Pith. sign in

REVIEW 4 major objections 8 minor 1 cited by

AviationLLM: An LLM-based Knowledge System for Aviation Training

T0 review · 4 major / 8 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 RALA-DPO combines Direct Preference Optimization with retrieval-augmented generation to make a Qwen-based LLM deliver aviation-training answers that experts score highest on accuracy and timeliness.

desk verdict Standard DPO+RAG recipe applied to aviation training with a useful new dataset, but the evaluation is too circular to support the headline claims. read the letter →

arxiv 2506.14336 v1 pith:VX5X7SY3 submitted 2025-06-17 cs.AI

Jia'ang Wan , Feng Shen , Fujuan Li , Yanjin Sun , Yan Li , Shiwen Zhang This is my paper
classification cs.AI
keywords largelanguagemodelsDirectPreferenceOptimizationRetrieval-AugmentedGenerationaviationtrainingQwendomainadaptationquestionansweringknowledgebase
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

This paper proposes RALA-DPO, a pipeline for building a large language model that answers professional aviation theory questions accurately and with current knowledge. The authors fine-tune the open-source Qwen2.5-14B with Direct Preference Optimization (DPO) on a preference dataset of 9,740 expert-annotated aviation answers, rather than relying on supervised fine-tuning (SFT) alone. They then add Retrieval-Augmented Generation (RAG), which embeds aviation manuals and regulations into a vector database and feeds the most relevant fragments into the prompt before the model answers. In their experiments, DPO beats SFT with a 57 percent win rate in pairwise comparison, and the DPO+RAG configuration receives the highest expert scores, 4.25 for fluency, 4.83 for accuracy, and 4.63 for timeliness out of 5. The intended consequence is that one LLM system can lessen dependence on limited instructor time while keeping training answers grounded in the latest authoritative documents.

What carries the argument

The load-bearing object is the DPO objective together with the retrieval-scoring rule. DPO replaces explicit reward modeling by optimizing $\log\sigma\left(\beta\log\frac{\pi(y_w\mid x)}{\pi_{\mathrm{ref}}(y_w\mid x)}-\beta\log\frac{\pi(y_l\mid x)}{\pi_{\mathrm{ref}}(y_l\mid x)}\right)$ over preference pairs, pushing the model toward expert-preferred continuations while a temperature $\beta$ controls how far it may drift from the reference model. RAG supplies timeliness: a text-embedding model indexes aviation knowledge fragments, a query is embedded and compared by cosine similarity, and the top $k$ fragments are inserted into a prompt template with the original question before the generator produces the answer. These two mechanisms do complementary work, one tuning generation preferences and the other grounding each answer in current retrievable evidence.

What would settle it

Re-run the same four model configurations on a question set assembled from aviation manuals and regulations that were not used to create the preference training data, with expert scorers blind to configuration; if DPO+RAG no longer clearly leads SFT+RAG on accuracy and timeliness, the claim that the combined framework is the source of improvement fails.

Watch

Extended reading notes

Core claim

The central claim is that alignment and retrieval address different failure modes, and their combination, RALA-DPO, outperforms either technique alone for domain-specific question answering. Using the implicit reward $R(x,y)=\beta\log \frac{\pi(y\mid x)}{\pi_{\mathrm{ref}}(y\mid x)}+\beta\log Z(x)$ and the DPO loss $L_{\mathrm{DPO}}=-\mathbb{E}_{(x_i,y_w^{(i)},y_l^{(i)})\sim D}\log\sigma\left(\beta\log\frac{\pi(y_w^{(i)}\mid x_i)}{\pi_{\mathrm{ref}}(y_w^{(i)}\mid x_i)}-\beta\log\frac{\pi(y_l^{(i)}\mid x_i)}{\pi_{\mathrm{ref}}(y_l^{(i)}\mid x_i)}\right)$, the model is trained to prefer expert-written answers over coarse responses while remaining close to the reference policy. The retrieval layer ranks knowledge fragments by cosine similarity $s(q,d)=\frac{\mathrm{Emb}_R(q)\cdot \mathrm{Emb}_R(d)}{\lVert \mathrm{Emb}_R(q)\rVert\, \lVert \mathrm{Emb}_R(d)\rVert}$ and concatenates the top $k$ fragments into the prompt before generation. The paper reports that this combined pipeline reaches the best expert ratings in Table 2, and that DPO alone wins 57 percent of pairwise comparisons against SFT, supporting the conclusion that preference alignment plus retrieval provides both accuracy and timeliness.

Load-bearing premise

The evaluation samples questions from the same aviation training dataset used to build the preference data, and no held-out partition is described, so the reported improvements assume the test questions genuinely measure the model's ability to answer new aviation questions rather than its familiarity with the training set.

Editorial extensions

If this is right

  • If RALA-DPO is correct, a DPO-aligned generator with a RAG knowledge base receives higher expert ratings for fluency, accuracy, and timeliness than SFT-based configurations with or without retrieval.
  • Aviation theory training can be supported by an LLM that answers routine professional questions immediately, reducing the bottleneck created by a limited number of instructors.
  • Updating the informational content of the system becomes a matter of adding or replacing documents in the knowledge base rather than retraining or fine-tuning the model.
  • Because RAG attaches retrieved source fragments to the prompt, answers can be traced back to authoritative manuals and regulations, making generated guidance auditable.

Reading between the lines

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

  • Extending beyond the paper: the zero-cost update claim is testable by inserting a newly issued regulation into the knowledge base and measuring whether expert-scored answers on that regulation improve without any model retraining, which the paper does not report.
  • Extending beyond the paper: a held-out evaluation split built from aviation documents never used to create the preference dataset would separate genuine generalization from memorization of the training distribution, and would be the natural stress test of the 57 percent win rate.
  • Extending beyond the paper: the same preference-alignment-plus-retrieval recipe could be applied to other regulated professions with authoritative document corpora, such as air traffic control, aircraft maintenance, and medical training, where timeliness and traceability matter as much as raw accuracy.
  • Extending beyond the paper: comparing RALA-DPO across base-model scales, for example a 7B parameter model against the 14B model used here, would show how much of the reported gain comes from the alignment and retrieval mechanism and how much from model size.
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

4 major / 8 minor

Summary. The paper proposes RALA-DPO, an LLM-based question-answering system for aviation theory training. The authors fine-tune Qwen2.5-14B with Direct Preference Optimization on a custom aviation training dataset (ATDS) that contains 9,740 preferred-response pairs, and they augment the fine-tuned model with Retrieval-Augmented Generation over an aviation knowledge base. They report that DPO outperforms SFT in an AI-evaluator pairwise comparison (Table 1, 57% win rate) and that DPO+RAG receives the highest expert scores for fluency, accuracy, and timeliness (Table 2). The paper concludes that the combination of DPO and RAG improves answer accuracy and timeliness for aviation training, with RAG providing zero-cost knowledge updates.

Significance. If the empirical claims were sound, the paper would offer a useful applied contribution: an integrated pipeline combining DPO fine-tuning and RAG for a high-stakes domain, with a clearly described training and retrieval architecture. The authors deserve credit for constructing a domain dataset and for comparing several model variants. However, the experimental evidence as presented does not establish the central generalization claim. The evaluation set is sampled from the same ATDS used for training, with no held-out partition or deduplication described; the proprietary AI evaluator is not validated against human judgment; and the expert evaluation is based on only 100 samples per condition without significance testing or inter-annotator agreement. The methodological novelty is limited: the DPO loss equations (Sections 4.1) are a standard restatement from Rafailov et al., and the RAG component is a standard cosine-similarity retrieval pipeline. No code or data are released, which further limits reproducibility. The paper is better viewed as an engineering report than as a validated scientific study in its current form.

major comments (4)
  1. [Section 5.2] The evaluation dataset is 'derived from random sampling of ATDS' (Section 5.2), while Section 5.1 states that the 9,740 preferred response pairs in ATDS were used to train the SFT and DPO models. The manuscript does not describe any train/eval split, deduplication, or topic-stratified holdout. Consequently, the reported improvements in Tables 1 and 2 may reflect memorization of the training distribution rather than generalization to new aviation questions. This is the load-bearing evidence for the paper's central claim, so the evaluation design must be corrected with a genuinely held-out set before the claim can be assessed.
  2. [Section 5.2, Table 2] The expert evaluation in Table 2 is based on only 100 sample question-answer pairs per model configuration, with no confidence intervals, inter-annotator agreement, or significance tests reported. Differences such as 4.43 vs. 4.38 for accuracy between DPO and SFT are within plausible sampling noise, so the claimed superiority of DPO over SFT, and of RAG over non-RAG, is not statistically established.
  3. [Section 5.2, Table 1] The AI evaluator Themis-turbo is a proprietary model, and the paper provides no validation of this evaluator against human expert judgments or against an independent benchmark. The paper states that six dimensions are scored, but Table 1 reports only aggregate win/lose/tie counts, with no dimension-level scores or agreement metrics. Without evidence that Themis-turbo's judgments align with expert preference, Table 1 cannot be interpreted as a reliable measure of answer quality.
  4. [Sections 4.2 and 5.3] The timeliness claim is not empirically tested. The paper argues that RAG provides 'zero-cost knowledge updates' and more timely answers, but the expert timeliness scores in Table 2 are subjective ratings and do not demonstrate that the system actually uses newer information than the training data. No temporal holdout, no test with documents newer than the training cutoff, and no evaluation of retrieval provenance is reported. The timeliness advantage therefore remains unsupported.
minor comments (8)
  1. [Abstract] The abstract contains a typo: 'the the instructors' should be 'the instructors'.
  2. [Section 1] The sentence 'We constructs a DPO dataset in the aviation training domain' mixes subject and verb agreement; it should be 'We construct'.
  3. [Tables 1 and 2] The table captions are formatted as 'T able 1' and 'T able 2'; the spacing should be corrected.
  4. [Section 4.1] There is a missing period after 'preserved in π ref' before 'The constrained update mechanism...'.
  5. [Section 4.1, Equations (1)-(7)] The DPO loss derivation is a standard restatement from Rafailov et al. (reference [16]); the authors should clarify what, if anything, is new in their formulation and cite the original derivation more explicitly.
  6. [Figure 1] The figure caption is very long and reads more like a body-text explanation; consider moving the retrieval description into Section 4.2 and keeping the caption concise.
  7. [References] References [12] and [22] are both Qwen technical reports; please ensure the in-text citations distinguish the OMNI report from the Qwen2.5 technical report clearly.
  8. [Section 5.2] The paper does not report the number of evaluation queries sampled from ATDS, the random seed, or the sampling procedure; provide these details to enable reproducibility.

Circularity Check

1 steps flagged · score 6.0 of 10

No held-out split is described: the evaluation set is sampled from the same ATDS used for training, so reported DPO/RAG gains may reflect memorization rather than generalization.

  1. fitted input called prediction [Section 5.2, Experimental Setup (basis for Table 1 and Table 2)]
    "Both methods used a learning rate of 0.0003 and a batch size of 16, training them on the ATDS dataset. ... For model evaluation, this paper utilizes the AI evaluator Themis-turbo from Alibaba Cloud's Bailian platform to assess two models, with the evaluation dataset derived from random sampling of ATDS."

    The paper trains both SFT and DPO on ATDS and then samples the evaluation set from ATDS. Since DPO is optimized to prefer the expert-annotated responses on ATDS preference pairs, evaluating on ATDS-derived queries measures memorization of the training distribution, not generalization to unseen aviation questions. No held-out partition, deduplication, or temporal split is described, so the central empirical claims that 'DPO significantly improves professional adaptability' and that RAG improves timeliness are not supported by independent out-of-sample evidence.

full rationale

The methodological derivation itself is mostly self-contained: Eqs. (1)-(6) are a standard restatement of DPO from Rafailov et al. [16], and Eqs. (8)-(9) define RAG retrieval and generation; neither is circular. No load-bearing self-citation or uniqueness-imported-from-authors pattern appears. The circular link is in the evaluation chain: the paper's headline empirical support for RALA-DPO comes from Table 1 and Table 2, but the evaluation data are explicitly 'derived from random sampling of ATDS,' the same dataset on which the models were trained. Because DPO training directly increases the probability of preferred responses for ATDS items, evaluating on ATDS-sampled questions makes the measured win rates and expert scores partly a reflection of training-set memorization. The paper reports no held-out set, no deduplication, and no temporal split for the timeliness claim, so the central predictions are not validated against an independent distribution. This warrants a partial-circularity score of 6 rather than a higher score, since the proposed architecture is not definitionally equivalent to its inputs and the flaw is in the evaluation design rather than the method formulation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on domain assumptions about data quality, evaluator validity, and retrieval sufficiency, none of which are independently verified. There are no fitted scientific constants and no invented physical or conceptual entities.

free parameters (3)
  • DPO temperature beta = not reported
    Appears in Eq. 2 and Eq. 6 as the deviation scale from the reference model; the paper never reports its value, and all DPO results depend on this hand-set hyperparameter.
  • learning rate = 0.0003
    Reported in Section 5.2 but no sweep or sensitivity analysis is provided.
  • batch size = 16
    Reported in Section 5.2 but no sweep or sensitivity analysis is provided.
assumptions (4)
  • domain assumption The 9,740 ATDS pairs reflect true aviation expertise and preference quality.
    Section 5.1 describes expert-guided manual annotation plus coarse responses from untrained LLMs, but no inter-annotator agreement or validation of preference correctness is reported.
  • domain assumption Themis-turbo, the proprietary AI evaluator, reliably measures answer quality.
    Section 5.2 treats the automated evaluator as ground truth for win rates without validating it against human judgments or independent benchmarks.
  • domain assumption The human expert scores are unbiased and representative.
    Section 5.2 says aviation experts scored 100 sample pairs per model, but the source of experts, blinding, and variance are not described.
  • domain assumption The retrieved top-k passages contain sufficient and correct knowledge for each query.
    Section 4.2 and Eq. 8 assume retrieval quality; no retrieval recall, precision, or failure analysis is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AviationLLM: An LLM-based Knowledge System for Aviation Training." pith.science (2026). https://pith.science/paper/VX5X7SY3

@misc{pith2026250614336,
  author       = {Pith},
  title        = {Pith review of: AviationLLM: An LLM-based Knowledge System for Aviation Training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VX5X7SY3}},
  note         = {Machine review of arXiv:2506.14336}
}
read the original abstract

Aviation training is a core link in ensuring flight safety, improving industry efficiency and promoting sustainable development. It not only involves flight simulation but also requires the learning of a great deal of professional aviation theory knowledge. In the existing training system, the knowledge is mainly imparted by the the instructors. However, the number of instructors is limited and the professional answers obtained from the Internet are not accurate enough, resulting in low training efficiency. To address this, we introduced LLM, but the basic pre-trained model cannot provide accurate answers to professional fields, so we fine-tuned it. Traditional Supervised Fine-Tuning (SFT) risk generating superficially plausible but factually incorrect responses due to insufficient data coverage. To address this, we employ Direct Preference Optimization(DPO). This paper proposes Retrieval-Augmented LLM Alignment via Direct Preference Optimization(RALA-DPO). We select open source pre-trained LLM Qwen and adapt it to aviation theory training through DPO-based domain alignment. Simultaneously, to mitigate hallucinations caused by training data biases, knowledge obsolescence, or domain knowledge gaps, we implement Retrieval-Augmented Generation(RAG) technology that combines generative and retrieval models. RALA-DPO effectively retrieves relevant information from external knowledge bases and delivers precise and high-quality responses through the generative model. Experimental results demonstrate that RALA-DPO can improve accuracy in response to professional aviation knowledge. With integrated RAG mechanisms, this system can further improve the accuracy of answers and achieve zero-cost knowledge updates simultaneously.

Figures

Figures reproduced from arXiv: 2506.14336 by the authors.

Figure 1
Figure 1. When a user submits a query, the system first performs semantic embedding on the query to generate a corresponding semantic embedding vector. Subsequently, cosine similarity is used to calculate the vector’s similarity to vectors in the vector database. This identifies the context most relevant to the query. These contexts are then substituted into a predefined prompt template. Finally, the augmented prompt, along w… view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. CAMB: A comprehensive industrial LLM benchmark on civil aviation maintenance

    cs.CL 2025-08 conditional novelty 5.0 of 10

    CAMB provides a seven-task, eight-dataset benchmark for assessing LLM and embedding model performance in civil aviation maintenance, with initial results showing large models top out near 69% on domain multiple-choice...

Reference graph

Works this paper leans on

23 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [1]

    Chemcrow: Augmenting large-language models with chemistry tools.Nature Machine Intelligence, 2023

    Andres M Bran, Sam Cox, Oliver Schilter, Carlo Baldassari, Andrew D White, and Philippe Schwaller. Chemcrow: Augmenting large-language models with chemistry tools.Nature Machine Intelligence, 2023

    work page 2023

  2. [2]

    SNFinLLM: Systematic and Nuanced Financial Domain Adaptation of Chinese Large Language Models

    Shujuan Zhao, Lingfeng Qiao, Kangyang Luo, Qian-Wen Zhang, Junru Lu, and Di Yin. Snfinllm: Systematic and nuanced financial domain adaptation of chinese large language models.arXiv preprint arXiv:2408.02302, 2024

    work page Pith review arXiv 2024

  3. [3]

    Biot5+: Towards generalized biological under- standing with iupac integration and multi-task tuning.Proceedingss of the ACL, 2024

    Qizhi Pei, Lijun Wu, Kaiyuan Gao, Xiaozhuan Liang, Yin Fang, Jinhua Zhu, Sh- ufang Xie, Tao Qin, and Rui Yan. Biot5+: Towards generalized biological under- standing with iupac integration and multi-task tuning.Proceedingss of the ACL, 2024

    work page 2024

  4. [4]

    Pubmed 200k rct: A dataset for sequential sentence classification in medical abstracts.Proceedings of the IJCNLP, 2017

    Franck Dernoncourt and Ji Young Lee. Pubmed 200k rct: A dataset for sequential sentence classification in medical abstracts.Proceedings of the IJCNLP, 2017

    work page 2017

  5. [5]

    Enhancing aviation english competency: A simulation-based approach for aspiring pilots.English for Specific Purposes, 76:106–121, 2024

    Derin Atay. Enhancing aviation english competency: A simulation-based approach for aspiring pilots.English for Specific Purposes, 76:106–121, 2024

    work page 2024

  6. [6]

    Design and development of aviation aircraft maintenance training plat- form based on vr technology.Procedia Computer Science, 228:898–906, 2023

    Sihai Li. Design and development of aviation aircraft maintenance training plat- form based on vr technology.Procedia Computer Science, 228:898–906, 2023. 12 Wan et al

    work page 2023

  7. [7]

    Instructfingpt: Financial sen- timent analysis by instruction tuning of general-purpose large language models

    Boyu Zhang, Hongyang Yang, and Xiao-Yang Liu. Instructfingpt: Financial sen- timent analysis by instruction tuning of general-purpose large language models. Proceedings of the IJCAI, 2023

    work page 2023

  8. [8]

    Cyberq: Generating questions and answers for cybersecurity education using knowledge graph-augmented llms.Proceedings of the AAAI, 2024

    Garima Agrawal, Kuntal Pal, Yuli Deng, Huan Liu, and Ying-Chih Chen. Cyberq: Generating questions and answers for cybersecurity education using knowledge graph-augmented llms.Proceedings of the AAAI, 2024

    work page 2024

Show all 23 references
  1. [9]

    Aiseckg: Knowledge graph dataset for cybersecurity education

    Garima Agrawal. Aiseckg: Knowledge graph dataset for cybersecurity education. Proceedings of the AAAI, 2023

    2023

  2. [10]

    Socraticlm: Exploring socratic personalized teaching with large language models.Proceedings of the NeurIPS, 2024

    Jiayu Liu, Zhenya Huang, Tong Xiao, Jing Sha, Jinze Wu, Qi Liu, Shijin Wang, and Enhong Chen. Socraticlm: Exploring socratic personalized teaching with large language models.Proceedings of the NeurIPS, 2024

    2024

  3. [11]

    Knowgpt: Knowledge graph based prompting for large language models

    Qinggang Zhang, Junnan Dong, Hao Chen, Daochen Zha, Zailiang Yu, and Xiao Huang. Knowgpt: Knowledge graph based prompting for large language models. Proceedings of the NeurIPS, 2024

    2024

  4. [12]

    Qwen2.5-omni technical report.arXiv preprint arXiv:2503.20215, 2025

    Jin Xu, Zhifang Guo, Jinzheng He, Hangrui Hu, Ting He, Shuai Bai, Keqin Chen, Jialin Wang, Yang Fan, Kai Dang, Bin Zhang, Xiong Wang, Yunfei Chu, and Junyang Lin. Qwen2.5-omni technical report.arXiv preprint arXiv:2503.20215, 2025

    2025 arXiv

  5. [13]

    Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schul- man, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, a...

    2022

  6. [14]

    Griffiths, Yuan Cao, and Karthik Narasimhan

    Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Thomas L. Griffiths, Yuan Cao, and Karthik Narasimhan. Tree of thoughts: Deliberate problem solving with large language models.Proceedings of the NeurIPS, 2024

    2024

  7. [15]

    Christiano, Miljan Martic, Jan Leike, Shane Legg, Tom B

    Paul F. Christiano, Miljan Martic, Jan Leike, Shane Legg, Tom B. Brown, and Dario Amodei. Deep reinforcement learning from human preferences.Advances in Neural Information Processing Systems, 30, 2017

    2017

  8. [16]

    Manning, and Chelsea Finn

    Rafael Rafailov, Archit Sharma, Eric Mitchell, Stefano Ermon, Christopher D. Manning, and Chelsea Finn. Direct preference optimization: Your language model is secretly a reward model.Advances in Neural Information Processing Systems, 36, 2024

    2024

  9. [17]

    Precise zero-shot dense retrieval without relevance labels.Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics, 1:1762–1777, 2023

    Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. Precise zero-shot dense retrieval without relevance labels.Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics, 1:1762–1777, 2023

    2023

  10. [18]

    Atlas: Few-shot learning with retrieval augmented language models.Journal of Machine Learning Research, 24:1–43, 2023

    Gautier Izacard, Patrick Lewis, Maria Lomeli, Lucas Hosseini, Fabio Petroni, Timo Schick, Jane Dwivedi-Yu, Armand Joulin, Sebastian Riedel, and Edouard Grave. Atlas: Few-shot learning with retrieval augmented language models.Journal of Machine Learning Research, 24:1–43, 2023

    2023

  11. [19]

    Self- rag: Learning to retrieve, generate, and critique through self-reflection.arXiv preprint arXiv:2310.11511, 2023

    Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi. Self- rag: Learning to retrieve, generate, and critique through self-reflection.arXiv preprint arXiv:2310.11511, 2023

    2023 arXiv

  12. [20]

    Ra-dit: Retrieval-augmented dual instruction tuning.Proceedings of the Twelfth International Conference on Learning Representations, 2023

    Xi Victoria Lin, Xilun Chen, Mingda Chen, Weijia Shi, Maria Lomeli, Rich James, Pedro Rodriguez, Jacob Kahn, Gergely Szilvasy, Mike Lewis, Luke Zettlemoyer, and Scott Yih. Ra-dit: Retrieval-augmented dual instruction tuning.Proceedings of the Twelfth International Conference o...

    2023

  13. [21]

    Retrieval-augmented generation-aided causal identification of aviation accidents: A large language model methodology.Expert Systems with Applications, 278:127306, 2025

    Tengfei Ren, Zhipeng Zhang, Bo Jia, and Shiwen Zhang. Retrieval-augmented generation-aided causal identification of aviation accidents: A large language model methodology.Expert Systems with Applications, 278:127306, 2025. AviationLLM: An LLM-based Knowledge System for Aviatio...

    2025

  14. [22]

    Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, ...

    2024 arXiv

  15. [23]

    Opseval: A compre- hensive task-oriented aiops benchmark for large language models.arXiv preprint arXiv:2310.07637, 2023

    Yuhe Liu, Changhua Pei, Longlong Xu, Bohan Chen, Mingze Sun, Zhirui Zhang, Yongqian Sun, Shenglin Zhang, Kun Wang, Haiming Zhang, Jianhui Li, Gaogang Xie, Xidao Wen, Xiaohui Nie, Minghua Ma, and Dan Pei. Opseval: A compre- hensive task-oriented aiops benchmark for large langua...

    arXiv

Pith tools

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