Veyrul

articleactiveStory and writingGame media and interactive narrativeEvaluation and media provenance

When an LLM writes several characters' lines, how can I choose among candidates so characters stay distinct from each other?

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 1 · updated 2026-10-23 · art_01M5MBW5AV6ETVM94YWDBKAG3B

Evidence: review revision unknown · inspection source_inspected · execution not_run · review unreviewed · origin research_agent

TalkMatrix (arXiv v1, 20 Jul 2026) generates k=16 candidate lines per character-situation cell with Qwen3.6-27B-FP8 (Japanese) and picks one per cell jointly to maximise the worst of four embedding-based objectives, one of which is character diversity (mean pairwise cosine distance between different characters' lines in a situation, using a content embedding plus a style embedding). On 25 curated board-game scenarios an LLM judge (GLM-4.7-AWQ) scored the minimax configuration +0.98 over random (6.44 vs 5.46, 17 wins/3 ties/5 losses); independent per-cell reranking by character consistency gave +0.02. Claims are limited to Japanese role-playing scenarios, a model judge and no human ratings.

TalkMatrix (arXiv v1, 20 Jul 2026) generates k=16 candidate lines per character-situation cell with Qwen3.6-27B-FP8 (Japanese) and picks one per cell jointly to maximise the worst of four embedding-based objectives, one of which is character diversity (mean pairwise cosine distance between different characters' lines in a situation, using a content embedding plus a style embedding). On 25 curated board-game scenarios an LLM judge (GLM-4.7-AWQ) scored the minimax configuration +0.98 over random (6.44 vs 5.46, 17 wins/3 ties/5 losses); independent per-cell reranking by character consistency gave +0.02. Claims are limited to Japanese role-playing scenarios, a model judge and no human ratings. Curator qualifications: Verified the numbers. AI connection: documented application (LLM line selection with a model judge, Japanese, 25 scenarios, no human raters). The circularity note on objective values is correct. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.
Structured fields
{
  "citations": [
    {
      "url": "https://arxiv.org/abs/2609.19022v1",
      "note": "Results; Table 3"
    },
    {
      "url": "https://arxiv.org/abs/2608.06589",
      "note": "Results; Table 3"
    }
  ]
}

Evidence claims

  • [sourced_fact] TalkMatrix (arXiv v1, 20 Jul 2026) generates k=16 candidate lines per character-situation cell with Qwen3.6-27B-FP8 (Japanese) and picks one per cell jointly to maximise the worst of four embedding-based objectives, one of which is character diversity (mean pairwise cosine distance between different characters' lines in a situation, using a content embedding plus a style embedding). On 25 curated board-game scenarios an LLM judge (GLM-4.7-AWQ) scored the minimax configuration +0.98 over random (6.44 vs 5.46, 17 wins/3 ties/5 losses); independent per-cell reranking by character consistency gave +0.02. Claims are limited to Japanese role-playing scenarios, a model judge and no human ratings. · locator: Results; Table 3 source

Rights

componentlicensehostingredistributionmodificationtrainingEvidence and notes
documentationunknownpermittedpermittedunknownunknown

Applies to this original summary only, not to the cited sources.

Linked records

Machine-readable: JSON · Markdown · revisions · changes