetis thread-safe.--store/--store-commit-cmd(defaultgit): where accepted candidates are committed.--gate-mode/--k-se,--no-regression,--seed: as hill-climb.--resume: rebuild pool/lineage/frontier fromgepa_state.json+ each accepted candidate's rollouts and continue the search. Preserved spend keeps the budget honest; the parent-sampling RNG stream restarts, so a resumed run is not byte-identical.
Known gaps (present tense — the shipped loop, not the paper)
Two. Re-derived against current main, because most of what this section used to
list has since been fixed in core: the hollow eval-cache reflection by #387, the
missing step record and the non-accumulating JOURNAL.md by #396
(harness.record_iteration is now the one place every algorithm ends an iteration,
and GEPA calls it), and the dirty snapshots plus the un-excluded optimizer-agent
dotfiles by #350 and #386. Check the two below before trusting them.
- The reflective dataset does not carry the task input, though
_write_reflection's own docstring says each failing task "contributes its input" (gepa.py:230). It writesAgent output/Trajectory/Feedbackand no input field (:263-267), so the optimizer sees a bad answer to a question it cannot read. Since #387 the output and trace survive an eval-cache hit, so the entry is no longer hollow — just anonymous. - If
splits.trainis empty the minibatch silently falls back to val ids (gepa.py:530), putting the gate split in front of the proposer, with no warning. Do not run GEPA with a zero-size train split.
How to run
python scripts/check.py # behavioral, offline (mock optimizer + synthetic adapter)
python scripts/run.py --run-dir .capevolve/run_X --project .capevolve/project \
--optimizer 'python .../run-optimizer/scripts/run.py --name mock --workdir {workdir} --prompt {prompt}' \
--max-metric-calls 400 --minibatch-size 4 --component-selector round_robin
Requires baseline first (reads the seed's full-val result from baseline.json).
Reports the pool, frontier_size, best candidate, accepts, merges, and
metric-calls spent.
Agent-mode loop
When orchestration_mode: agent, follow orchestrate/orchestrate §Agent-mode
loop for the shared rules, and make each round GEPA-shaped: pick the parent by
per-instance win count; sample minibatch-size train ids; evaluate parent
then child on that same minibatch and drop the child unless sum(child) > sum(parent); only then pay for a full val eval and its gate. Reflect on the
train minibatch, never on val — val is the judge, not the teacher.
References
references/concepts.md— the paper's thesis (language as a richer learning medium than a scalar), the frequency-weighted per-instance frontier, the system-aware merge and its tie-breaking, the metric-call/eval-cache accounting, and how the pieces relate to the hill-climb / skillopt siblings. Load when you need the reasoning behind a knob rather than its value. Cites arXiv:2507.19457.
Signals
- GitHub stars
- 56
- Forks
- 16
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
gepa- Source
- github.com/skillberry-ai/cap-evolve