{
  "abstract": "Aggregating several LLM responses to the same question can improve accuracy, but every method that does so must decide where the responses should differ. Established test-time scaling methods such as self-consistency vary only the reasoning path, resampling from a fixed input. The input modification offers several further options: the wording of the question can be rephrased, the spelling can be perturbed, and the accompanying images can be transformed. We study Test-Time Augmentation (TTA), which aggregates predictions across transformed versions of the input, and use it to ask where diversity is best injected. Comparing these options across six benchmarks at a matched number of inference calls, we find that where the diversity comes from matters more than how much of it there is. Meaning-preserving paraphrases of the input yield the largest accuracy gains (+1.80 pp over a single call) and reach peak accuracy with the fewest samples. Yet, they are not the most diverse: character-level noise makes the responses disagree far more often (37.8% versus 27.4%) while gaining less accuracy, and perturbing two modalities at once falls below the single-call baseline. Not all diversity is equally useful: perturbations that keep the meaning intact produce disagreement that aggregation can use, while those that corrupt it produce disagreement that is wrong. Overall, our results suggest that for mid-tier LLMs such as Claude Haiku, varying how a question is phrased is a better use of an inference call than varying the reasoning path. Our TTA implementation is available at github.com/aws-samples/sample-genai-reflection-for-bedrock.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2608.09351",
  "author_total": 3,
  "authors": [
    {
      "name": "Nikita Kozodoi"
    },
    {
      "name": "Zainab Afolabi"
    },
    {
      "name": "Jack Butler"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/fkg6qdksxh1cw.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2608.09351",
  "fulltext_url": null,
  "id": "fkg6qdksxh1cw",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2608.09351",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Topic Modeling"
    ]
  ],
  "openalex": "W7202197917",
  "pdf_url": "https://arxiv.org/pdf/2608.09351",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-19",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/fkg6qdksxh1cw.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:33:54Z",
      "native_id": "2608.09351",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2608.09351"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "Test-Time Augmentation for LLMs: Where to Inject Diversity at Inference Time",
  "title_text": "Test-Time Augmentation for LLMs: Where to Inject Diversity at Inference Time",
  "type": "preprint",
  "updated_at": "2026-10-30T12:33:54Z",
  "url": "https://research.draconitic.com/paper/fkg6qdksxh1cw",
  "venue": "arXiv",
  "versions": []
}