LLM MOE models for RAG search

References

Sources cited by the report.

Appendix D. References

[1] P. Lewis, E. Perez, A. Piktus, et al., "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks," Advances in Neural Information Processing Systems, vol. 33, 2020.

[2] V. Karpukhin, B. Oguz, S. Min, et al., "Dense Passage Retrieval for Open-Domain Question Answering," Proceedings of EMNLP, 2020.

[3] O. Khattab and M. Zaharia, "ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT," Proceedings of SIGIR, 2020.

[4] Google DeepMind, "Gemma 4 Model Card," Google AI for Developers, 2026. https://ai.google.dev/gemma/docs/core/model_card_4

[5] Google DeepMind, "Gemma 4 Model Overview and Quantization-Aware Training," Google AI for Developers, 2026. https://ai.google.dev/gemma/docs/core

[6] Google DeepMind, "Thinking Mode in Gemma" and "Gemma 4 Prompt Formatting," Google AI for Developers, 2026. https://ai.google.dev/gemma/docs/capabilities/thinking

[7] ggml-org, "llama.cpp Server README," 2026. https://github.com/ggml-org/llama.cpp/blob/master/tools/server/README.md

[8] Qwen Team, "Qwen3: Think Deeper, Act Faster" and Qwen3-30B-A3B Model Card, 2025. https://qwenlm.github.io/blog/qwen3/

[9] Mistral AI, "Mistral-Small-3.2-24B-Instruct-2506 Model Card," 2025. https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506

[10] Mistral AI, "Mistral-Small-3.1-24B Model Card," 2025. https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Base-2503

[11] Meta, "Llama 4 Scout Model Card," 2025. https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct

[12] Mistral AI, "Ministral-3-14B-Instruct-2512 Model Card," 2025. https://huggingface.co/mistralai/Ministral-3-14B-Instruct-2512

[13] Microsoft, "Phi-4-mini-instruct Model Card," 2025. https://huggingface.co/microsoft/Phi-4-mini-instruct

[14] Attached artifact, prompt2(5), "Master Prompt: Academic Evaluation of Local-LLM CV Extraction," supplied July 2026.

[15] Attached artifacts, aiextract.sys.prompt and aiextract.user.prompt, supplied July 2026.

[16] Attached source set, 24 category-specific local-LLM extraction evaluation reports covering 2,484 PDF CVs, associated prompts, and the combined Excel report, July 2026.