executing-plans
executing-plans
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available
Skills 둘러보기
생산성 및 협업 · SKILL.md
pplm_candidate_reranking
pplm_candidate_rerankingUse when generated candidates must be ranked by attribute likelihood with Dist-n diversity filtering as in PPLM BCR.
Independent GitHub snapshot · Automated classification · Original documentation governs
Skill 문서
GitHubname: pplm_candidate_reranking description: Use when generated candidates must be ranked by attribute likelihood with Dist-n diversity filtering as in PPLM BCR.
PPLM Candidate Reranking
Use this skill for PPLM BR/BCR-style candidate selection.
Inputs
- Candidate token sequences.
- Attribute scores where larger is better.
- Minimum Dist-n diversity threshold.
Outputs
- Dist-1/2/3 metrics.
- Filtered ranked list.
- Selected candidate and reason.
Workflow
- Compute Dist-n as unique n-grams divided by total n-grams.
- Filter candidates below the requested Dist threshold.
- Select the highest attribute score among surviving candidates.
- If all candidates fail diversity, select the highest score and record fallback.
Validation
Run python tests/test_candidate_reranking.py.
Limitations
Reduced proxy implementations must declare that they are not full GPT-2 345M reproduction.