{
  "schema_version": "1.1",
  "id": "s13:https://arxiv.org/abs/2608.05138v1",
  "slug": "2608-05138v1-0bvu6le",
  "url": "https://feed7.dev/p/2608-05138v1-0bvu6le",
  "title": "Teaching Nemotron Greek: Mining a Corpus, Adapting Retrieval, and Grounding Generation for Modern Greek across Specialist Domains",
  "why_included": "A Greek RAG study shows generic multilingual embeddings can lose to BM25 in specialist domains, while targeted adaptation sharply improves retrieval and grounded answers. Language-specific evaluation still matters.",
  "summary": "The authors trained a Nemotron 1B embedder on **65,773 Greek retrieval pairs**, raising nDCG@10 from **0.362 to 0.835**. A LoRA-tuned Nemotron 30B-A3B reader raised judged answer correctness from **29.4% to 66.9%**.",
  "practical_implication": "For specialized or lower-resource languages, benchmark a lexical baseline before choosing a dense retriever, then adapt retrieval, reranking and generation as separate stages. The released **HERA benchmark** provides a Greek RAG test bed.",
  "agent_context": "The authors trained a Nemotron 1B embedder on **65,773 Greek retrieval pairs**, raising nDCG@10 from **0.362 to 0.835**. A LoRA-tuned Nemotron 30B-A3B reader raised judged answer correctness from **29.4% to 66.9%**.\n\nFor specialized or lower-resource languages, benchmark a lexical baseline before choosing a dense retriever, then adapt retrieval, reranking and generation as separate stages. The released **HERA benchmark** provides a Greek RAG test bed.\n\nBM25 beat several off-the-shelf multilingual dense models on the specialist corpora, and the adapted embedder's advantage over BM25 remained domain-dependent. The reported evaluation is specific to Modern Greek and the studied legal, energy, financial and medical material.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2608.05138v1",
    "published_at": "2026-08-05T17:56:40.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "context",
  "domains": [
    "research",
    "data"
  ],
  "topics": [
    "retrieval",
    "context-engineering",
    "model-selection"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "BM25 beat several off-the-shelf multilingual dense models on the specialist corpora, and the adapted embedder's advantage over BM25 remained domain-dependent. The reported evaluation is specific to Modern Greek and the studied legal, energy, financial and medical material."
  ],
  "connected_context": {
    "meaning": "This confirms that multilingual dense retrieval is not automatically superior to a lexical baseline in specialist, lower-resource corpora. It turns RAG adaptation into a staged selection problem across retrieval, reranking, and generation, with HERA offering a focused test bed. The domain-dependent margin over BM25 narrows the result to workload-specific evaluation rather than a general prescription for Greek or multilingual RAG.",
    "corpus_size": 368,
    "generated_at": "2026-08-06T10:06:10.366Z",
    "connections": [
      {
        "title": "How Temperature Shapes Ideological Discourse in Retrieval-Augmented Generation?",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2607.11783v1",
        "feed7_url": "https://feed7.dev/p/2607-11783v1-170rjdc",
        "reason": "The Greek study separates retrieval and generation adaptation, while the ideology study shows decoding can alter how retrieved material appears in answers; together they support evaluating source selection and generation settings jointly."
      },
      {
        "title": "AI on Your Lakehouse: Context Comes in Shapes, Not Queries — Zach Blumenfeld, Neo4j",
        "source_name": "AI Engineer",
        "source_url": "https://www.youtube.com/watch?v=kRkcNOsRyYg",
        "feed7_url": "https://feed7.dev/p/ai-on-your-lakehouse-context-comes-in-shapes-not-queries-zach-blumenfeld-0r4s9u5",
        "reason": "Both reject a universal retrieval method: the Greek results require lexical and dense systems to be compared by domain, while graph-shaped context is another retrieval surface that must be benchmarked on the actual data."
      },
      {
        "title": "Citation Needed: Provenance for LLM-Built Knowledge Graphs — Daniel Chalef, Zep AI",
        "source_name": "AI Engineer",
        "source_url": "https://www.youtube.com/watch?v=H7puB0RwJMM",
        "feed7_url": "https://feed7.dev/p/citation-needed-provenance-for-llm-built-knowledge-graphs-daniel-chalef-1iob5t8",
        "reason": "Stage-specific RAG evaluation identifies which component improves correctness, while provenance supplies the complementary mechanism needed to trace which specialist sources support generated claims."
      }
    ]
  },
  "lifecycle": "Current",
  "published_at": "2026-08-05T17:56:40.000Z",
  "modified_at": "2026-08-05T17:56:40.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2608-05138v1-0bvu6le",
    "json": "https://feed7.dev/p/2608-05138v1-0bvu6le.json",
    "markdown": "https://feed7.dev/p/2608-05138v1-0bvu6le.md"
  }
}