Pith. sign in

REVIEW 4 major objections 5 minor 8 cited by

LoRA-LiteE: A Computationally Efficient Framework for Chatbot Preference-Tuning

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A two-model LoRA ensemble reaches 80.2% accuracy on chatbot preference prediction, matching un-finetuned GPT-4 and beating larger models when training time is short.

desk verdict Plainly written, sensible LoRA+SFT ensemble for Chatbot Arena preference prediction, but the headline 'comparable to GPT-4' rests on a single accuracy point with empirically fitted ensemble weights and no error bars, so the evidence is thinner than the claim. read the letter →

arxiv 2411.09947 v2 pith:EHD3A5RB submitted 2024-11-15 cs.CL

Yahe Yang , Chunliang Tao , Xiaojing Fan This is my paper · ORCID
classification cs.CL
keywords LoRAensemblelearningpreferencetuningChatbotArenasupervisedfine-tuningparameter-efficientLLMalignmentevaluation
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 aims to show that chatbot preference tuning does not have to go through the costly reinforcement-learning-from-human-feedback (RLHF) pipeline. It proposes LoRA-LiteE, a framework that takes two small chat models (Gemma-2-9b and Llama-3-8b), fine-tunes each with Low-Rank Adaptation (LoRA) on the Chatbot Arena preference dataset, and combines their predictions by weighted averaging. On that task the ensemble reaches 80.2% accuracy and a log loss of 0.99, which is close to the un-finetuned GPT-4 baseline (78.3% accuracy, 1.07 log loss) and well above the individual small models. The paper also reports that when fine-tuning time is capped, the ensemble overtakes the larger single models (Gemma-2-27b and Llama-3-70b), which only end up higher when given ample training time. A sympathetic reading is that parameter-efficient fine-tuning plus ensembling is a practical route to preference-aligned chatbots in resource-constrained settings.

What carries the argument

The load-bearing machinery is the LoRA-Lite ensemble: two lightweight chat models, Gemma-2-9b and Llama-3-8b, each fine-tuned with Low-Rank Adaptation (LoRA), which freezes the pretrained weights and injects trainable low-rank matrices so that only a small fraction of parameters is updated. After task-specific supervised fine-tuning on the Chatbot Arena training split, the two models output three-class probability vectors (model A wins, model B wins, tie), and the ensemble combines them through the weighted average $P_{\mathrm{final}} = 0.7P_{\mathrm{gemma}} + 0.3P_{\mathrm{llama}}$. This aggregation converts two moderate-quality single models (72.3% and 75.1% accuracy) into an 80.2% predictor, and it does so with each member trainable on a single A100 GPU. The same equation is also the point where the argument is most delicate, since the 0.7/0.3 weights are reported as empirically determined rather than as the output of a validation procedure.

What would settle it

On a held-out split of Chatbot Arena, re-run LoRA-LiteE with the ensemble weight selected by a validation search over $\alpha \in [0,1]$ in $P_{\mathrm{final}} = \alpha P_{\mathrm{gemma}} + (1-\alpha)P_{\mathrm{llama}}$; if test accuracy only reaches 80.2% when $\alpha$ is chosen on the test set, the match to GPT-4 is an artifact of weight tuning.

Watch

Extended reading notes

Core claim

The central discovery is that a weighted average of LoRA-adapted predictions from two small chat models can rival a much larger, RLHF-trained model on the specific task of predicting which of two chatbot responses a human judge prefers. LoRA-LiteE achieves 80.2% accuracy and 0.99 log loss on Chatbot Arena, compared with 78.3% and 1.07 for the un-finetuned GPT-4 baseline, 84.8% and 0.72 for Gemma-2-27b, 86.9% and 0.79 for Llama-3-70b, and 72.3%/75.1% for the two small models fine-tuned individually. The mechanism is the ensemble rule $P_{\mathrm{final}} = 0.7P_{\mathrm{gemma}} + 0.3P_{\mathrm{llama}}$, which the paper describes as empirically determined. Because the trained ensemble is cheaper to produce than an RLHF run, the authors argue it offers a scalable and accessible alternative for preference tuning under limited compute.

Load-bearing premise

The central result depends on the ensemble weights (0.7 and 0.3) having been determined without access to the test set; the paper only says they were empirically determined and gives no validation procedure, so if they were tuned on the evaluation data, the reported 80.2% accuracy would not transfer to new data.

Editorial extensions

If this is right

  • If the reported results hold, preference prediction for chatbots can be run with two small open models and a single GPU, rather than with an RLHF-trained model such as GPT-4.
  • Under short training budgets, an ensemble of small LoRA-tuned models can beat a single much larger model, so model scale alone does not determine preference-prediction quality.
  • The ensemble also improves probability estimates, lowering log loss from 1.07 (GPT-4) to 0.99, which matters for downstream uses where confidence scores rather than hard choices are consumed.
  • Resource-constrained organizations that cannot run RLHF can still deploy preference-aligned chatbots by using supervised fine-tuning plus LoRA plus ensembling.
  • The practical recommendation is to choose between a small ensemble and a large single model based on available training time, since the ranking flips when the budget is small.

Reading between the lines

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

  • The general lesson that a heterogeneous pair of model families (Gemma and Llama) produces an ensemble gain is testable: swapping in two checkpoints from the same family should reduce or eliminate the gain if diversity is what matters.
  • A validation-based search over the ensemble weight would make the method more reproducible and would clarify whether 80.2% is stable or is an upper bound selected on the test set; the paper leaves this unresolved.
  • The time-versus-accuracy crossover suggests a practical scheduling rule: train the small ensemble first and only launch a larger model if the budget comfortably exceeds the crossover time, a threshold the paper does not identify.
  • The same weighted-averaging recipe could be transferred to other preference datasets or to direct preference optimization objectives, but the paper only demonstrates it on Chatbot Arena, so any claim of generality is extrapolation.
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 / 5 minor

Summary. The paper proposes LoRA-LiteE, an ensemble of two parameter-efficiently fine-tuned small language models (Gemma-2-9b and Llama-3-8b) for predicting human preferences on the Chatbot Arena dataset. The final prediction is a weighted average of the two models' softmax outputs, with weights 0.7 and 0.3. The authors report that this ensemble reaches 80.2% accuracy, slightly above un-finetuned GPT-4's 78.3%, and they claim that under limited fine-tuning time it outperforms larger fine-tuned models (Gemma-2-27b, Llama-3-70b). The paper includes training-mode comparisons, LoRA hyperparameters, and a limitations paragraph.

Significance. The practical question the paper addresses is timely: can a small LoRA ensemble replace a much larger model for preference prediction in resource-constrained settings? If the empirical claims were rigorously supported, the result would be useful and accessible, especially because the authors use a public benchmark and report their fine-tuning hyperparameters. The claimed benefits of reduced computational cost and comparable accuracy to GPT-4 are potentially significant. However, the significance is currently limited by the absence of any uncertainty quantification, a described train/test split, a validation procedure for the ensemble weights, and the missing Figure 1 that is central to the resource-constrained claim.

major comments (4)
  1. [Section IV-A, Eq. (1)] The ensemble weights in Equation (1) are described only as 'empirically determined,' with no mention of a validation set, cross-validation, or any separation between weight selection and final evaluation. Since the headline result (80.2% vs. 78.3% for GPT-4) depends on these two free parameters, the reported accuracy cannot be taken as an unbiased estimate of generalization unless the weights were selected on data disjoint from the test set. Please specify exactly how the 0.7/0.3 weights were chosen and report performance on a truly held-out test set.
  2. [Table III] Table III reports only point estimates of accuracy and log loss, with no error bars, confidence intervals, significance tests, or test-set size. The central claim of 'comparable performance to GPT-4' rests on a 1.9-percentage-point difference, which may be within sampling noise. Report the number of test samples, and provide bootstrapped confidence intervals or repeated-seed results so the reader can assess whether the difference is meaningful.
  3. [Section IV-C and Figure 1] Figure 1, which is the sole evidence for the claim that LoRA-LiteE outperforms larger models under limited resource constraints, is not present in the manuscript. The text asserts that the figure reveals a crossover within the first seven hours of fine-tuning, but no actual figure, axes, or numerical thresholds are given. Add the figure with error bars and a precise description of how training time and early stopping were measured for each model; without this evidence the resource-constrained claim cannot be evaluated.
  4. [Section IV-C] The comparison of fine-tuning time across models is not adequately controlled. The paper does not specify batch sizes, number of GPUs, whether larger models had the same early-stopping criteria, or how often checkpoints were evaluated. Because the efficiency claim is about 'limited resource constraints,' the authors should report total compute (e.g., GPU-hours) and, ideally, inference cost for the ensemble versus each single model, rather than wall-clock hours to a particular accuracy.
minor comments (5)
  1. [Section III-A] The dataset is described as containing 57,477 training samples, but the test-set size and the train/test split procedure are never given; this information should be added to Section IV-A.
  2. [Equation (4)] The indicator function in the accuracy formula appears as '⊮', which looks like a typographical artifact; it should be typeset as a standard indicator function (e.g., \mathbb{1}).
  3. [Throughout] The dataset name is spelled inconsistently as both 'ChatBot Arena' and 'Chatbot Arena'; please standardize it.
  4. [References] Reference [33] contains the typo 'Syste m design' in the title.
  5. [Section IV-A] No random seeds or number of repeated training runs are reported, which makes it impossible to assess training variability even with the same hyperparameters.

Circularity Check

2 steps flagged · score 6.0 of 10

Test-set early stopping and empirically fitted ensemble weights make the reported headline accuracy a fitted value rather than an independent prediction.

  1. fitted input called prediction [Section IV-A, Equation (1)]
    "The final ensemble prediction is calculated using empirically determined weights in Equation 1. P_final = 0.7 × P_gemma + 0.3 × P_llama (1)"

    Equation (1) is the entire ensemble definition, and its only free parameters are the two weights. The paper states they are 'empirically determined' but never reports a validation split, cross-validation, or any protocol separating weight selection from evaluation. The headline result (LoRA-LiteE 80.2% vs GPT-4 78.3%) is the accuracy of this weighted prediction against the test labels. Since no validation protocol is reported, the paper does not establish that the reported accuracy is independent of the weight-fitting step; the reported accuracy is the value obtained by plugging the fitted weights into Equation (1). The central comparison is therefore indexed by a fitted parameter whose selection procedure is not separated from the evaluation.

  2. fitted input called prediction [Section IV-C, Table III and Figure 1]
    "Best Performance Comparison Table III summarizes the best performance achieved with early stopping criteria for each model. ... we analyze the relationship between fine-tuning time (in hours) and test set accuracy of the LoRA-LiteE and the two larger counterparts."

    Early stopping is a model-selection step. The paper's Figure 1 and Table III use test-set accuracy to pick and report the best checkpoint for each fine-tuned model, so the reported accuracy is the maximum over training time of the test-set accuracy. That maximum is, by construction, a function of the test labels, making the reported 'best performance' a fitted statistic rather than an out-of-sample prediction. The comparison against GPT-4, which is not fine-tuned and has no such selection, and the resource-constrained 'outperforms larger models' claim are therefore biased by test-set selection.

full rationale

The paper's core methodology is not circular in its derivation: LoRA fine-tuning and ensembling are standard techniques, and the individual base models are externally specified. There is no load-bearing self-citation, no imported uniqueness theorem, and no ansatz smuggled in via citation. However, the empirical evaluation of the central claim is partially circular in the fitted-input sense. The ensemble prediction is defined by Equation (1) with explicitly 'empirically determined' weights, and no validation procedure is described to show the weights were fixed independently of the test labels. Additionally, Table III and Figure 1 report the best test-set accuracy achieved with early stopping, which means the test labels were used to select the reported checkpoint. Both steps make the headline accuracy a fitted value rather than an independent prediction: the 80.2% versus GPT-4's 78.3% margin, and the resource-constrained outperformance claim, are not shown to be robust to the fitting of these parameters. Because the individual model accuracies and the GPT-4 comparison are independent inputs, the circularity is partial rather than total, but it directly affects the paper's main conclusion.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The paper's central claim rests on a small number of background domain assumptions and on two empirically fitted ensemble weights. No new entities are introduced.

free parameters (2)
  • ensemble weight for Gemma-2-9b prediction = 0.7
    In Equation 1, P_final = 0.7 * P_gemma + 0.3 * P_llama; the paper states this weight is 'empirically determined', making the headline performance dependent on a fitted scalar.
  • ensemble weight for Llama-3-8b prediction = 0.3
    Companion weight in Equation 1, also empirically determined.
assumptions (3)
  • domain assumption LoRA fine-tuning preserves the base model's conversational understanding while reducing trainable parameters
    The paper relies on this claim from the LoRA literature [13] to justify the small-model approach.
  • domain assumption Weighted averaging of probability outputs from multiple models improves preference prediction accuracy
    The ensemble strategy assumes that complementary model strengths can be combined via a fixed weighted average.
  • domain assumption The Chatbot Arena dataset provides reliable, unbiased human preference labels
    The entire evaluation is based on this dataset's labels without discussion of label noise or selection bias.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LoRA-LiteE: A Computationally Efficient Framework for Chatbot Preference-Tuning." pith.science (2026). https://pith.science/paper/EHD3A5RB

@misc{pith2026241109947,
  author       = {Pith},
  title        = {Pith review of: LoRA-LiteE: A Computationally Efficient Framework for Chatbot Preference-Tuning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EHD3A5RB}},
  note         = {Machine review of arXiv:2411.09947}
}
read the original abstract

Effective preference tuning is pivotal in aligning chatbot responses with human expectations, enhancing user satisfaction and engagement. Traditional approaches, notably Reinforcement Learning from Human Feedback (RLHF) as employed in advanced models like GPT-4, have demonstrated considerable success in this domain. However, RLHF methods are often computationally intensive and resource-demanding, limiting their scalability and accessibility for broader applications. To address these challenges, this study introduces LoRA-Lite Ensemble (LoRA-LiteE), an innovative framework that combines Supervised Fine-tuning (SFT) with Low-Rank Adaptation (LoRA) and Ensemble Learning techniques to effectively aggregate predictions of lightweight models, which aim to achieve a balance between the performance and computational cost. Utilizing the Chatbot Arena benchmark dataset, we conduct a comprehensive comparative analysis among our LoRA-LiteE model, corresponding base models at different scales, and GPT-4 trained with RLHF. Our empirical results demonstrate that the proposed LoRA-LiteE model achieves comparable performance to un-finetuned GPT-4 and outperforms the single larger-scale models under limited resource constraints. These findings highlight that our LoRA-LiteE provides a feasible and efficient methodology for human preference prediction in chatbot systems, enhancing scalability and accessibility, and thereby broadening the applicability of preference-tuned chatbots in resource-constrained environments.

Figures

Figures reproduced from arXiv: 2411.09947 by the authors.

Figure 1
Figure 1. Comparative Accuracy Gains during Fine-tuning Across Models [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 8 Pith papers

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

  1. Feature Alignment-Based Knowledge Distillation for Efficient Compression of Large Language Models

    cs.CL 2024-12 reject novelty 3.0 of 10

    A multi-layer feature alignment knowledge distillation loss is reported to make a compact student model nearly match GPT-4 on perplexity, BLEU, ROUGE, and CER, but the experimental details are not provided.

  2. Dynamic Adaptation of LoRA Fine-Tuning for Efficient and Task-Specific Optimization of Large Language Models

    cs.CL 2025-01 reject novelty 2.0 of 10

    Dynamic LoRA, a layer-wise adaptive variant of LoRA, reportedly improves GLUE accuracy from 87.4% to 88.1% at only 0.1% more trainable parameters, but the write-up lacks reproducibility.

  3. Deep Learning in Image Classification: Evaluating VGG19's Performance on Complex Visual Data

    cs.CV 2024-12 reject novelty 2.0 of 10

    On a public chest X-ray dataset, VGG19 is reported to outperform SVM, XGBoost, MLP, and ResNet50 in accuracy, AUC, F1, and recall, but without a reproducible evaluation protocol.

  4. Optimizing Large Language Models with an Enhanced LoRA Fine-Tuning Algorithm for Efficiency and Robustness in NLP Tasks

    cs.CL 2024-12 reject novelty 2.0 of 10

    A modified LoRA update with per-matrix learning rates and an object-detection-style density term is reported to slightly improve QQP accuracy over GPT-4 baselines.

  5. Computer Vision-Driven Gesture Recognition: Toward Natural and Intuitive Human-Computer

    cs.CV 2024-12 reject novelty 2.0 of 10

    A CNN-LSTM gesture recognizer with a decorative 3D skeleton visualization that reports unverifiable accuracy and speed numbers.

  6. Leveraging Convolutional Neural Network-Transformer Synergy for Predictive Modeling in Risk-Based Applications

    q-fin.RM 2024-12 reject novelty 2.0 of 10

    A CNN+Transformer hybrid is reported to outperform tree-based baselines on credit default prediction, with accuracy 0.8197 and AUC 0.7921.

  7. Accurate Medical Named Entity Recognition Through Specialized NLP Models

    cs.CL 2024-12 reject novelty 2.0 of 10

    The paper reports BioBERT as the best among five models on MIMIC-III NER, but the experimental description is too sparse to verify the numbers.

  8. Optimizing Multi-Task Learning for Enhanced Performance in Large Language Models

    cs.CL 2024-12 reject novelty 2.0 of 10

    A multi-task GPT-4 model is said to beat single-task GPT-4, GPT-3, BERT, and Bi-LSTM on classification and summarization, but the experimental evidence is not reported.

Reference graph

Works this paper leans on

45 extracted references · 21 canonical work pages · cited by 8 Pith papers

  1. [1]

    An overview of chatbot technology

    Eleni Adamopoulou and Lefteris Moussiades. An overview of chatbot technology. In IFIP international conference on artificial intelligence applications and innovations , pages 373–383. Springer, 2020

    work page 2020

  2. [2]

    Conversational agents in healthcare: a systematic review

    Liliana Laranjo, Adam G Dunn, Huong Ly Tong, Ahmet Baki Kocaballi, Jessica Chen, Rabia Bashir, Didi Surian, Blanca Gallego, Farah Magrabi, Annie YS Lau, et al. Conversational agents in healthcare: a systematic review. Journal of the American Medical Informatics Association , 25(9):1248–1258, 2018

    work page 2018

  3. [3]

    Unleashing the potential of chatbots in education: A state-of-the-art analysis

    Rainer Winkler and Matthias S ¨ollner. Unleashing the potential of chatbots in education: A state-of-the-art analysis. In Academy of Management Proceedings , volume 2018, page 15903. Academy of Management Briarcliff Manor, NY 10510, 2018

    work page 2018

  4. [4]

    Neural approaches to conversational ai

    Jianfeng Gao, Michel Galley, and Lihong Li. Neural approaches to conversational ai. In The 41st international ACM SIGIR conference on research & development in information retrieval , pages 1371–1374, 2018

    work page 2018

  5. [5]

    Learning from dialogue after deployment: Feed yourself, chatbot! arXiv preprint arXiv:1901.05415 , 2019

    Braden Hancock, Antoine Bordes, Pierre-Emmanuel Mazare, and Jason Weston. Learning from dialogue after deployment: Feed yourself, chatbot! arXiv preprint arXiv:1901.05415 , 2019

    arXiv 1901

  6. [6]

    Fine- tuning language models from human preferences

    Daniel M Ziegler, Nisan Stiennon, Jeffrey Wu, Tom B Brown, Alec Radford, Dario Amodei, Paul Christiano, and Geoffrey Irving. Fine- tuning language models from human preferences. arXiv preprint arXiv:1909.08593, 2019

    arXiv 1909

  7. [7]

    A reduction of imitation learning and structured prediction to no-regret online learning

    St ´ephane Ross, Geoffrey Gordon, and Drew Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. In Proceedings of the fourteenth international conference on artificial intelligence and statistics , pages 627–635. JMLR Workshop and Conference Proceedings, 2011

    work page 2011

  8. [8]

    Dialog-based language learning

    Jason E Weston. Dialog-based language learning. Advances in Neural Information Processing Systems , 29, 2016

    work page 2016

Show all 45 references
  1. [9]

    Deep reinforcement learning from human prefer- ences

    Paul F Christiano, Jan Leike, Tom Brown, Miljan Martic, Shane Legg, and Dario Amodei. Deep reinforcement learning from human prefer- ences. Advances in neural information processing systems , 30, 2017

    2017

  2. [10]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wain- wright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. Advances in neural information processing systems, 35:27...

    2022

  3. [11]

    A contrastive deep learning approach to cryptocurrency portfolio with us treasuries

    Zichao Li, Bingyang Wang, and Ying Chen. A contrastive deep learning approach to cryptocurrency portfolio with us treasuries. Journal of Computer Technology and Applied Mathematics , 1(3):1–10, 2024

    2024

  4. [12]

    Gpt-4 technical report

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023

    2023 arXiv

  5. [13]

    Lora: Low-rank adaptation of large language models

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 , 2021

    2021 arXiv

  6. [14]

    Chatbot arena: An open platform for evaluating llms by human preference

    Wei-Lin Chiang, Lianmin Zheng, Ying Sheng, Anastasios Nikolas An- gelopoulos, Tianle Li, Dacheng Li, Hao Zhang, Banghua Zhu, Michael Jordan, Joseph E Gonzalez, et al. Chatbot arena: An open platform for evaluating llms by human preference. arXiv preprint arXiv:2403.04132, 2024

    2024 arXiv

  7. [15]

    Incorporating economic indicators and market sentiment effect into us treasury bond yield prediction with machine learning

    Zichao Li, Bingyang Wang, and Ying Chen. Incorporating economic indicators and market sentiment effect into us treasury bond yield prediction with machine learning. Journal of Infrastructure, Policy and Development, 8(9):7671, 2024

    2024

  8. [16]

    Llama: Open and efficient foundation language models

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie- Anne Lachaux, Timoth´ee Lacroix, Baptiste Rozi `ere, Naman Goyal, Eric Hambro, Faisal Azhar, et al. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971 , 2023

    2023 arXiv

  9. [17]

    Gemma 2: Im- proving open language models at a practical size

    Gemma Team, Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupatiraju, L ´eonard Hussenot, Thomas Mes- nard, Bobak Shahriari, Alexandre Ram ´e, et al. Gemma 2: Im- proving open language models at a practical size. arXiv preprint arXiv:2408.00118, 2024

    2024 arXiv

  10. [18]

    Ratt: Athought structure for coherent and correct llmreasoning

    Jinghan Zhang, Xiting Wang, Weijieying Ren, Lu Jiang, Dongjie Wang, and Kunpeng Liu. Ratt: Athought structure for coherent and correct llmreasoning. arXiv preprint arXiv:2406.02746 , 2024

    2024 arXiv

  11. [19]

    Thought space explorer: Navigating and expanding thought space for large language model reasoning

    Jinghan Zhang, Fengran Mo, Xiting Wang, and Kunpeng Liu. Thought space explorer: Navigating and expanding thought space for large language model reasoning. arXiv preprint arXiv:2410.24155 , 2024

    2024

  12. [20]

    Direct preference optimization: Your language model is secretly a reward model

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

    2024

  13. [21]

    In-context time series predictor

    Jiecheng Lu, Yan Sun, and Shihao Yang. In-context time series predictor. arXiv preprint arXiv:2405.14982 , 2024

    2024

  14. [22]

    Utilizing large language models for information extraction from real estate transactions

    Yu Zhao and Haoxiang Gao. Utilizing large language models for information extraction from real estate transactions. arXiv preprint arXiv:2404.18043, 2024

    2024 arXiv

  15. [23]

    Using large language models in real estate transactions: A few-shot learning approach

    Shiqi Yang, Yu Zhao, and Haoxiang Gao. Using large language models in real estate transactions: A few-shot learning approach. OSF Preprints, May 2024

    2024

  16. [24]

    Blendsql: A scalable dialect for unifying hybrid question answering in relational algebra

    Parker Glenn, Parag Pravin Dakle, Liang Wang, and Preethi Raghavan. Blendsql: A scalable dialect for unifying hybrid question answering in relational algebra. arXiv preprint arXiv:2402.17882 , 2024

    2024 arXiv

  17. [25]

    Accurate training of web-based question answering systems with feedback from ranked users

    Liang Wang, Ivano Lauriola, and Alessandro Moschitti. Accurate training of web-based question answering systems with feedback from ranked users. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 5: Industry Track) , pages 660–667, 2023

    2023

  18. [26]

    Predicting stock prices with finbert-lstm: Integrating news sentiment analysis

    Wenjun Gu, Yihao Zhong, Shizun Li, Changsong Wei, Liting Dong, Zhuoyue Wang, and Chao Yan. Predicting stock prices with finbert-lstm: Integrating news sentiment analysis. arXiv preprint arXiv:2407.16150 , 2024

    2024 arXiv

  19. [27]

    Autonomous navigation of unmanned vehicle through deep reinforcement learning

    Letian Xu, Jiabei Liu, Haopeng Zhao, Tianyao Zheng, Tongzhou Jiang, and Lipeng Liu. Autonomous navigation of unmanned vehicle through deep reinforcement learning. arXiv preprint arXiv:2407.18962 , 2024

    2024 arXiv

  20. [28]

    Can speculative sampling accelerate react without compromising reasoning quality? In The Second Tiny Papers Track at ICLR 2024

    Han Xu, Jingyang Ye, Yutong Li, and Haipeng Chen. Can speculative sampling accelerate react without compromising reasoning quality? In The Second Tiny Papers Track at ICLR 2024

    2024

  21. [29]

    Cops: Empowering llm agents with provable cross-task experience sharing

    Chen Yang, Chenyang Zhao, Quanquan Gu, and Dongruo Zhou. Cops: Empowering llm agents with provable cross-task experience sharing. arXiv preprint arXiv:2410.16670 , 2024

    2024 arXiv

  22. [30]

    Integrated optimization of large language models: Synergizing data utilization and compression techniques

    Xinjin Li, Yu Ma, Yangchen Huang, Xingqi Wang, Yuzhen Lin, and Chenxi Zhang. Integrated optimization of large language models: Synergizing data utilization and compression techniques. 2024

    2024

  23. [31]

    Harnessing llms for api interactions: A framework for classification and synthetic data generation

    Chunliang Tao, Xiaojing Fan, and Yahe Yang. Harnessing llms for api interactions: A framework for classification and synthetic data generation. arXiv preprint arXiv:2409.11703 , 2024

    2024 arXiv

  24. [32]

    Towards resilient and efficient llms: A comparative study of efficiency, performance, and adversarial robustness

    Xiaojing Fan and Chunliang Tao. Towards resilient and efficient llms: A comparative study of efficiency, performance, and adversarial robustness. arXiv preprint arXiv:2408.04585 , 2024

    2024 arXiv

  25. [33]

    Towards federated learning at scale: Syste m design

    Keith Bonawitz. Towards federated learning at scale: Syste m design. arXiv preprint arXiv:1902.01046 , 2019

    1902 arXiv

  26. [34]

    Fednlp: A research platform for federated learning in natural language processing

    Bill Yuchen Lin, Chaoyang He, Zihang Zeng, Hulin Wang, Yufen Huang, Mahdi Soltanolkotabi, Xiang Ren, and Salman Avestimehr. Fednlp: A research platform for federated learning in natural language processing. arXiv preprint arXiv:2104.08815 , 2021

    2021 arXiv

  27. [35]

    Pmfl: Partial meta-federated learning for heterogeneous tasks and its applications on real-world medical records

    Tianyi Zhang, Shirui Zhang, Ziwei Chen, Yoshua Bengio, and Dianbo Liu. Pmfl: Partial meta-federated learning for heterogeneous tasks and its applications on real-world medical records. In 2022 IEEE International Conference on Big Data (Big Data) , pages 4453–4462. IEEE, 2022

    2022

  28. [36]

    Uncertainty-based extensible codebook for discrete federated learning in heterogeneous data silos

    Tianyi Zhang, Yu Cao, and Dianbo Liu. Uncertainty-based extensible codebook for discrete federated learning in heterogeneous data silos. arXiv preprint arXiv:2402.18888 , 2024

    2024 arXiv

  29. [37]

    Roberta: A robustly optimized bert pretraining approach

    Yinhan Liu. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692 , 364, 2019

    1907 arXiv

  30. [38]

    Electra: Pre-training text encoders as discriminators rather than generators

    K Clark. Electra: Pre-training text encoders as discriminators rather than generators. arXiv preprint arXiv:2003.10555 , 2020

    2003 arXiv

  31. [39]

    Ensemble methods in machine learning

    Thomas G Dietterich. Ensemble methods in machine learning. In International workshop on multiple classifier systems , pages 1–15. Springer, 2000

    2000

  32. [40]

    Restful- llama: Connecting user queries to restful apis

    Han Xu, Ruining Zhao, Jindong Wang, and Haipeng Chen. Restful- llama: Connecting user queries to restful apis. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Process- ing: Industry Track, pages 1433–1443, 2024

    2024

  33. [41]

    Dual learning for machine translation

    Di He, Yingce Xia, Tao Qin, Liwei Wang, Nenghai Yu, Tie-Yan Liu, and Wei-Ying Ma. Dual learning for machine translation. Advances in neural information processing systems , 29, 2016

    2016

  34. [42]

    An ensemble approach to stock price prediction using deep learning and time series models

    Mujie Sui, Cheng Zhang, Li Zhou, Shuhan Liao, and Changsong Wei. An ensemble approach to stock price prediction using deep learning and time series models. 2024

    2024

  35. [43]

    Meta learning enabled adversarial defense

    Yiyi Tao. Meta learning enabled adversarial defense. In 2023 IEEE International Conference on Sensors, Electronics and Computer Engi- neering (ICSECE), pages 1326–1330. IEEE, 2023

    2023

  36. [44]

    Steerdiff: Steering towards safe text-to-image diffusion models

    Hongxiang Zhang, Yifeng He, and Hao Chen. Steerdiff: Steering towards safe text-to-image diffusion models. arXiv preprint arXiv:2410.02710 , 2024

    2024

  37. [45]

    Nevlp: Noise- robust framework for efficient vision-language pre-training

    Yiyi Tao, Zhuoyue Wang, Hang Zhang, and Lun Wang. Nevlp: Noise- robust framework for efficient vision-language pre-training. arXiv preprint arXiv:2409.09582, 2024

    2024 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.