Teaching Nemotron Greek: Mining a Corpus, Adapting Retrieval, and Grounding Generation for Modern Greek across Specialist Domains
A Greek-tuned Nemotron RAG stack beat its unadapted retrieval baseline by a wide margin on specialist corpora.
The paper adapts Nemotron for Modern Greek across mining, synthetic supervision, retrieval, reranking, and grounded generation. Fine-tuning a 1B embedder on 65,773 Greek retrieval pairs raised nDCG@10 from 0.362 to 0.835. A LoRA-tuned Nemotron 30B-A3B reader lifted judged answer correctness from 29.4% to 66.9%, with better faithfulness and citations. The authors also release HERA, a large-scale Greek RAG benchmark, along with adapted Sophea Nemo RAG models. ArXiv · AI/CL/LG's note
The paper adapts Nemotron for Modern Greek across mining, synthetic supervision, retrieval, reranking, and grounded generation. Fine-tuning a 1B embedder on 65,773 Greek retrieval pairs raised nDCG@10 from 0.362 to 0.835. A LoRA-tuned Nemotron 30B-A3B reader lifted judged answer correctness from 29.4% to 66.9%, with better faithfulness and citations. The authors also release HERA, a large-scale Greek RAG benchmark, along with adapted Sophea Nemo RAG models. ArXiv · AI/CL/LG's note
score 4