Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Continual Hyperbolic Learning of Instances and Classes

T0 review · 4 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read By embedding the instance–class hierarchy in hyperbolic space, HyperCLIC lifts continual instance accuracy from 20.05% to 41.76% on EgoObjects, with stronger class-level recognition and milder mistakes.

desk verdict A genuinely new task and large empirical gains on EgoObjects, but the hyperbolic-specific claim is under-supported because the method bundles a fixed future-class hierarchy with hyperbolic geometry and then tests with Euclidean inference. read the letter →

arxiv 2506.10710 v1 pith:4W77OL6B submitted 2025-06-12 cs.CV cs.AIcs.LG

classification cs.CVcs.AIcs.LG
keywords continuallearninghyperbolicgeometryinstancerecognitionhierarchicalclassificationclass-incrementalknowledgedistillationPoincaréballEgoObjects
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 a new continual-learning task: a model must keep recognizing both specific object instances and the broader classes they belong to as new instances arrive over time, rather than mastering one granularity at the expense of the other. It observes that instances and classes form a single tree-shaped hierarchy, and argues that hyperbolic space — a geometry whose volume grows exponentially, like a branching tree — is the natural place to store that hierarchy while learning incrementally. The proposed method, HyperCLIC, first embeds the class–instance tree into fixed hyperbolic prototypes; it then trains with a hyperbolic classification loss that pulls each sample toward its instance prototype and a hyperbolic distillation loss that keeps the hyperbolic logits of older classes stable. On EgoObjects, with 1,110 instances across 277 classes, the method reports 41.76% instance-level accuracy trained from scratch versus 20.05% for iCaRL, together with higher class- and superclass-level accuracy, lower forgetting, and hierarchically less severe errors. Read sympathetically, the paper's claim is that living in hyperbolic space is what lets a single continual learner serve both levels of granularity at once.

What carries the argument

The load-bearing object is the precomputed set of hyperbolic prototypes for every node of the class–instance tree. It is built in three passes: the Poincaré distance loss of Nickel & Kiela places nodes so that hyperbolic distance tracks graph distance, the max-margin entailment-cone loss of Ganea et al. forces each child's embedding to lie inside its parent's cone, and a separation loss, borrowed from Long et al., stops the instance prototypes from collapsing together. During continual training the Euclidean backbone output is mapped into the ball by the exponential map, the hyperbolic logit for label $y$ is $h(z,y) = -d_{\mathbb{B}}(z, P_y)/\tau$, and both the classification and distillation losses are cross-entropies over these hyperbolic logits. This mechanism is what is supposed to let one feature space recognize instances, classes, and superclasses simultaneously while resisting catastrophic forgetting.

What would settle it

Train the same iCaRL-style pipeline on EgoObjects with the hyperbolic prototype loss replaced by an ordinary Euclidean prototype loss — identical distillation, exemplars, and nearest-mean classifier — and compare instance-, class-, and superclass-level accuracy plus LCA distance; if the Euclidean variant matches HyperCLIC, the reported gains are not attributable to hyperbolic geometry. A second check is to run nearest-exemplar-mean classification in hyperbolic space at test time and see whether it at least ties the Euclidean version; if it does not, the Euclidean representations the deployed model uses are not in fact aligned with the hierarchy.

Watch

Extended reading notes

Core claim

HyperCLIC treats instances, classes, superclasses, and ancestors as nodes of one tree and embeds that tree in advance into the Poincaré ball, a model of hyperbolic space in which the distance between two prototypes mirrors the graph distance between their nodes. Visual features from a Euclidean backbone are pushed into the same ball by the exponential map $\exp_0$, and the logit for a label is the negative hyperbolic distance to that label's prototype, scaled by temperature $\tau = 0.1$. Training combines a hyperbolic classification loss for the current task with a hyperbolic distillation loss over exemplars of previous tasks, following the iCaRL recipe but with every logit computed hyperbolically. On EgoObjects this yields 41.76% instance-level accuracy from scratch versus 20.05% for iCaRL, 45.91% versus 21.39% at class level, and 48.04% versus 22.24% at superclass level, with the lowest common-ancestor distance (4.93) and lowest forgetting (4.17%); with a pretrained backbone the instance accuracy is 84.81% versus 81.63%, and the paper reports $p < 0.05$ against iCaRL and DER over five seeds. The paper's central claim is that aligning continual classification and distillation with a precomputed hyperbolic hierarchy produces this joint multi-granularity performance, and that even the mistakes are hierarchically mild.

Load-bearing premise

The method assumes that Euclidean nearest-exemplar-mean classification at test time is already aligned with the hyperbolic prototypes used during training, so the hyperbolic geometry never needs to be computed at inference — an alignment the paper asserts in Section 4.2 but does not quantitatively verify.

Editorial extensions

If this is right

  • If a single model can classify at instance and class level at once, systems like robotics and self-driving perception gain a graceful-degradation property: a wrong instance prediction can still land in the correct class, which is precisely what the lower lowest-common-ancestor distance — the average edge distance to the deepest shared ancestor of prediction and ground truth — quantifies.
  • The continual hierarchical metrics proposed here — instance-, class-, and superclass-level accuracy plus LCA distance — give the field a standard way to report not only how often a model is right but how far wrong it is when it errs.
  • Because the hierarchy is embedded once before training, HyperCLIC needs no hierarchical information at test time and adds no extra inference computation; the paper shows the gains hold across backbones from ResNet34 to ViT-B/16.
  • HyperCLIC's stability across 15- and 37-task streams, with 4.17% forgetting from scratch, suggests that prototype-anchored hyperbolic distillation slows the erosion of fine-grained instance knowledge that class-level continual learners typically sacrifice.

Reading between the lines

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

  • The paper compares against baselines that were given no hierarchy at all; a test the authors leave implicit is whether iCaRL or DER closes the gap once they receive the same WordNet tree in Euclidean form, which would separate the contribution of hyperbolic geometry from the mere use of hierarchical information.
  • Because the hierarchy is built automatically by matching class names to WordNet synsets, the two-stage scheme transfers to any dataset with a taxonomy; a natural extension is to test it on deeper ImageNet-style trees, where the class- and superclass-level metrics have more levels to climb and hierarchy errors can be scored more finely.
  • The paper positions its work as complementary to hierarchical task-ordering methods; combining a coarse-to-fine task schedule with hyperbolic prototypes could plausibly compound the gains, since ordering and geometry would then push representations toward the same tree.
  • If the Euclidean-alignment assumption used at test time were verified explicitly — say, by reporting the correlation between Euclidean feature distances and hyperbolic prototype distances — the method would become easier to adopt, since the exponential map could be dropped at inference with confidence.
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 / 4 minor

Summary. The paper defines a new continual learning task in which a model must recognize both object instances and object classes across a sequence of tasks. It proposes HyperCLIC, which first embeds a WordNet-derived class-instance hierarchy as fixed prototypes in the Poincaré ball, then trains a feature extractor with a hyperbolic prototype-based classification loss and a hyperbolic distillation loss. Evaluation on EgoObjects with 15, 6, and 37 tasks reports large gains over iCaRL, DER, and other baselines on instance-, class-, and superclass-level accuracy, LCA distance, and forgetting. The appendix provides 5-seed averages and significance tests against iCaRL and DER.

Significance. If the causal role of hyperbolic geometry were established, this would be a meaningful contribution: a new joint instance-and-class continual learning formulation, a practical method that works across multiple granularities, and evaluation metrics for hierarchical mistake severity. The paper ships code and metadata, runs multi-seed experiments with p-values, and provides thorough ablations over temperature, loss weighting, distillation losses, and backbone architectures. The main weakness is that the experiments do not currently isolate the hyperbolic mechanism from the hierarchical prototype prior, so the central mechanistic claim is supported only partially by the empirical results.

major comments (4)
  1. [Sections 4.1-4.2, Tables 1-2] The comparison does not isolate the hyperbolic mechanism. HyperCLIC differs from the baselines in at least two bundled ways: it uses fixed prototypes for every node in the full EgoObjects hierarchy, including instance classes that will only appear in future tasks (Eq. (7) computes logits against all |C| prototypes in every task), and it places those prototypes in hyperbolic space. A Euclidean control that uses the same hierarchical prototypes, the same softmax over all prototypes, and the same distillation loss is needed to determine whether the from-scratch gain of 41.76% versus 20.05% for iCaRL comes from hyperbolic geometry or from the hierarchical prototype prior alone. The statement in Section 5 that 'there is no non-trivial way to integrate hierarchical structure into their existing frameworks' is an argument, not a substitute for such a control.
  2. [Section 4.2, inference paragraph] Test-time predictions use Euclidean nearest-exemplar-mean classification, not the hyperbolic logits used in training. The paper asserts that 'the Euclidean representations are already aligned with the hierarchy' but provides no quantitative evidence for this alignment. Since the deployed decision rule is Euclidean, the reported gains cannot be attributed to the hyperbolic training objective unless this transfer is measured, for example by comparing hyperbolic-versus-Euclidean logit accuracy on the test set or by measuring the correlation between Euclidean distances and graph distances in the hierarchy.
  3. [Section 3 and Algorithm 2] The method uses the complete set of future instance labels during every task: the hyperbolic prototypes for all instance nodes are precomputed from the full hierarchy in Section 4.1, and Algorithm 2 computes classification logits against all prototypes, including classes not yet observed. This is a protocol advantage over the class-incremental baselines, which do not have access to future class labels or their geometric layout. Please clarify whether this is intended as a permitted prior, and ideally include an ablation in which prototypes are added only when a class first appears, to show that the gains persist without future-class information.
  4. [Appendix A.3, Tables 8-9] The main text (Table 1) reports a from-scratch instance accuracy of 41.76% for HyperCLIC, while Table 9 reports a 5-seed average of 38.55±2.37 (and 21.49±0.91 for iCaRL). The paper does not state whether Table 1 is a single seed or the best seed, nor does it explain how the p-values in Table 8 relate to these numbers. Please report mean±std in the main table or otherwise reconcile the headline number with the averaged result.
minor comments (4)
  1. [Section 4.1, Eq. (3)] The formula for the Poincaré loss L_Poincaré is missing from the main text; it appears only in Algorithm 1. Please insert the equation and its definition of the negative set N(u) at the appropriate place.
  2. [Equation (8)] The notation in Eq. (8) uses y_i both as the target for sample i and as the summation index over classes; please use separate symbols for the sample index and the class index to avoid confusion.
  3. [Section 5, Hierarchies paragraph] There is a typo in 'benhcmark'; additionally, the description of hierarchy construction could state more explicitly how the 91 classes without instances were removed and how the resulting tree depth of 12 relates to the 15-task split.
  4. [Section 5, Qualitative examples] The sentence 'HyperCLIC makes more instance-level mistakes' appears to contradict the higher instance accuracy reported elsewhere; please rephrase to clarify that this refers to the specific examples shown rather than the overall accuracy.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: HyperCLIC's central instance-level results are external benchmark outcomes, and the hierarchical-metric concern is a design confound rather than a circular derivation.

full rationale

The paper's derivation chain is not circular. Stage 1 constructs fixed hyperbolic prototypes by minimizing Poincaré, entailment, and separation losses over the WordNet-derived tree T (Eqs. 3-5, Algorithm 1); Stage 2 trains a feature extractor with hyperbolic classification and distillation losses (Eqs. 6-10, Algorithm 2); and the final test-time decision rule is Euclidean nearest-exemplar-mean classification, which does not even use the hyperbolic prototypes. The headline empirical claims are instance-level accuracies (41.76% vs 20.05% from scratch; 84.81% vs 81.63% pretrained) against non-hierarchical baselines iCaRL and DER on EgoObjects—these are external benchmark numbers, not algebraic consequences of the method's inputs. The reader's concern that class-, superclass-, and LCA metrics are computed from the same WordNet hierarchy used to construct the prototypes is a legitimate experimental-design confound: any hierarchy-aware method would be expected to score higher on those metrics, so they do not independently prove that hyperbolic geometry, rather than the hierarchical prototype prior, causes the gains. However, confounding is not circularity: the reported metric values are empirical outcomes of trained models, not identities forced by the equations. Also, Section 4.2 asserts without quantitative evidence that 'the Euclidean representations are already aligned with the hierarchy,' which weakens the causal attribution to hyperbolic geometry, but this is an unsupported shortcut rather than a circular step. Self-citations (e.g., Ghadimi Atigh et al. 2021 for sibling/cousin accuracy, Long et al. 2020 for separation loss) are not load-bearing because the relevant equations are given explicitly in the paper. No fitted parameter is renamed as a prediction, no uniqueness result is imported from the authors, and no ansatz is smuggled in solely via self-citation.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The method introduces no new physical or conceptual entities beyond the standard hyperbolic prototype machinery. It does rely on several hand-set hyperparameters, a constructed WordNet hierarchy, and a transfer assumption from hyperbolic training to Euclidean inference. The central contribution is empirical and inherits all of these choices.

free parameters (4)
  • Temperature tau = 0.1
    Controls softmax peakedness over hyperbolic logits. Table 5 shows strong sensitivity, and 0.1 was selected because it gave the best accuracy in both from-scratch and pretrained settings.
  • Loss balancing weight lambda = 0.5
    Balances L_distil and L_cls. Table 3 shows that 0.5 yields the best instance-, class-, and superclass-level accuracy, while other values favor stability or plasticity less effectively.
  • Poincare curvature c = 1
    The curvature of the Poincare ball is fixed to 1 for all experiments. It affects every hyperbolic distance used in the classification and distillation losses, but it is taken from the library default rather than fit to data.
  • Fine-tuned layers = all layers for HyperCLIC; classifier-only for iCaRL in pretrained comparisons
    Appendix A.2 shows the number of fine-tuned layers was chosen per method to give each method its best pretrained score. This protocol choice can alter relative results.
assumptions (5)
  • domain assumption The joint label structure of instances, classes, superclasses, and other nodes is a tree with each instance a leaf having exactly one parent class.
    Section 3 formalizes the hierarchy as T = (V, E) with directed edges and one parent per node; if EgoObjects labels do not form a clean tree, the prototypes and the hierarchical metrics built on it are distorted.
  • domain assumption The WordNet-derived hierarchy for EgoObjects is an accurate ground-truth hierarchy, and manual insertions plus pruning of single-child nodes do not introduce harmful noise.
    Section 5 describes automatic WordNet matching, manual additions, and pruning of 91 classes and single-child nodes. The method and all hierarchical metrics inherit this constructed hierarchy.
  • standard math Poincare ball embeddings with entailment cones can embed the object tree with low distortion and preserve parent-child partial order.
    Section 4.1 relies on Nickel and Kiela (2017), Ganea et al. (2018a), and a cited low-distortion result from Sarkar (2011), without re-deriving the embedding guarantees for this specific hierarchy.
  • domain assumption Backbone features lie in the tangent space at the origin, so the exponential map produces hyperbolic representations comparable to the precomputed prototypes.
    Section 4.2, Eq. 6: Euclidean backbone output is exp-mapped into the Poincare ball. If the features are not in the tangent space or are not scale-compatible, distances to prototypes are not meaningful classifiers.
  • ad hoc to paper Euclidean nearest-exemplar-mean inference is consistent with the hyperbolic training objective because Euclidean representations are already aligned with the hierarchy.
    Section 4.2, final paragraph: the paper asserts this alignment without quantitative evidence, while test-time predictions use Euclidean nearest means rather than the hyperbolic logits used during training.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Continual Hyperbolic Learning of Instances and Classes." pith.science (2026). https://pith.science/paper/4W77OL6B

@misc{pith2026250610710,
  author       = {Pith},
  title        = {Pith review of: Continual Hyperbolic Learning of Instances and Classes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4W77OL6B}},
  note         = {Machine review of arXiv:2506.10710}
}
read the original abstract

Continual learning has traditionally focused on classifying either instances or classes, but real-world applications, such as robotics and self-driving cars, require models to handle both simultaneously. To mirror real-life scenarios, we introduce the task of continual learning of instances and classes, at the same time. This task challenges models to adapt to multiple levels of granularity over time, which requires balancing fine-grained instance recognition with coarse-grained class generalization. In this paper, we identify that classes and instances naturally form a hierarchical structure. To model these hierarchical relationships, we propose HyperCLIC, a continual learning algorithm that leverages hyperbolic space, which is uniquely suited for hierarchical data due to its ability to represent tree-like structures with low distortion and compact embeddings. Our framework incorporates hyperbolic classification and distillation objectives, enabling the continual embedding of hierarchical relations. To evaluate performance across multiple granularities, we introduce continual hierarchical metrics. We validate our approach on EgoObjects, the only dataset that captures the complexity of hierarchical object recognition in dynamic real-world environments. Empirical results show that HyperCLIC operates effectively at multiple granularities with improved hierarchical generalization.

Figures

Figures reproduced from arXiv: 2506.10710 by the authors.

Figure 1
Figure 1. Recognizing instances and classes simultane￾ously is important in many real-world applications. By adding instances as an additional layer to the object hierarchy and learning representations that capture the joint hierarchy, we can classify samples at multiple levels of granularity. Regardless of continual scenarios, the main chal￾lenge of continual learning is to alleviate catastrophic forgetting with only limited… view at source ↗
Figure 2
Figure 2. Overview of HyperCLIC . The class-instance hierarchy is projected into a shared hyperbolic space. When learning an instance, its samples go through the feature extractor and are mapped into the shared hyperbolic space using the exponential map. These samples are then pushed toward their hyperbolic instance prototype via classification loss and are encouraged to maintain hyperbolic logits from previous classes throug… view at source ↗
Figure 3
Figure 3. Left: the pair-wise distances of the instance-level hyperbolic prototypes. Middle & right: the class-level predictions only for the wrong instance-level pre￾dictions. The squares highlight the hierarchical structure of HyperCLIC mistakes compared with iCaRL and its similarity to the original hierarchy. Qualitative examples: success and failure cases [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative examples from EgoObjects comparing instance and class-level predictions for ground truth (1st row), iCaRL (2nd row), and our method (3rd row). Green indicates correct predictions, while red and orange denote severe and less severe hierarchical errors, respe…
Figure 5
Figure 5. Figure 5: Performance of HyperCLIC and iCaRL across four metrics with varying numbers of fine-tuned layers. HyperCLIC excels when all layers are fine-tuned, while iCaRL performs better with more fixed layers. The [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]

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. After the Euclidean Highway: Hyperbolic Expert AI as the Next Innovation

    cs.LG 2026-07 conditional novelty 6.0 of 10

    Hyperbolic losses at the loss layer alone train stably at 18M-sample scale, while the same geometry inside a trainable adapter is claimed to collapse training.

Reference graph

Works this paper leans on

15 extracted references · 5 canonical work pages · cited by 1 Pith paper

  1. [6]

    Skip-gram word embeddings in hyperbolic space.arXiv preprint arXiv:1809.01498,

    Matthias Leimeister and Benjamin J Wilson. Skip-gram word embeddings in hyperbolic space.arXiv preprint arXiv:1809.01498,

  2. [7]

    Disentangling the causes of plasticity loss in neural networks.arXiv preprint arXiv:2402.18762,

    Clare Lyle, Zeyu Zheng, Khimya Khetarpal, Hado van Hasselt, Razvan Pascanu, James Martens, and Will Dabney. Disentangling the causes of plasticity loss in neural networks.arXiv preprint arXiv:2402.18762,

  3. [9]

    clvision-challenge-2022 reports.https://github.com/ContinualAI/ clvision-challenge-2022/tree/main/reports,

    Lorenzo Pellegrini. clvision-challenge-2022 reports.https://github.com/ContinualAI/ clvision-challenge-2022/tree/main/reports,

  4. [11]

    Hyperbolic neural networks++.arXiv preprint arXiv:2006.08210,

    Ryohei Shimizu, Yusuke Mukuta, and Tatsuya Harada. Hyperbolic neural networks++.arXiv preprint arXiv:2006.08210,

  5. [12]

    Poincar\’e glove: Hyperbolic word embeddings

    Alexandru Tifrea, Gary B ´ecigneul, and Octavian-Eugen Ganea. Poincar\’e glove: Hyperbolic word embeddings. arXiv preprint arXiv:1810.06546,

  6. [14]

    Hyperbolic contrastive learning.arXiv preprint arXiv:2302.01409,

    Yun Yue, Fangzhou Lin, Kazunori D Yamada, and Ziming Zhang. Hyperbolic contrastive learning.arXiv preprint arXiv:2302.01409,

  7. [15]

    Forward compatible few-shot class-incremental learning

    Da-Wei Zhou, Fu-Yun Wang, Han-Jia Ye, Liang Ma, Shiliang Pu, and De-Chuan Zhan. Forward compatible few-shot class-incremental learning. InCVPR, 2022a. Da-Wei Zhou, Qi-Wei Wang, Han-Jia Ye, and De-Chuan Zhan. A model or 603 exemplars: Towards memory-efficient class-incremental learning.arXiv preprint arXiv:2205.13218, 2022b. Da-Wei Zhou, Qi-Wei Wang, Zhi-H...

  8. [2017]

    Geoopt: Riemannian optimization in pytorch.arXiv preprint arXiv:2005.02819,

    Max Kochurov, Rasul Karimov, and Serge Kozlukov. Geoopt: Riemannian optimization in pytorch.arXiv preprint arXiv:2005.02819,

Show all 15 references
  1. [2018]

    Leveraging hierarchical taxonomies in prompt-based continual learning.arXiv preprint arXiv:2410.04327,

    Quyen Tran, Hoang Phan, Minh Le, Tuan Truong, Dinh Phung, Linh Ngo, Thien Nguyen, Nhat Ho, and Trung Le. Leveraging hierarchical taxonomies in prompt-based continual learning.arXiv preprint arXiv:2410.04327,

  2. [2019]

    Riemannian walk for incremental learning: Understanding forgetting and intransigence

    Arslan Chaudhry, Puneet K Dokania, Thalaiyasingam Ajanthan, and Philip HS Torr. Riemannian walk for incremental learning: Understanding forgetting and intransigence. InECCV, 2018a. Arslan Chaudhry, Marc’Aurelio Ranzato, Marcus Rohrbach, and Mohamed Elhoseiny. Efficient lifelon...

  3. [2020]

    Embedding text in hyperbolic spaces.arXiv preprint arXiv:1806.04313,

    Bhuwan Dhingra, Christopher J Shallue, Mohammad Norouzi, Andrew M Dai, and George E Dahl. Embedding text in hyperbolic spaces.arXiv preprint arXiv:1806.04313,

  4. [2021]

    Apo-vae: Text generation in hyperbolic space.arXiv preprint arXiv:2005.00054,

    Shuyang Dai, Zhe Gan, Yu Cheng, Chenyang Tao, Lawrence Carin, and Jingjing Liu. Apo-vae: Text generation in hyperbolic space.arXiv preprint arXiv:2005.00054,

  5. [2022]

    3rd continual learning workshop challenge on egocentric category and instance level object understanding.arXiv preprint arXiv:2212.06833,

    Lorenzo Pellegrini, Chenchen Zhu, Fanyi Xiao, Zhicheng Yan, Antonio Carta, Matthias De Lange, Vincenzo Lomonaco, Roshan Sumbaly, Pau Rodriguez, and David Vazquez. 3rd continual learning workshop challenge on egocentric category and instance level object understanding.arXiv pre...

  6. [2024]

    An empirical analysis of forgetting in pre-trained models with incremental low-rank updates.arXiv preprint arXiv:2405.18069,

    Simone Magistri, Joost van de Weijer, Andew D Bagdanov, et al. An empirical analysis of forgetting in pre-trained models with incremental low-rank updates.arXiv preprint arXiv:2405.18069,

  7. [2025]

    Distilling the knowledge in a neural network.arXiv preprint arXiv:1503.02531,

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network.arXiv preprint arXiv:1503.02531,

Pith tools

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