prompting · group-candidates-execution-result-then-permute-within-groups TEC-0132 · proposed

Group candidates by execution result, then permute within gr

Replaces selector fine-tuning with retrieved structured criteria (schema grounding, SQL operations, expected output form) prepended to a listwise prompt, and mitigates position bias by shuffling only inside execution-result groups, aggregat

Proposed. Filed by the drafting stage because a paper introduces or tests it. Nobody has vouched for it. The claim drafted from that paper is its first evidence, and what the paper used to validate it is in that claim’s source note. Say whether it held up for you, below.

Replaces selector fine-tuning with retrieved structured criteria (schema grounding, SQL operations, expected output form) prepended to a listwise prompt, and mitigates position bias by shuffling only inside execution-result groups, aggregat Filed by the drafting stage from Replacing Training with Memory: Listwise Selection for Text-to-SQL (2609.00834) with status proposed: the paper introduces or tests it, and the claim drafted from the paper is its first evidence. The paper validated it by: Execution accuracy, selector calls and input tokens per query against greedy, majority voting, pointwise (Contextual-RM-32B), pairwise (CHASE-SQL style) and R3-SQL on the same candidate sets: 73.08 vs 71.97 for R3-SQL on BIRD-dev at n=32, +2.02 over pairwise, 2.92x fewer tokens; ablations drop 0.55 (no memory) and 0.32 (no permutation); order-consistency rises 18.98 -> 23.04 (global shuffle) -> 29.04 (group-based) -> 33.17 (with tie-break).. Nobody has vouched for it; add support or contest it on this page.

Addresses: Writing a reusable locator: regex, SQL, XPath

Does it work?

nothing measured0 supporting · 0 contesting sources

Efficacy claims — what this technique actually moves, under which conditions, and whether that has been contested.

No efficacy claim filed yet. The technique is catalogued; whether it moves the capability, and when, is a separate assertion that needs its own sources.

No search recorded either, so this says nothing about the literature — only that nobody has looked here yet.

Code

No repository linked yet. Contribute one.

Sources

Either way, bring evidence: a link, or what you saw.

Whether a technique works is a claim. What you file here becomes a claim about this technique under the capability it addresses, with your evidence as its source, and moves the standing above the same way a paper would.