Pith. sign in

REVIEW 3 major objections 5 minor 41 references

FedRAG: A Framework for Fine-Tuning Retrieval-Augmented Generation Systems

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read FedRAG claims to be a framework that makes fine-tuning RAG retrievers and generators straightforward in both centralized and federated settings, with seamless conversion between them and deep integration with the HuggingFace-based tooling…

desk verdict A clean, well-integrated RAG fine-tuning framework with released code, but the federated path — the headline feature — is never executed and the centralized validation is two noisy runs on 100 examples. read the letter →

arxiv 2506.09200 v2 pith:GMQNTUQE submitted 2025-06-10 cs.LG cs.CL

Val Andrei Fajardo , David B. Emerson , Amandeep Singh , Veronica Chatrath , Marcelo Lotif , Ravi Theja , Alex Cheung , Izuki Matsuba This is my paper
classification cs.LGcs.CL
keywords retrieval-augmentedgenerationfine-tuningfederatedlearningRAGframeworkRA-DITRALTLSRLLM
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces FedRAG, a software framework whose central claim is that fine-tuning retrieval-augmented generation systems—both the retriever and the generator—can be done through one clean set of abstractions, and that the same centralized training task can be converted into a federated one with minimal code changes. If this works as claimed, practitioners would be able to adapt RAG systems to domain-specific data, including data that cannot be centrally collected, without learning a new training stack. The framework wraps state-of-the-art objectives like RALT and LSR, integrates with popular tools, and exposes a trainer manager that orchestrates which model is updated while the other is frozen. The paper's own validation is a lightweight centralized experiment: QLoRA fine-tuning with the RALT objective improved average exact match on the MMLU global facts subset from 22.0 to 30.5, while the federated path is described but never executed.

What carries the argument

The central object is the RAGSystem, composed of a KnowledgeStore, a Retriever, and a Generator, along with a TrainerManager that orchestrates fine-tuning: it prepares the target model for training, freezes the other model, and exposes a train() method and a get_federated_task() method. The fine-tuning methods implemented are RALT (Retrieval-Augmented Language Model Training), which teaches the generator to use retrieved context, and LSR (Language Model Supervised Retriever Training), which minimizes the KL divergence between retriever scores over retrieved chunks and generator target probabilities. The manager's ability to emit a federated task object is the mechanism that carries the paper's central claim of seamless centralized-to-federated conversion.

What would settle it

Run the framework's federated path with two or more clients on disjoint subsets of Web Questions and compare exact-match accuracy on the MMLU global facts split against the centralized RALT result; if the federated model fails to train or clearly underperforms the 30.5 average, the claim of seamless federated conversion is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that FedRAG provides a single workflow spanning the whole RAG fine-tuning lifecycle: assemble a RAGSystem from a retriever, a generator, and a knowledge store; fine-tune either component with objectives such as RALT (generator training on retrieved context) or LSR (retriever training via distribution matching); and then convert the centralized trainer into a federated task by extracting an FL task object that plugs into a Flower-based server and client. The authors argue this fills a gap because existing RAG frameworks focus on assembly and inference while leaving fine-tuning and federated conversion to ad hoc user effort.

Load-bearing premise

The seamless conversion from centralized to federated fine-tuning works as advertised, but the paper never executes a federated run, so the entire federated benefit rests on that untested assumption.

Editorial extensions

If this is right

  • A user can assemble a RAG system, fine-tune the retriever or generator with RALT or LSR, and then extract an FL task to run the same training federated across clients with minimal code changes.
  • The framework integrates with HuggingFace models and datasets, Unsloth for fast generator fine-tuning, Qdrant for the knowledge store, and LlamaIndex for inference bridging.
  • The lightweight RALT reproduction improved average exact match on the MMLU global facts subset from 22.0 to 30.5, consistent with the direction of earlier RA-DIT results.
  • Because the components are modular, new trainers, losses, benchmarks, and knowledge stores can be added without rewriting the core pipeline.
  • If the federated conversion works, FedRAG would give organizations a practical route to fine-tuning RAG systems on data that must remain decentralized.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper does not show an executed federated experiment, so the strongest practical payoff—training on private or fragmented data—rests on an untested assumption; a two-client run on disjoint data would be the natural next check.
  • The reported MMLU gain comes from 100 test items and two noisy runs, so the 22-to-30.5 improvement is suggestive but not a stable estimate; a multi-seed run on a larger benchmark would settle the effect size.
  • The same trainer-manager pattern could serve as a testbed for comparing generator-side objectives against retriever-side losses within identical RAG configurations, isolating which component actually drives downstream gains.
  • The roadmap's planned ReSearch-style reinforcement-learning trainer would extend the framework from supervised fine-tuning to reasoning-with-search, a larger design space the current centralized experiment does not touch.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript introduces FedRAG, a library framework for fine-tuning retrieval-augmented generation (RAG) systems in both centralized and federated settings. It describes a modular design (core, evals, FL tasks, generators, knowledge stores, retrievers, trainers), shows usage patterns via code snippets, and reports a lightweight RA-DIT-inspired experiment in Appendix A in which a Llama-2-7B generator is fine-tuned with the RALT objective using QLoRA and evaluated on the MMLU global-facts subset. The paper claims that FedRAG offers state-of-the-art fine-tuning methods, seamless conversion from centralized to federated training, and fills a critical gap in RAG tooling.

Significance. If the framework delivers what is claimed, it would be a valuable systems contribution: a single API for retriever/generator fine-tuning, integrated with HuggingFace, Unsloth, Qdrant, and LlamaIndex, plus a straightforward route to federated training. The release of library code, a containerized knowledge store, and experimental code are concrete strengths that aid reproducibility. However, the current evidence is far too thin to support the central claims. The federated training path is never executed; the only experiment is two noisy runs of a single fine-tuning method on 100 examples; and the 'critical gap' claim is not supported by a systematic survey of existing tools. The paper is therefore best viewed as a framework proposal with an initial proof-of-concept, rather than a validated system.

major comments (3)
  1. [Section 4.2, Figure 3] The claim of a 'seamless conversion' from centralized to federated RAG fine-tuning is the paper's distinguishing feature, but the federated path is never exercised. Figure 3 only shows that manager.get_federated_task() returns an object with server and client launched via flwr.start_server and flwr.start_client; no experiment, log, or ablation with this path appears in the paper or appendices. Federated averaging over HuggingFace trainers with QLoRA adapters, frozen retriever/generator, and optimizer state is non-trivial, and a mismatch in initialization, adapter handling, or freezing logic would silently break the central claim. I recommend adding a minimal federated experiment with at least two clients on the same 100-item MMLU subset, reporting accuracy and comparing with the centralized run.
  2. [Appendix A, Table 3] The empirical validation consists of two runs on 100 MMLU global-facts examples, with baseline exact match of 17.0 and 27.0 and RALT exact match of 27.0 and 34.0. The between-run spread (10 and 7 points) is as large as the reported mean improvement (8.5 points). The paper reports no random seeds, hyperparameters (learning rate, epochs, batch size, LoRA rank/alpha), or evaluation settings, despite attributing variability to 'sampling parameters used for generation' (Section A.3). This is insufficient to substantiate Section 5's conclusion that the framework 'can be used to successfully and flexibly execute RAG fine-tuning tasks.'
  3. [Sections 1, 2.2, and 5] The abstract and introduction claim that FedRAG fills 'a critical gap in available tools' and supports 'state-of-the-art fine-tuning methods,' but the evidence does not match these claims. Only RALT generator fine-tuning is demonstrated; LSR, the combined RA-DIT sequence, and retriever fine-tuning are never run, and ReSearch is only listed on the roadmap (Table 4). Additionally, the related work (Section 2.2) is too narrow to establish that no existing framework offers similar functionality; a comparison with general federated LLM fine-tuning platforms (e.g., FedML, NVIDIA FLARE, OpenFL) is missing. Please either temper the claims or provide evidence for them.
minor comments (5)
  1. [Section 4.2, Figure 2] The import statement in Figure 2 is missing commas between the trainer class names, reading 'HuggingFaceTrainerForRALT HuggingFaceTrainerForLSR' instead of a proper Python import list.
  2. [Section A.2] The MMLU evaluation with exact match is ambiguous; please clarify whether the model outputs a multiple-choice letter or a free-form answer, and how exact match is computed.
  3. [Section A.2] The few-shot examples are said to be 'randomly drawn from the validation split and held fixed,' but no seed or the actual examples are given, which hinders reproducibility.
  4. [Section 4.3] The sentence 'Of the integrations listed in Table 2, only the LlamaIndex integration had not been represented in the preceding patterns' is grammatically unclear; it likely means that the LlamaIndex bridge is not shown in Figures 1-4.
  5. [Footnote 1] The reference to the library code appears only in a footnote; consider stating the repository URL and a versioned release in the main text for reproducibility.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: FedRAG's claims are supported by external benchmarks and adopted methods, not by fitting or self-referential definitions; the only self-citation is incidental and not load-bearing.

full rationale

FedRAG contains no derivation chain whose output is defined by its input. The paper presents a software framework; its empirical claim in Appendix A is an external comparison (fine-tuned vs. no-fine-tuning on the MMLU global-facts subset), not a quantity defined inside the paper. The methods wrapped by the framework (RALT, LSR, federated averaging) are adopted from prior external literature, not derived from FedRAG's own results. The sole self-citation, Ref. [11] (llama-index-networks, co-authored by the first author), appears only in Related Work as an example of decentralized RAG inference and does not justify FedRAG's central value proposition; the stated gap ('no existing tools provide a simple interface for converting centralized RAG fine-tuning to federated tasks') is an assertion about the ecosystem, not a conclusion forced by that citation. The federated training path is never executed or validated in the paper, but that is an evidence/validation gap rather than circularity: no fitted parameter is renamed as a prediction, and no equation reduces to its own input. Accordingly, no circular step meets the quotable-reduction bar; the score of 1 reflects the incidental non-load-bearing self-citation and otherwise zero circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

Central claim rests on the software functioning as described and on the assumptions above. No free parameters are fitted; the experiment has no fitted constants. No new physical or conceptual entities are introduced.

assumptions (3)
  • domain assumption RALT and LSR objectives as implemented in FedRAG match the original RA-DIT paper (Lin et al., 2023).
    The framework's trainers rely on the correctness of the published objectives and their loss formulations, which are not re-derived in this paper.
  • domain assumption Federated averaging via the Flower backend preserves the training dynamics of the wrapped HuggingFace trainer.
    The federated conversion feature is never tested in the paper; the API assumes the flwr server/client pattern in Figure 3 will converge and produce a useful global model.
  • domain assumption The DRAGON+ retriever and Llama-2-7B generator are suitable base models for reproducing RA-DIT-style improvements.
    The lightweight experiment uses a 4-bit Llama-2 with QLoRA, which differs from the original RA-DIT setup; the positive result is assumed to transfer.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FedRAG: A Framework for Fine-Tuning Retrieval-Augmented Generation Systems." pith.science (2026). https://pith.science/paper/GMQNTUQE

@misc{pith2026250609200,
  author       = {Pith},
  title        = {Pith review of: FedRAG: A Framework for Fine-Tuning Retrieval-Augmented Generation Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GMQNTUQE}},
  note         = {Machine review of arXiv:2506.09200}
}
read the original abstract

Retrieval-augmented generation (RAG) systems have been shown to be effective in addressing many of the drawbacks of relying solely on the parametric memory of large language models. Recent work has demonstrated that RAG systems can be improved via fine-tuning of their retriever and generator models. In this work, we introduce FedRAG, a framework for fine-tuning RAG systems across centralized and federated architectures. FedRAG supports state-of-the-art fine-tuning methods, offering a simple and intuitive interface and a seamless conversion from centralized to federated training tasks. FedRAG is also deeply integrated with the modern RAG ecosystem, filling a critical gap in available tools.

Figures

Figures reproduced from arXiv: 2506.09200 by the authors.

Figure 1
Figure 1. Creating a RAGSystem with FedRAG. Not depicted here, but the retriever is also used to popu￾late embedded reference chunks into the knowledge store, prior to querying. train () method that can be called without using manager providing an interface similar to that of HuggingFace. Federated Fine-Tuning: With the centralized fine￾tuning pattern established, we show the simple process for converting the previous task to… view at source ↗
Figure 4
Figure 4. Benchmarking with FedRAG. LlamaIndex integration had not been represented in the preceding patterns of Figures 1–4. FedRAG supports a bridge to convert a RAGSystem object to a LlamaIndex RAG system equivalent, thus enabling users to leverage their powerful inference features and ecosystem. These in￾tegrations allow FedRAG users to leverage existing tools while gaining RAG fine-tuning capabilities, aligning with the … view at source ↗
Figure 2
Figure 2. Fine-tuning a RAGSystem with FedRAG. ... # Keep code from Figures 1 and 2 import flwr as fl # The FL backend # fl_task fl_task = manager.get_federated_task() # Build fl server and client server = fl_task.server( model=retriever_trainer.model ) client = fl_task.client( model=retriever_trainer.model, train_dataset=train_dataset, ) # Spin up client and server using flwr fl.start_server(server) fl.start_client(client) … view at source ↗
Figures from the paper (1 more)
Figure 3
Figure 3. Figure 3: Federated fine-tuning of a RAGSystem with FedRAG. 4.3 Integrations In this section, we briefly outline the existing integra￾tions to popular tools and frameworks within the RAG ecosystem. Of the integrations listed in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

41 extracted references · 25 canonical work pages

  1. [1]

    Model context protocol, 2024

    Anthropic. Model context protocol, 2024. URL https://github.com/modelcontextprotocol/ modelcontextprotocol

    work page 2024

  2. [2]

    Semantic parsing on freebase from question-answer pairs

    Jonathan Berant, Andrew Chou, Roy Frostig, and Percy Liang. Semantic parsing on freebase from question-answer pairs. In Proceedings of the 2013 conference on empirical methods in natural language processing, pages 1533–1544, 2013

    work page 2013

  3. [3]

    Enhancing RAG pipelines in Haystack: Introducing diversityranker and lostinthemiddleranker, 2023

    Vladimir Blagojevic. Enhancing RAG pipelines in Haystack: Introducing diversityranker and lostinthemiddleranker, 2023. https: //towardsdatascience.com/enhancing-rag- pipelines-in-haystack-45f14e2bc9f5

    work page 2023

  4. [4]

    Brown, B

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J.D. Kaplan, P. Dhariwal, A. Neelakantan, and et al. Lan- guage models are few-shot learners. In Advances in Neural Information Processing Systems , volume 33, pages 1877–1901, 2020

    work page 1901

  5. [5]

    LangChain, October 2022

    Harrison Chase. LangChain, October 2022. URL https://github.com/langchain-ai/langchain

    work page 2022

  6. [6]

    Pan, Wen Zhang, Huajun Chen, Fan Yang, Zenan Zhou, and Weipeng Chen

    Mingyang Chen, Tianpeng Li, Haoze Sun, Yijie Zhou, Chenzheng Zhu, Haofen Wang, Jeff Z. Pan, Wen Zhang, Huajun Chen, Fan Yang, Zenan Zhou, and Weipeng Chen. Research: Learning to reason with search for llms via reinforcement learning, 2025. URL https://arxiv.org/abs/2503.19470

    arXiv 2025

  7. [7]

    Suchanek, and Chlo´ e Clavel

    Cyril Chhun, Pierre Colombo, Fabian M. Suchanek, and Chlo´ e Clavel. Of human criteria and automatic metrics: A benchmark of the evaluation of story gen- eration. In Nicoletta Calzolari, Chu-Ren Huang, Hansaem Kim, James Pustejovsky, Leo Wanner, Key-Sun Choi, Pum-Mo Ryu, Hsin-Hsi Chen, Lu- cia Donatelli, Heng Ji, Sadao Kurohashi, Patrizia Paggio, Nianwen ...

    work page 2022

  8. [8]

    Deepseek-r1: Incentivizing rea- soning capability in llms via reinforcement learning,

    DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, and et al. Deepseek-r1: Incentivizing rea- soning capability in llms via reinforcement learning,

    work page

Show all 41 references
  1. [9]

    Qlora: Efficient finetuning of quantized llms

    Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. Qlora: Efficient finetuning of quantized llms. Advances in neural information processing systems, 36:10088–10115, 2023

    2023

  2. [10]

    Omkar Dige, John Willes, and D. B. Emerson. Evaluating RAG system performance: The impact of knowledge cut-off and fine-tuning. In Adap- tive Foundation Models: Evolving AI for Person- alized and Efficient Learning , 2024. URL https: //openreview.net/forum?id=2K6I3317QV

    2024

  3. [11]

    LlamaIndex Networks, 11 2024

    Val Andrei Fajardo, Jerry Liu, Logan Markewich, Simon Suo, Haotian Zhang, and Sourabh Desai. LlamaIndex Networks, 11 2024. URL https://github.com/run- llama/llama_index/llama-index-networks. Extension for LlamaIndex

    2024

  4. [12]

    A survey on rag meeting llms: To- wards retrieval-augmented large language models

    Wenqi Fan, Yujuan Ding, Liangbo Ning, Shijie Wang, Hengyun Li, Dawei Yin, Tat-Seng Chua, and Qing Li. A survey on rag meeting llms: To- wards retrieval-augmented large language models. In Proceedings of the 30th ACM SIGKDD Con- ference on Knowledge Discovery and Data Mining , ...

  5. [13]

    Agent2agent (a2a) protocol, 2025

    Google. Agent2agent (a2a) protocol, 2025. URL https://github.com/google/A2A

    2025

  6. [14]

    The llama 3 herd of models, 2024

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, and et al. The llama 3 herd of models, 2024. URL https://arxiv.org/ abs/2407.21783

    2024 arXiv

  7. [15]

    Measuring massive multitask language under- standing

    Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Stein- hardt. Measuring massive multitask language under- standing. arXiv preprint arXiv:2009.03300 , 2020

    2009 arXiv

  8. [16]

    A survey on hallucination in large lan- guage models: Principles, taxonomy, challenges, and open questions

    Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and Ting Liu. A survey on hallucination in large lan- guage models: Principles, taxonomy, challenges, and open questions. ACM Trans. Inf. Syst....

    2025 doi

  9. [17]

    Few-shot learning with retrieval augmented language models

    Gautier Izacard, Patrick Lewis, Maria Lomeli, Lucas Hosseini, Fabio Petroni, Timo Schick, Jane Dwivedi- Yu, Armand Joulin, Sebastian Riedel, and Edouard 5 Grave. Few-shot learning with retrieval augmented language models. arXiv preprint arXiv:2208.03299 , 1(2):4, 2022

    2022 arXiv

  10. [18]

    Dense passage retrieval for open-domain question answering

    Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. Dense passage retrieval for open-domain question answering. In Bonnie Webber, Trevor Cohn, Yulan He, and Yang Liu, edi- tors, Proceedings of the 2020 Conference on...

    2020 doi

  11. [19]

    Komal Kumar, Tajamul Ashraf, Omkar Thawakar, Rao Muhammad Anwer, Hisham Cholakkal, Mubarak Shah, Ming-Hsuan Yang, Phillip H. S. Torr, Fahad Shahbaz Khan, and Salman Khan. Llm post- training: A deep dive into reasoning large language models, 2025. URL https://arxiv.org/abs/2502. 21321

    2025

  12. [20]

    Retrieval-augmented generation for knowledge-intensive NLP tasks

    Patrick Lewis, Ethan Perez, Aleksandra Pik- tus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich K¨ uttler, Mike Lewis, Wen-tau Yih, Tim Rockt¨ aschel, Sebastian Riedel, and Douwe Kiela. Retrieval-augmented generation for knowledge-intensive NLP tasks. In Proceedings ...

    2020

  13. [21]

    How to train your dragon: Diverse augmentation towards generalizable dense retrieval

    Sheng-Chieh Lin, Akari Asai, Minghan Li, Barlas Oguz, Jimmy Lin, Yashar Mehdad, Wen-tau Yih, and Xilun Chen. How to train your dragon: Diverse augmentation towards generalizable dense retrieval. arXiv preprint arXiv:2302.07452 , 2023

    2023 arXiv

  14. [22]

    Ra- dit: Retrieval-augmented dual instruction tuning

    Xi Victoria Lin, Xilun Chen, Mingda Chen, Weijia Shi, Maria Lomeli, Richard James, Pedro Rodriguez, Jacob Kahn, Gergely Szilvasy, Mike Lewis, et al. Ra- dit: Retrieval-augmented dual instruction tuning. In The Twelfth International Conference on Learning Representations, 2023

    2023

  15. [23]

    LlamaIndex, November 2022

    Jerry Liu. LlamaIndex, November 2022. URL https: //github.com/jerryjliu/llama_index

    2022

  16. [24]

    Query rewriting in retrieval- augmented large language models

    Xinbei Ma, Yeyun Gong, Pengcheng He, Hai Zhao, and Nan Duan. Query rewriting in retrieval- augmented large language models. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Pro- ceedings of the 2023 Conference on Empirical Meth- ods in Natural Language Processing , pages...

    2023 doi

  17. [25]

    H. B. McMahan, E. Moore, D. Ramage, S. Hamp- son, and B. A. y Arcas. Communication-efficient learning of deep networks from decentralized data. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS) , 2017

    2017

  18. [26]

    Fine-tuning or retrieval? com- paring knowledge injection in llms, 2024

    Oded Ovadia, Menachem Brief, Moshik Mishaeli, and Oren Elisha. Fine-tuning or retrieval? com- paring knowledge injection in llms, 2024. URL https://arxiv.org/abs/2312.05934

    2024 arXiv

  19. [27]

    REFINER: Reasoning feedback on in- termediate representations

    Debjit Paul, Mete Ismayilzada, Maxime Peyrard, Beatriz Borges, Antoine Bosselut, Robert West, and Boi Faltings. REFINER: Reasoning feedback on in- termediate representations. In Yvette Graham and Matthew Purver, editors, Proceedings of the 18th Conference of the European Chapt...

    2024

  20. [28]

    Graph retrieval-augmented generation: A survey, 2024

    Boci Peng, Yun Zhu, Yongchao Liu, Xiaohe Bo, Haizhou Shi, Chuntao Hong, Yan Zhang, and Siliang Tang. Graph retrieval-augmented generation: A survey, 2024. URL https://arxiv.org/abs/2408. 08921

    2024

  21. [29]

    In-context retrieval-augmented language models

    Ori Ram, Yoav Levine, Itay Dalmedigos, Dor Muhlgay, Amnon Shashua, Kevin Leyton-Brown, and Yoav Shoham. In-context retrieval-augmented language models. Transactions of the Associa- tion for Computational Linguistics , 11:1316–1331,

  22. [30]

    Liu, and Bal- aji Lakshminarayanan

    Jie Ren, Yao Zhao, Tu Vu, Peter J. Liu, and Bal- aji Lakshminarayanan. Self-evaluation improves selective generation in large language models. In Javier Antor´ an, Arno Blaas, Kelly Buchanan, Fan Feng, Vincent Fortuin, Sahra Ghalebikesabi, An- dreas Kriegler, Ian Mason, David ...

    2023

  23. [31]

    Lorenzo Sani, Alex Iacob, Zeyu Cao, Royson Lee, Bill Marino, Yan Gao, Dongqi Cai, Zexi Li, Wanru Zhao, Xinchi Qiu, and Nicholas D. Lane. Photon: Federated llm pre-training, 2024. URL https:// arxiv.org/abs/2411.02908. 6

    2024

  24. [32]

    Llama 2: Open founda- tion and fine-tuned chat models

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Niko- lay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. Llama 2: Open founda- tion and fine-tuned chat models. arXiv preprint arXiv:2307.09288, 2023

    2023 arXiv

  25. [33]

    Le, Ed H

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc V. Le, Ed H. Chi, Sharan Narang, Aakanksha Chowd- hery, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models. In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigal...

    2023

  26. [34]

    Rossi, Alexa Siu, Ruiyi Zhang, and Tyler Derr

    Yu Wang, Nedim Lipka, Ryan A. Rossi, Alexa Siu, Ruiyi Zhang, and Tyler Derr. Knowledge graph prompting for multi-document question answering,

  27. [35]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. Chain-of-thought prompting elicits reasoning in large language models. In Proceedings of the 36th International Conference on Neural Information Processing Sys...

    2022

  28. [36]

    Raft: Adapting language model to domain specific rag

    Tianjun Zhang, Shishir G Patil, Naman Jain, Sheng Shen, Matei Zaharia, Ion Stoica, and Joseph E Gon- zalez. Raft: Adapting language model to domain specific rag. In First Conference on Language Mod- eling, 2024

    2024

  29. [37]

    Open-source large language mod- els are strong zero-shot query likelihood models for document ranking

    Shengyao Zhuang, Bing Liu, Bevan Koopman, and Guido Zuccon. Open-source large language mod- els are strong zero-shot query likelihood models for document ranking. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the As- sociation for Computational Linguistics...

    2023 doi

  30. [38]

    URL https://arxiv.org/abs/2308.11730

    arXiv

  31. [2023]

    URL https: //aclanthology.org/2023.tacl-1.75

    doi: 10.1162/tacl a 00605. URL https: //aclanthology.org/2023.tacl-1.75

    2023 doi

  32. [2024]

    ISBN 9798400704901

    Association for Computing Machinery. ISBN 9798400704901. doi: 10.1145/3637528.3671470. URL https://doi.org/10.1145/3637528.3671470

  33. [2025]

    URL https://arxiv.org/abs/2501.12948

    arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.