LoRA-BAM: Input Filtering for Fine-tuned LLMs via Boxed Abstraction Monitors over LoRA Layers

Changshun Wu; Chih-Hong Cheng; Saddek Bensalem; Tianyi Duan

REVIEW 4 major objections 3 minor 26 references

LoRA-BAM: Input Filtering for Fine-tuned LLMs via Boxed Abstraction Monitors over LoRA Layers

T0 review · 4 major / 3 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 LoRA-BAM filters out-of-scope questions for LoRA fine-tuned LLMs by wrapping k-means clusters of LoRA feature vectors in axis-aligned boxes, and reports that at matched 95% true-positive calibration it rejects 55-91% of near-OoD medical…

desk verdict A neat, small-scale extension of the authors' box-abstraction monitor to LoRA features, with a useful paraphrase-alignment regularizer and surprisingly strong near-OoD numbers at matched FPR95, but the evaluation is too thin and the axis-aligned box orientation is an unexamined liability. read the letter →

arxiv 2506.00998 v1 pith:4EDF4ZY2 submitted 2025-06-01 cs.LG

Changshun Wu , Tianyi Duan , Saddek Bensalem , Chih-Hong Cheng This is my paper

classification cs.LG

keywords LoRAfine-tuningout-of-distributiondetectionboxedabstractionfeature-spacemonitoringparaphraseregularizationlargelanguagemodels

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

LoRA-BAM attaches an out-of-distribution (OoD) filter to a LoRA fine-tuned LLM by reading the feature vectors that flow through the LoRA branch. During fine-tuning it clusters those vectors with k-means and wraps each cluster in an axis-aligned box; at inference time, any query whose feature vector lands outside every enlarged box is flagged as OoD. Because the decision region is a union of boxes rather than a single Gaussian ellipsoid, it can carve away holes that convex boundaries miss. The paper reports that at a matched 95% true-positive calibration, LoRA-BAM rejects 55-91% of near-OoD medical questions versus 25-35% for the Mahalanobis baseline, while rejecting only 3% of paraphrased in-distribution questions. A regularization loss that pulls paraphrased questions closer in LoRA feature space is what makes the boxes tight enough to be selective without rejecting valid rephrasings.

What carries the argument

The machinery is a boxed abstraction monitor over the LoRA branch's feature vector $A\vec{v}_{in}$ (the low-rank projection applied to the input activation). Each cluster of fine-tuning features is enclosed by an axis-aligned bounding box whose per-dimension min/max are expanded by $\Delta\cdot\sigma_{i,j}$, the per-dimension standard deviation of that cluster. The union of enlarged boxes forms a non-convex acceptance region, checked at inference in $O(md)$ time, where $m$ is the number of clusters and $d$ the feature dimension. The accompanying regularization loss $\|f_A(q)-f_A(q_p)\|_2$ between original and paraphrased questions shapes the feature space so that semantically equivalent inputs land close together, letting the boxes stay selective.

What would settle it

Take a LoRA-tuned model where the in-distribution feature vectors lie along a strongly correlated low-dimensional manifold, such as a thin diagonal in 2-D. If the per-dimension boxes are large enough to cover the diagonal's extent, the union of boxes will also admit off-manifold OoD points; if the boxes are shrunk, valid paraphrases near the diagonal's ends get rejected. A direct check: compute the maximal off-diagonal correlation within each k-means cluster on the real MedQA features; if the condition number of the per-cluster covariance is high, box expansion will mis-calibrate. The method's claim would fail if a near-OoD set living inside the box hull but outside the Gaussian fit achieves lower rejection than Mahalanobis at matched FPR95.

Watch

Extended reading notes

Core claim

The central discovery is that a non-convex decision boundary built from axis-aligned boxes over LoRA-layer feature vectors is a more sensitive near-OoD detector for fine-tuned LLMs than the convex boundaries (Mahalanobis ellipsoid, cosine spherical cap) used by prior methods. The paper constructs one box per k-means cluster of fine-tuning feature vectors, expands each box by a hyperparameter times the per-dimension standard deviation, and declares a query OoD if its LoRA feature vector is outside all boxes. The expansion is calibrated on in-distribution data at FPR95, and the paraphrase-alignment loss during fine-tuning is shown to be necessary: without it, the boxed monitor's rejection rates collapse, especially at small expansion margins. Far-OoD detection remains competitive with the baselines, and the in-distribution paraphrase rejection is the lowest of the three methods.

Load-bearing premise

In-distribution LoRA feature vectors must form compact clusters that axis-aligned boxes, expanded by per-dimension standard deviation without any covariance or dimensionality modeling, can separate from out-of-distribution vectors.

Editorial extensions

If this is right

A union of axis-aligned boxes is a strictly more expressive decision region than a single ellipsoid for the same feature space, so any convex-boundary OoD method can be strengthened by swapping in boxed abstraction.
The regularization loss is necessary for boxed monitors to work at small expansion margins; without it, rejection drops sharply at $\Delta=1$ across all OoD domains.
The monitor adds only $O(md)$ containment checks per query, so it can run as a lightweight pre-filter before confidence calibration or retrieval-augmented generation.
Because the monitor reads only the LoRA branch output, it applies to LoRA and its quantized version without retraining the base weights.

Reading between the lines

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

The same boxed-abstraction construction should transfer to other parameter-efficient adapters that expose a low-rank feature vector, such as DoRA or IA3, since the monitor never touches the base-model weights.
Extending the boxes to low-dimensional PCA space, which the paper lists as future work, would let the monitor's decision region be visualized while trading off some detected OoD samples that live in discarded dimensions.
A natural hybrid: use the box union as a cheap coarse filter and run Mahalanobis distance only on queries that fall inside some box, cutting the near-OoD miss rate of both methods at the same computational budget.
The k-means cluster count $m$ is a free parameter the paper does not tune systematically; near-OoD rejection likely peaks at a $m$ that matches the number of latent medical subtopics, a testable hypothesis.

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.

Desk Editor's Note

read the letter

You should know two things about this paper. First, it is a legitimate extension of the authors' prior BAM work, not a new framework: they take the boxed-abstraction monitor from object detection, apply it to LoRA feature vectors (A*v_in), and add a regularization loss that pulls paraphrased ID questions together in feature space. Second, the headline results at matched FPR95 are real and worth attention: LoRA-BAM rejects 55% and 91% of near-OoD samples on Anatomy and Nutrition versus 25% and 35% for Mahalanobis, while rejecting only 3% of paraphrased ID questions. Those are meaningful gains, and the inference cost is trivial O(md).

What it does well: the method is simple, interpretable, and the paraphrase-alignment loss is a sensible way to reduce false alarms on valid ID variations. The paper is also honest about its limits, listing the small model, limited seeds, and single dataset.

Now the soft spots. The stress-test concern about axis-aligned boxes in an arbitrarily oriented LoRA feature space is legitimate and unaddressed. The coordinates of f_A(q) inherit the basis of the LoRA matrix, and nothing in the paper checks whether within-cluster covariance is close to diagonal. If it is not, a +3-sigma corner vector can fall inside the box while being far in Mahalanobis distance, and the reported advantage over Mahalanobis may partly reflect coordinate choice rather than expressiveness. A PCA-rotation rerun at matched FPR is the natural check, and its absence is the biggest gap. Second, the evaluation is tiny: one 0.5B model, 100 ID examples per domain, no error bars. The limitations section concedes this, but it still caps what you can conclude. Third, the ablation for the unregularized model shows only fixed-Delta rows, not a matched-FPR row, so you cannot compare operating points fairly. Fourth, the abstract says "substantially improved" based on this narrow benchmark, which oversells it.

None of this is fatal. The central claim—that at a matched FPR, the union-of-boxes detector beats Mahalanobis on near-OoD in this setting—appears supported by the table. The paper deserves a serious referee, but it needs major revisions: a robustness check against basis rotation, more models and data, error bars, and a properly matched ablation. I would send it to review, expecting the authors to tighten the empirical work before acceptance. For a reading group, it is a useful case study in how coordinate choice can affect OoD detector comparisons.

Referee Report

4 major / 3 minor

Summary. The paper proposes LoRA-BAM, an out-of-distribution (OoD) detector for LoRA fine-tuned LLMs. Feature vectors are extracted from the LoRA layer, clustered with k-means, and each cluster is enclosed by an axis-aligned box expanded by Δ·σ per dimension; a query is flagged OoD if its feature vector falls outside all boxes. A regularization loss penalizes Euclidean distance between original and paraphrased queries in the LoRA feature space during fine-tuning. Experiments with Qwen2.5-0.5B-Instruct fine-tuned on 100 MedQA Q-A pairs compare against Mahalanobis distance and cosine similarity baselines; at a matched operating point, LoRA-BAM reports higher rejection on near-OoD Anatomy/Nutrition (55%/91% vs 25%/35%) and lower rejection of paraphrased ID questions (3% vs 7%/9%).

Significance. If the reported results survive the concerns below, the paper makes a useful contribution: a lightweight (O(md) inference), interpretable, non-convex alternative to ellipsoidal OoD detectors, with a regularizer that visibly changes the feature-space geometry. The box construction is algorithmic rather than fitted to the test OoD set, the FPR95 calibration is a standard operating-point choice, and the footnote reporting a λ sweep suggests some robustness checking. The headline advantage over Mahalanobis distance in Table 1 is large and internally consistent. However, the paper does not currently rule out the possibility that the advantage stems from the coordinate system of the LoRA projection rather than from the boxed-abstraction model, and the evidence base is too narrow to support the abstract's general claim of 'substantially improved' OoD detection.

major comments (4)

[Section 2, Eq. (1) and box enlargement] The ID region is a union of axis-aligned boxes in the coordinates of f_A(q)=A v_in, a space whose basis is an artifact of the LoRA weight A. The paper provides no covariance or cluster-shape diagnostic, so the box over-approximation can be very loose: with correlated coordinates, a vector at +3σ along every coordinate can fall inside the enlarged box while being about 17σ away in Mahalanobis distance in a 32-dimensional space, so near-OoD queries can be accepted or ID paraphrases rejected purely because of axis alignment. Since Table 1's headline comparison (e.g., Anatomy 55% vs 25%, Nutrition 91% vs 35% at matched operating point) is against the covariance-aware Mahalanobis detector, the missing decisive experiment is a rerun under a PCA rotation or whitening of the LoRA features at matched FPR95; without it, the claimed advantage may reflect coordinate choice rather than the expressiveness of a union of boxes.
[Section 3.2 / Table 1] The empirical core is a single model (Qwen2.5-0.5B-Instruct), a single ID dataset of 100 Q-A pairs (Q(Med)), and no error bars or seed variation; the Limitations section concedes this. With FPR95 calibrated on 100 ID examples, the reported percentages (e.g., the 3% ID rejection rate) carry nontrivial sampling error, and the cross-domain generalization claim rests on one architecture. At least one additional model size or family and multiple seeds with confidence intervals are needed to support the abstract's 'substantially improved' claim; otherwise the conclusion should be explicitly scoped to this benchmark.
[Section 3.2 / Table 1, upper half] The ablation compares unregularized BAM at fixed Δ = 0.2–1.0, whereas the comparative claim is made at a matched operating point (TPR = 95%). A fixed-Δ comparison confounds detector quality with threshold calibration, so the text's conclusion that the regularization is 'essential' is not demonstrated at matched FPR. A TPR=95% row for the unregularized model, or an AUROC/AUPR comparison, is required before attributing the gains to the regularization term.
[Section 2, FPR95 definition] The FPR95 definition is internally inconsistent: the text first calls it 'false positive rate at 95% true positive rate' and then says Δ is adjusted so that the filter has a 95% success rate on ID-only calibration data (i.e., a 5% false positive rate); Table 1 labels rows 'TPR = 95%' without specifying which set is used to select the threshold. If the operating points are not matched in exactly the same way for all three methods, the comparison in Table 1 is not apples-to-apples. Please state the calibration protocol precisely (calibration set, target metric, and whether the threshold is global or per-domain) and report the resulting Δ or threshold values.

minor comments (3)

[Section 3.2] The text says LoRA-BAM trails the best far-OoD method by 2% on the Law domain, but Table 1 shows 95% vs 96%, a difference of one percentage point; please correct the numbers or the phrasing.
[Footnote 3] The λ sweep is reported only qualitatively ('always better', λ∈{0.1,0.5,1,5,7,10}); since the regularization is a core contribution, please include the actual numbers or a plot showing FPR95-matched performance as a function of λ.
[Section 2, Eq. (1)] The dimensionality notation is inconsistent: the LoRA matrices are defined with rank r, the feature space is called R^k, and the box definition uses R^d; please align these symbols throughout the paper.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the box monitor is algorithmically defined and benchmarked on external MedQA/MMLU data; only non-load-bearing self-citations are present.

full rationale

LoRA-BAM's construction is self-contained and algorithmic: Eq. (1) defines each axis-aligned box directly from per-dimension min/max of clustered LoRA feature vectors, and the enlargement by Delta times sigma_i,j is an explicit hyperparameter choice. The FPR95 criterion is a standard threshold calibration performed on an ID-only calibration set, not on the OoD test sets, so the reported near-OoD and far-OoD rejection rates are measured on external benchmark domains (Anatomy, Biology, Nutrition, Law, CS) rather than forced by the construction. The paraphrase regularization is a training objective that shapes the feature space; it does not by itself place any held-out test paraphrase inside a box, so the 3% ID-paraphrase rejection rate remains an empirical result. The two self-citations (Wu et al., IROS 2024; He et al., 2025) are references to prior box-abstraction and OoD works, but Section 2 provides full definitions and no uniqueness theorem or load-bearing assertion is imported from those papers. The skeptical PCA-orientation concern is a robustness/correctness issue about whether axis-aligned boxes capture the feature geometry, not a circularity: it does not show that any prediction is equivalent to an input by construction.

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

The method relies on three unstated modeling choices: clusterability of LoRA features, adequacy of axis-aligned boxes, and semantic meaning of Euclidean distance after paraphrase regularization. These are domain assumptions rather than proven properties; the paper gives only small-scale empirical evidence.

free parameters (4)

Number of clusters m
K-means cluster count used to build boxes; not reported in the paper; the decision boundary depends directly on it.
Box enlargement factor Delta = calibrated via FPR95 (exact value not reported)
Controls how much each box is expanded along each dimension; in the TPR=95 rows it is set so that 95% of ID calibration queries are accepted.
Regularization weight lambda = 5
Weight of the Euclidean-distance loss for paraphrase alignment; authors evaluated {0.1, 0.5, 1, 5, 7, 10} and used 5 for the main result.
LoRA feature extraction point (layer and dimension) = not specified
The method requires choosing which LoRA projection layer's intermediate vector A v_in is used; the paper does not specify the layer or the resulting feature dimension.

assumptions (3)

domain assumption LoRA feature vectors of in-distribution queries form compact clusters in the chosen feature space.
The method builds axis-aligned boxes around k-means clusters; if features are not clusterable in raw coordinates, boxes over-reject or under-reject. Invoked in Section 2.
domain assumption Axis-aligned boxes with per-dimension standard-deviation enlargement can separate ID paraphrases from near-OoD queries.
The method assumes the ID region is well-approximated by a union of boxes in the chosen coordinate system; no dimensionality reduction or correlation modeling is used. Section 2.
domain assumption Euclidean-distance regularization on paraphrases preserves generation quality and improves OoD discrimination.
The training objective assumes closeness in LoRA feature space corresponds to semantic equivalence; the paper reports improved results but provides no formal or large-scale evidence. Section 2, Regularization for Paraphrasing.

how reviews work

0 comments

Cite this review

Pith. "Pith review of LoRA-BAM: Input Filtering for Fine-tuned LLMs via Boxed Abstraction Monitors over LoRA Layers." pith.science (2026). https://pith.science/paper/4EDF4ZY2

@misc{pith2026250600998,
  author       = {Pith},
  title        = {Pith review of: LoRA-BAM: Input Filtering for Fine-tuned LLMs via Boxed Abstraction Monitors over LoRA Layers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4EDF4ZY2}},
  note         = {Machine review of arXiv:2506.00998}
}

read the original abstract

Fine-tuning large language models (LLMs) improves performance on domain-specific tasks but can lead to overfitting, making them unreliable on out-of-distribution (OoD) queries. We propose LoRA-BAM - a method that adds OoD detection monitors to the LoRA layer using boxed abstraction to filter questions beyond the model's competence. Feature vectors from the fine-tuning data are extracted via the LLM and clustered. Clusters are enclosed in boxes; a question is flagged as OoD if its feature vector falls outside all boxes. To improve interpretability and robustness, we introduce a regularization loss during fine-tuning that encourages paraphrased questions to stay close in the feature space, and the enlargement of the decision boundary is based on the feature variance within a cluster. Our method complements existing defenses by providing lightweight and interpretable OoD detection.

Figures

Figures reproduced from arXiv: 2506.00998 by the authors.

**Figure 2.** LoRA and extracting fracture vectors A⃗vin for constructing OoD detectors (a), and an illustrative example indicating the decision boundary between LoRABAM (two boxes) and Mahalanobis distance (one ellipse) when enclosing the feature vectors (b) ently a convex set (an ellipsoid or spherical cap), making existing methods inherently limited in their detection power. Contrarily, the decision boundary in LoRA-BAM is … view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

26 extracted references · 20 canonical work pages

[1]

Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, and 1 others. 2023. GPT-4 technical report. arXiv preprint arXiv:2303.08774

arXiv 2023
[2]

Haoyue Bai, Yifei Ming, Julian Katz-Samuels, and Yixuan Li. 2024. HYPO : Hyperspherical out-of-distribution generalization. In International Conference on Learning Representations (ICLR)

work page 2024
[3]

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

2000
[4]

Xuefeng Du, Zhaoning Wang, Mu Cai, and Sharon Li. 2022. VOS : Learning what you don't know by virtual outlier synthesis. In International Conference on Learning Representations (ICLR)

work page 2022
[5]

Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yixin Dai, Jiawei Sun, Haofen Wang, and Haofen Wang. 2023. Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997, 2:1

arXiv 2023
[6]

Jiahui Geng, Fengyu Cai, Yuxia Wang, Heinz Koeppl, Preslav Nakov, and Iryna Gurevych. 2023. A survey of confidence estimation and calibration in large language models. arXiv preprint arXiv:2311.08298

arXiv 2023
[7]

Mingrong Gong, Chaoqi Chen, Qingqiang Sun, Yue Wang, and Hui Huang. 2025. Out-of-distribution detection with prototypical outlier proxy. In AAAI Conference on Artificial Intelligence (AAAI), volume 39, pages 16835--16843

work page 2025
[8]

Weicheng He, Changshun Wu, Chih-Hong Cheng, Xiaowei Huang, and Saddek Bensalem. 2025. Mitigating hallucinations in YOLO -based object detection models: A revisit to out-of-distribution detection. arXiv preprint arXiv:2503.07330

work page arXiv 2025

Show all 26 references

[9]

Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2021. Measuring massive multitask language understanding. In International Conference on Learning Representations (ICLR)

2021
[10]

Dan Hendrycks and Kevin Gimpel. 2017. A baseline for detecting misclassified and out-of-distribution examples in neural networks. In International Conference on Learning Representations (ICLR)

2017
[11]

Edward J Hu, yelong shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. Lo RA : Low-rank adaptation of large language models. In International Conference on Learning Representations (ICLR)

2022
[12]

Kimin Lee, Kibok Lee, Honglak Lee, and Jinwoo Shin. 2018. A simple unified framework for detecting out-of-distribution samples and adversarial attacks. Advances in neural information processing systems (NeurIPS), 31

2018
[13]

Shiyu Liang, Yixuan Li, and R. Srikant. 2018. Enhancing the reliability of out-of-distribution image detection in neural networks. In International Conference on Learning Representations (ICLR)

2018
[14]

Bo Liu, Li-Ming Zhan, Zexin Lu, Yujie Feng, Lei Xue, and Xiao-Ming Wu. 2024. How good are LLMs at out-of-distribution detection? In Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING), pages 8211--8222

2024
[15]

Haodong Lu, Dong Gong, Shuo Wang, Jason Xue, Lina Yao, and Kristen Moore. 2024. Learning with mixture of prototypes for out-of-distribution detection. In International Conference on Learning Representations (ICLR)

2024
[16]

Thomas Mesnard, Cassidy Hardin, Robert Dadashi, Surya Bhupatiraju, Shreya Pathak, Laurent Sifre, Morgane Rivi \`e re, Mihir Sanjay Kale, Juliette Love, and 1 others. 2024. Gemma: Open models based on gemini research and technology. arXiv preprint arXiv:2403.08295

2024 arXiv
[17]

Hieu V Nguyen and Li Bai. 2010. Cosine similarity metric learning for face verification. In Asian Conference on Computer Vision (ACCV), pages 709--720. Springer

2010
[18]

Bart omiej Olber, Krystian Radlak, Adam Popowicz, Michal Szczepankiewicz, and Krystian Chachu a. 2023. Detection of out-of-distribution samples using binary neuron activation patterns. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3378--3387

2023
[19]

Ankit Pal, Logesh Kumar Umapathi, and Malaikannan Sankarasubbu. 2022. Medmcqa: A large-scale multi-subject multi-choice dataset for medical domain question answering. In Conference on Health, Inference, and Learning (CHIL), pages 248--260. PMLR

2022
[20]

Etienne Salimbeni, Francesco Craighero, Renata Khasanova, Milos Vasic, and Pierre Vandergheynst. 2024. Beyond fine-tuning: LoRA modules boost near-ood detection and llm security. In ICLR 2024 Workshop on Secure and Trustworthy Large Language Models

2024
[21]

Kristina P Sinaga and Miin-Shen Yang. 2020. Unsupervised k-means clustering algorithm. IEEE access, 8:80716--80727

2020
[22]

Yiyou Sun, Yifei Ming, Xiaojin Zhu, and Yixuan Li. 2022. Out-of-distribution detection with deep nearest neighbors. In International Conference on Machine Learning (ICML), pages 20827--20840. PMLR

2022
[23]

Changshun Wu, Weicheng He, Chih-Hong Cheng, Xiaowei Huang, and Saddek Bensalem. 2024. BAM : box abstraction monitors for real-time OoD detection in object detection. In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 2632--2638. IEEE

2024
[24]

Kai Xu, Rongyu Chen, Gianni Franchi, and Angela Yao. 2024. Scaling for training time and post-hoc out-of-distribution detection enhancement. In International Conference on Learning Representations (ICLR)

2024
[25]

An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, and 1 others. 2024. Qwen2.5 technical report. arXiv preprint arXiv:2412.15115

2024 arXiv
[26]

Jinsong Zhang, Qiang Fu, Xu Chen, Lun Du, Zelin Li, Gang Wang, xiaoguang Liu, Shi Han, and Dongmei Zhang. 2023. Out-of-distribution detection based on in-distribution data patterns memorization with modern hopfield energy. In International Conference on Learning Representations (ICLR)

2023

Pith tools

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

[1] [1]

Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, and 1 others. 2023. GPT-4 technical report. arXiv preprint arXiv:2303.08774

arXiv 2023

[2] [2]

Haoyue Bai, Yifei Ming, Julian Katz-Samuels, and Yixuan Li. 2024. HYPO : Hyperspherical out-of-distribution generalization. In International Conference on Learning Representations (ICLR)

work page 2024

[3] [3]

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

2000

[4] [4]

Xuefeng Du, Zhaoning Wang, Mu Cai, and Sharon Li. 2022. VOS : Learning what you don't know by virtual outlier synthesis. In International Conference on Learning Representations (ICLR)

work page 2022

[5] [5]

Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yixin Dai, Jiawei Sun, Haofen Wang, and Haofen Wang. 2023. Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997, 2:1

arXiv 2023

[6] [6]

Jiahui Geng, Fengyu Cai, Yuxia Wang, Heinz Koeppl, Preslav Nakov, and Iryna Gurevych. 2023. A survey of confidence estimation and calibration in large language models. arXiv preprint arXiv:2311.08298

arXiv 2023

[7] [7]

Mingrong Gong, Chaoqi Chen, Qingqiang Sun, Yue Wang, and Hui Huang. 2025. Out-of-distribution detection with prototypical outlier proxy. In AAAI Conference on Artificial Intelligence (AAAI), volume 39, pages 16835--16843

work page 2025

[8] [8]

Weicheng He, Changshun Wu, Chih-Hong Cheng, Xiaowei Huang, and Saddek Bensalem. 2025. Mitigating hallucinations in YOLO -based object detection models: A revisit to out-of-distribution detection. arXiv preprint arXiv:2503.07330

work page arXiv 2025