Entity · technique

Retrieval-Augmented Fine-Tuning

techniqueactiveprovisionalretrieval-augmented-fine-tuning-e49ab2eb·1 events·first seen 23h ago

Aliases: Retrieval-Augmented Fine-Tuning

Co-occurring entities

U.S. Nuclear Regulatory Commission BM25 Google Gemma 4 Retrieval-Augmented Generation NRC Generic Fundamentals Examination

More like this (12)

Learning to Reason by Analogy via Retrieval-Augmented Reinforcement Fine-Tuning reinforcement fine-tuning supervised fine-tuning Parameter-Efficient Fine-Tuning Reference-Augmented Training Super-Tuning: From Activation-Aware Pruning to Sparse Fine-Tuning Retrieval-Augmented Generation Chain-of-Thought Fine-Tuning fine-tuning OpenAI Fine-Tuning behavioral fine-tuning importance-weighted supervised fine-tuning

Recent events (1)

4arXiv · cs.CL·23h ago·source ↗

Gemma 4 31B benchmarked on NRC Reactor Operator Licensing Exam via fine-tuning and RAG configurations

Researchers benchmark eight fine-tuning and retrieval configurations of Gemma 4 31B-IT against the U.S. NRC Reactor Operator licensing examination across 14 Generic Fundamentals Examinations from 2015–2021. The best configuration—supervised fine-tuning with Gemini-distilled chain-of-thought rationales combined with fixed-size chunking RAG—passed 8 of 14 exams and reached 79.7% aggregate accuracy, near the 80% human passing threshold. Key findings include that no configuration without fine-tuning passed any exam, that preferred chunking strategy reverses depending on training state, and that RAFT underperforms standard SFT in matched retrieval environments.

Evaluation and Benchmarking Enterprise Deployment Patterns U.S. Nuclear Regulatory Commission BM25 Google +4 more