Pith. sign in

REVIEW 3 major objections 5 minor 41 references

Mixed-Precision Graph Neural Quantization for Low Bit Large Language Models

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a graph neural network can assign quantization bit-widths per weight column and beat GPTQ at 2-bit post-training quantization of LLaMA models.

desk verdict Novel GNN-based bit-allocation idea, but Algorithm 1's compensation step is a no-op and the reported gains aren't explained by the written method. read the letter →

arxiv 2501.18154 v1 pith:Y2EUDACT submitted 2025-01-30 cs.CL

Wanlong Liu , Yichen Xiao , Dingyi Zeng , Hongyang Zhao , Wenyu Chen , Malu Zhang This is my paper · ORCID
classification cs.CL
keywords Post-TrainingQuantizationGraphNeuralNetworksMixed-PrecisionLargeLanguageModelsLow-bitLLaMAHessian-based
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

Post-training quantization of large language models degrades sharply below 3 bits per weight. This paper claims that a graph neural network can decide, column by column, how many bits each weight column deserves, and that this mixed-precision allocation beats the standard GPTQ method at the same average bit budget. The supporting evidence is Table I: at 2.0 bits, MG-PTQ reports lower perplexity than GPTQ on every tested LLaMA model on both WikiText2 and C4. The practical stake is that 2-bit deployment of models in the 7B-to-13B range may become usable without retraining. The paper also claims that the graph component itself, not just the bit-allocation classifier, drives the improvement, based on an ablation that replaces the GCN with an MLP.

What carries the argument

The load-bearing object is the feature graph built from the Cholesky factor $H_c = \mathrm{Cholesky}((2X_F^T X_F+\lambda I)^{-1})$, where $X_F$ is calibration data. Each column of the target weight matrix is a node; $H_c$ plays two roles at once: as the weighted adjacency matrix in the GCN propagation $X_G^{(l+1)}=\sigma(H_c X_G^{(l)} W^{(l)})$ and as the divisor and multiplier in the blockwise compensation update $E:=(W-B)/H_c$, then $W:=W-E\cdot H_c$ (Algorithm 1). The GCN's message passing is meant to let each column's bit assignment reflect dependencies on other columns, and the Gumbel-Softmax approximation supplies gradients through the discrete bit-width choice. The training loss combines quantization error with a mean-squared-error constraint on average bit-width, so the allocator can be tuned to a target bit budget.

What would settle it

Run the same GNN allocator on LLaMA-7b at 2.0 bits but replace the Algorithm 1 update with GPTQ's standard Hessian-inverse compensation; if WikiText2 perplexity jumps back to GPTQ's roughly 152 rather than staying near MG-PTQ's 130, the claimed gain comes from the unusual update, not from the GNN allocation.

Watch

Extended reading notes

Core claim

The central claim is that weight-column importance for quantization can be learned by message passing over a graph whose adjacency matrix is the Cholesky factor of the inverse second-order Hessian, and that the resulting per-column bit assignment yields lower perplexity than GPTQ's uniform low-bit quantization. Concretely, MG-PTQ forms one node per column of the target weight matrix, uses mean-pooled column values as node features, propagates them through a two-layer GCN, and maps the output to bit-widths per column. The same Cholesky factor is then used in the blockwise error-compensation update inherited from the GPTQ-style pipeline. Table I reports the headline numbers: at 2.0 bits, MG-PTQ achieves perplexity of 130.27 on LLaMA1-7b on WikiText2 versus GPTQ's 152.31, with similar improvements across LLaMA1-13b, LLaMA2-7b, LLaMA2-13b and LLaMA3-8B on both WikiText2 and C4. The authors interpret these results as showing that GNN-based importance perception captures dependencies among weights that uniform or heuristic schemes miss.

Load-bearing premise

The result stands on the unproven claim that the same Cholesky factor can both carry information in the graph network and serve as the divisor and multiplier in the blockwise error-correction update; if that correction is mathematically unsound, the reported perplexity gains are not explained by the stated algorithm.

Editorial extensions

If this is right

  • At 2.0 bits, MG-PTQ reports lower perplexity than GPTQ on every LLaMA variant tested, on both WikiText2 and C4, so the claimed advantage is consistent across models rather than a single favorable case.
  • Because average bit-width is controlled by an MSE penalty, the same trained allocator can be re-targeted to different budgets such as 1.6, 1.8, 2.0, and 2.5 bits without re-architecting the method.
  • The ablation with an MLP in place of the GCN shows a significant perplexity drop, which supports the paper's claim that graph propagation itself, not merely per-column classification, drives the improvement.
  • The efficiency analysis reports quantization time close to GPTQ's, so the method is presented as practical for deployment-scale models.

Reading between the lines

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

  • The same learned importance scores could be reused for pruning or activation quantization, since the GCN is effectively learning which weight columns are costly to distort; this is a natural follow-up the paper does not test.
  • A head-to-head against salience-driven mixed-precision heuristics rather than uniform GPTQ would separate what the learned allocator adds over a handcrafted importance rule such as activation magnitudes.
  • The unproven compensation update is the place to look first: a direct check of layer-wise reconstruction error on a single LLaMA layer would settle whether the gains come from the allocator or from an undocumented error-correcting effect of the Cholesky factor.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes MG-PTQ, a post-training quantization method that uses a graph neural network (GNN) to assign per-column mixed-precision bit-widths to the weights of large language models. The GNN takes a Cholesky-decomposed Hessian as the adjacency matrix and weight-derived features as node features, and is trained to minimize blockwise quantization error subject to an average-bit-width penalty. Quantization is performed blockwise following a GPTQ-like procedure. Experiments on five LLaMA models at 1.6-2.5 bits on WikiText2 and C4 report perplexity improvements over GPTQ at 2 bits.

Significance. The central idea - learning a mixed-precision bit allocation with a GNN using second-order information - is timely, and the experimental coverage (five models, two datasets) is reasonably broad. The paper also introduces a concrete mechanism (Gumbel-Softmax approximate gradients) for training through discrete bit-width choices. However, the load-bearing algorithmic claim is not supported by the written procedure: the compensation step in Algorithm 1 cancels algebraically, and the ablation intended to isolate the GCN's contribution changes the input features along with the architecture. If the algorithm description were corrected and the ablations made controlled, the approach could be a useful contribution; as it stands, the reported gains over GPTQ are not explained by the stated method.

major comments (3)
  1. [III-B, Algorithm 1 lines 15-16] The 'Block-wise Output Compensation' step is algebraically a no-op. For any standard meaning of '/' (right matrix division, elementwise division, or left pseudo-inverse), E·Hc_{b:b+β,b:b+β} = W_{:,b:b+β} − B_{:,b:b+β}, so the update W_{:,b:b+β} := W_{:,b:b+β} − E·Hc_{b:b+β,b:b+β} sets the block equal to B_{:,b:b+β}, which was already fixed on line 14. Because columns beyond the current block are never touched, no quantization error is propagated to the remaining weights as in GPTQ's update. Consequently, Algorithm 1 as written reduces to per-column mixed-precision t-bit rounding without any error compensation, and the perplexity improvements over GPTQ reported in Table I cannot be explained by the stated algorithm. The authors must either provide a corrected compensation step (e.g., updating the remaining columns using the off-diagonal blocks of the inverse Hessian, as in GPTQ) together with a derivation, or revise the description and claims to reflect the actual procedure used.
  2. [IV-C-1, Fig. 2(a)] The ablation comparing GCN with MLP is not a controlled comparison. The MLP receives as input the 'second-order Hessian matrix' while the GCN receives weight-derived features (X_G^(0) from Eq. 3) as node features and uses the Hessian as adjacency. Thus the performance difference could be due to the different input features rather than the graph architecture. To support the conclusion that 'the GCN module is crucial', the MLP should be given the same node features as the GCN, or the GCN should be ablated by removing the message-passing step while keeping the input features identical.
  3. [Table I] The baseline set is insufficient to support the claim of state-of-the-art performance. Only RTN, AWQ, and GPTQ are compared; recent low-bit PTQ methods such as BiLLM, PB-LLM, SpQR, and Slim-LLM are not included, and there are no GPTQ results at 1.6 or 1.8 bits to match MG-PTQ's bit levels. The abstract's claim that MG-PTQ 'outperforms previous state-of-the-art PTQ method GPTQ' may be true at 2 bits, but the broader claim of setting new benchmarks requires comparison with current SOTA methods at the same bit widths.
minor comments (5)
  1. [Algorithm 1] The output of Algorithm 1 is described as 'binarized weights' but B contains mixed-precision values; this should be clarified.
  2. [III-B, Eq. (2) and (4)] The Cholesky factor Hc is lower-triangular and non-symmetric; using it directly as a GCN adjacency matrix is unconventional and the paper does not discuss how directed message passing is handled. The authors should justify this design choice or use a symmetrized version.
  3. [Table I] The table formatting is confusing because the GPTQ rows and MG-PTQ rows are not visually separated, and the bit labels 2.5, 2, 1.8, 1.6 appear to belong to GPTQ; also, GPTQ results at 1.6 and 1.8 bits are missing, which limits comparability.
  4. [IV-A-3] Reproduction details are incomplete: the Gumbel-Softmax temperature schedule is not specified, the value of the Hessian regularizer λ in Eq. (2) is not reported, and the actual average bit-width achieved after training is not given.
  5. [References] The reference list contains duplicates: [34] and [40] are the same GCN paper, and AWQ is listed as both [27] and [41] with different bibliographic details.

Circularity Check

1 steps flagged · score 4.0 of 10

Algorithm 1's Block-wise Output Compensation cancels by construction (E := (W−B)/Hc; W := W−E·Hc), so the stated method reduces to per-column quantization; the external perplexity comparison remains independent, giving partial circularity.

  1. self definitional [Algorithm 1, lines 14–16; Section III-B3 'Quantization']
    "B:,b:b+β := B1 + B2 + ... + Btmax E := (W:,b:b+β − B:,b:b+β)/Hc W:,b:b+β := W:,b:b+β −E·Hc // Block-wise Output Compensation (OBC)"

    Line 15 defines E so that E·Hc_block = W_block − B_block under the elementwise or right-division reading implied by the pseudocode. Line 16 therefore sets W_block := B_block, which was already fixed on line 14, and later blocks are never updated. The 'Block-wise Output Compensation' is zero by construction, so the algorithm as written reduces to per-column mixed-precision t-bit rounding with no GPTQ-style Hessian-inverse error propagation to remaining columns. Any reported low-bit improvement cannot be attributed to the stated compensation mechanism.

full rationale

The only step that reduces to its own input by construction is Algorithm 1 lines 15–16: E is defined as (W−B)/Hc, making the compensation update W := W − E·Hc identically equal to W := B. Since B is already fixed for the block and no other columns are modified, the claimed OBC has no effect; the stated method degenerates to per-column rounding plus bit allocation. This is a definitional no-op rather than a fitted prediction, but it is load-bearing because the paper attributes its low-bit gains to the GPTQ-style blockwise scheme. The headline perplexity numbers are measured on held-out WikiText2/C4 text and are not themselves constructed from the method's equations, so the circularity is partial. The GNN is trained on the inference model's own weights, but PTQ is permitted to use target weights and the final metric is external; that raises overfitting/generalization concerns, not a definitional equivalence. No load-bearing self-citation chain appears in the references; [16] is an external GPTQ citation, and the authors' own GNN works are background only.

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

The method's central assumption is that a Cholesky factor of the inverse Hessian can act as both graph adjacency and compensation matrix; this is unproven and mathematically unusual. Key hyperparameters (alpha, lambda, block size, hidden dimension, max bit classes) are hand-chosen, and the ablation's input mismatch invalidates the MLP control. No new physical or conceptual entities are introduced.

free parameters (5)
  • alpha (bit-width penalty weight) = 1
    Balances quantization error and average bit-width loss in Eq. (6); chosen by hand without sensitivity analysis.
  • Hessian regularizer lambda = not specified
    Added in Eq. (2) to stabilize the inverse Hessian; the paper gives no value, making the exact adjacency matrix undefined.
  • GCN hidden/input dimension dgnn = 512
    Sets the node feature dimensionality and the padding target in Eq. (3); chosen by hand.
  • Block size beta = 128
    Controls the granularity of the blockwise compensation in Algorithm 1; set by hand following GPTQ.
  • Max bit-width classes = 4
    Number of classes in the bit-width classifier in Eq. (5); chosen by hand and may limit the allocator's flexibility.
assumptions (5)
  • domain assumption Calibration data X_F approximates the true activation distribution, so H = 2 X_F^T X_F + lambda I captures the loss curvature.
    Used in Eq. (2) to compute the Hessian; standard in GPTQ-style PTQ but not verified for the models tested.
  • ad hoc to paper The Cholesky factor Hc can serve as both graph adjacency matrix and error-compensation matrix.
    Invoked in Eq. (4) and Algorithm 1 lines 15-16 without derivation; a non-symmetric triangular matrix is not a conventional adjacency matrix, and the compensation update is not shown to minimize error.
  • standard math Gumbel-Softmax provides a valid differentiable approximation to the argmax bit selection.
    Section III.C; standard technique for discrete optimization, but the temperature parameter is not specified.
  • domain assumption Block-wise independence of weight columns, inherited from GPTQ.
    Algorithm 1 processes blocks independently; the paper cites GPTQ for this design without re-deriving its validity.
  • ad hoc to paper The MLP ablation with Hessian input is an equivalent control for the GCN.
    Section IV.C.1; the MLP receives different input features than the GCN (Hessian matrix vs. mean-pooled weights), so the comparison does not isolate the effect of the graph architecture.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mixed-Precision Graph Neural Quantization for Low Bit Large Language Models." pith.science (2026). https://pith.science/paper/Y2EUDACT

@misc{pith2026250118154,
  author       = {Pith},
  title        = {Pith review of: Mixed-Precision Graph Neural Quantization for Low Bit Large Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y2EUDACT}},
  note         = {Machine review of arXiv:2501.18154}
}
read the original abstract

Post-Training Quantization (PTQ) is pivotal for deploying large language models (LLMs) within resource-limited settings by significantly reducing resource demands. However, existing PTQ strategies underperform at low bit levels < 3 bits due to the significant difference between the quantized and original weights. To enhance the quantization performance at low bit widths, we introduce a Mixed-precision Graph Neural PTQ (MG-PTQ) approach, employing a graph neural network (GNN) module to capture dependencies among weights and adaptively assign quantization bit-widths. Through the information propagation of the GNN module, our method more effectively captures dependencies among target weights, leading to a more accurate assessment of weight importance and optimized allocation of quantization strategies. Extensive experiments on the WikiText2 and C4 datasets demonstrate that our MG-PTQ method outperforms previous state-of-the-art PTQ method GPTQ, setting new benchmarks for quantization performance under low-bit conditions.

Figures

Figures reproduced from arXiv: 2501.18154 by the authors.

Figure 1
Figure 1. The overall architecture of MG-PTQ model. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Further experimental analysis. Sub-figure (a) presents [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 22 canonical work pages

  1. [1]

    Gpt-4 technical report,

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat et al. , “Gpt-4 technical report,” arXiv preprint arXiv:2303.08774 , 2023

    arXiv 2023

  2. [2]

    Open and ef- ficient foundation language models,

    H. Touvron, T. Lavril, G. Izacard, X. Martinet, M. Lachaux, T. Lacroix, B. Rozi `ere, N. Goyal, E. Hambro, F. Azhar et al. , “Open and ef- ficient foundation language models,” Preprint at arXiv. https://doi. org/10.48550/arXiv, vol. 2302, 2023

    doi:10.48550/arxiv

  3. [3]

    The application of large language models in recommendation systems,

    P. Yu, Z. Xu, J. Wang, and X. Xu, “The application of large language models in recommendation systems,” arXiv preprint arXiv:2501.02178 , 2025

    arXiv 2025

  4. [4]

    Mt-bench-101: A fine-grained benchmark for evaluating large language models in multi-turn dialogues,

    G. Bai, J. Liu, X. Bu, Y . He, J. Liu, Z. Zhou, Z. Lin, W. Su, T. Ge, B. Zheng et al. , “Mt-bench-101: A fine-grained benchmark for evaluating large language models in multi-turn dialogues,”arXiv preprint arXiv:2402.14762, 2024

    arXiv 2024

  5. [5]

    Graphreader: Building graph-based agent to en- hance long-context abilities of large language models,

    S. Li, Y . He, H. Guo, X. Bu, G. Bai, J. Liu, J. Liu, X. Qu, Y . Li, W. Ouyang et al. , “Graphreader: Building graph-based agent to en- hance long-context abilities of large language models,” arXiv preprint arXiv:2406.14550, 2024

    arXiv 2024

  6. [6]

    A hybrid attention framework for fake news detection with large language models,

    X. Xu, P. Yu, Z. Xu, and J. Wang, “A hybrid attention framework for fake news detection with large language models,” arXiv preprint arXiv:2501.11967, 2025

    arXiv 2025

  7. [7]

    D. Wang, Information Science and Electronic Engineering: Proceedings of the 3rd International Conference of Electronic Engineering and Information Science (ICEEIS 2016), January 4-5, 2016, Harbin, China . CRC Press, 2016

    work page 2016

  8. [8]

    Measuring massive multitask language understanding,

    D. Hendrycks, C. Burns, S. Basart, A. Zou, M. Mazeika, D. Song, and J. Steinhardt, “Measuring massive multitask language understanding,” in 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . OpenReview.net, 2021

    work page 2021

Show all 41 references
  1. [9]

    Conceptmath: A bilingual concept-wise bench- mark for measuring mathematical reasoning of large language models,

    Y . Wu, J. Liu, X. Bu, J. Liu, Z. Zhou, Y . Zhang, C. Zhang, Z. Bai, H. Chen, T. Ge et al., “Conceptmath: A bilingual concept-wise bench- mark for measuring mathematical reasoning of large language models,” arXiv preprint arXiv:2402.14660 , 2024

    2024 arXiv

  2. [10]

    Optimization of transformer heart disease prediction model based on particle swarm optimization algorithm,

    P. Yu, J. Yi, T. Huang, Z. Xu, and X. Xu, “Optimization of transformer heart disease prediction model based on particle swarm optimization algorithm,” arXiv preprint arXiv:2412.02801 , 2024

    2024 arXiv

  3. [11]

    Enhancing user intent for recommendation systems via large language models,

    X. Xu, Z. Xu, P. Yu, and J. Wang, “Enhancing user intent for recommendation systems via large language models,” arXiv preprint arXiv:2501.10871, 2025

    2025 arXiv

  4. [12]

    How good are low-bit quantized llama3 models? an empirical study,

    W. Huang, X. Ma, H. Qin, X. Zheng, C. Lv, H. Chen, J. Luo, X. Qi, X. Liu, and M. Magno, “How good are low-bit quantized llama3 models? an empirical study,” CoRR, vol. abs/2404.14047, 2024

    2024 arXiv

  5. [13]

    Spqr: A sparse- quantized representation for near-lossless LLM weight compression,

    T. Dettmers, R. Svirschevski, V . Egiazarian, D. Kuznedelev, E. Frantar, S. Ashkboos, A. Borzunov, T. Hoefler, and D. Alistarh, “Spqr: A sparse- quantized representation for near-lossless LLM weight compression,” in The Twelfth International Conference on Learning Representati...

    2024

  6. [14]

    Degree-quant: Quantization-aware training for graph neural networks,

    S. A. Tailor, J. Fern ´andez-Marqu´es, and N. D. Lane, “Degree-quant: Quantization-aware training for graph neural networks,” in 9th Inter- national Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . OpenReview.net, 2021

    2021

  7. [15]

    Harnessing earnings reports for stock predictions: A qlora-enhanced llm approach,

    H. Ni, S. Meng, X. Chen, Z. Zhao, A. Chen, P. Li, S. Zhang, Q. Yin, Y . Wang, and Y . Chan, “Harnessing earnings reports for stock predictions: A qlora-enhanced llm approach,” arXiv preprint arXiv:2408.06634, 2024

    2024 arXiv

  8. [16]

    GPTQ: accu- rate post-training quantization for generative pre-trained transformers,

    E. Frantar, S. Ashkboos, T. Hoefler, and D. Alistarh, “GPTQ: accu- rate post-training quantization for generative pre-trained transformers,” CoRR, vol. abs/2210.17323, 2022

    2022 arXiv

  9. [17]

    Billm: Pushing the limit of post-training quantization for llms,

    W. Huang, Y . Liu, H. Qin, Y . Li, S. Zhang, X. Liu, M. Magno, and X. Qi, “Billm: Pushing the limit of post-training quantization for llms,” in Forty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024 . OpenReview.net, 2024

    2024

  10. [18]

    PB-LLM: partially binarized large language models,

    Z. Yuan, Y . Shang, and Z. Dong, “PB-LLM: partially binarized large language models,” in The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024 . Open- Review.net, 2024

    2024

  11. [19]

    Slim-llm: Salience-driven mixed-precision quantization for large language models,

    W. Huang, H. Qin, Y . Liu, Y . Li, X. Liu, L. Benini, M. Magno, and X. Qi, “Slim-llm: Salience-driven mixed-precision quantization for large language models,” CoRR, vol. abs/2405.14917, 2024

    2024 arXiv

  12. [20]

    Analysis of the cholesky decomposition of a semi- definite matrix,

    N. J. Higham, “Analysis of the cholesky decomposition of a semi- definite matrix,” 1990

    1990

  13. [21]

    Pointer sentinel mixture models,

    S. Merity, C. Xiong, J. Bradbury, and R. Socher, “Pointer sentinel mixture models,” in 5th International Conference on Learning Repre- sentations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings. OpenReview.net, 2017

    2017

  14. [22]

    Exploring the limits of transfer learning with a unified text-to-text transformer,

    C. Raffel, N. Shazeer, A. Roberts, K. Lee, S. Narang, M. Matena, Y . Zhou, W. Li, and P. J. Liu, “Exploring the limits of transfer learning with a unified text-to-text transformer,” Journal of machine learning research, vol. 21, no. 140, pp. 1–67, 2020

    2020

  15. [23]

    LLM-QAT: data-free quantization aware training for large language models,

    Z. Liu, B. Oguz, C. Zhao, E. Chang, P. Stock, Y . Mehdad, Y . Shi, R. Krishnamoorthi, and V . Chandra, “LLM-QAT: data-free quantization aware training for large language models,” in Findings of the Associa- tion for Computational Linguistics, ACL 2024, Bangkok, Thailand and vi...

    2024

  16. [24]

    Qlora: Efficient finetuning of quantized llms,

    T. Dettmers, A. Pagnoni, A. Holtzman, and L. Zettlemoyer, “Qlora: Efficient finetuning of quantized llms,” Advances in Neural Information Processing Systems, vol. 36, 2024

    2024

  17. [25]

    BRECQ: pushing the limit of post-training quantization by block reconstruction,

    Y . Li, R. Gong, X. Tan, Y . Yang, P. Hu, Q. Zhang, F. Yu, W. Wang, and S. Gu, “BRECQ: pushing the limit of post-training quantization by block reconstruction,” in 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . Open...

    2021

  18. [26]

    Zeroquant: Efficient and affordable post-training quantization for large- scale transformers,

    Z. Yao, R. Yazdani Aminabadi, M. Zhang, X. Wu, C. Li, and Y . He, “Zeroquant: Efficient and affordable post-training quantization for large- scale transformers,” Advances in Neural Information Processing Sys- tems, vol. 35, pp. 27 168–27 183, 2022

    2022

  19. [27]

    Awq: Activation-aware weight quanti- zation for on-device llm compression and acceleration,

    J. Lin, J. Tang, H. Tang, S. Yang, W.-M. Chen, W.-C. Wang, G. Xiao, X. Dang, C. Gan, and S. Han, “Awq: Activation-aware weight quanti- zation for on-device llm compression and acceleration,” Proceedings of Machine Learning and Systems , vol. 6, pp. 87–100, 2024

    2024

  20. [28]

    Supervised neural networks for the clas- sification of structures,

    A. Sperduti and A. Starita, “Supervised neural networks for the clas- sification of structures,” IEEE transactions on neural networks , vol. 8, no. 3, pp. 714–735, 1997

    1997

  21. [29]

    Substructure aware graph neural networks,

    D. Zeng, W. Liu, W. Chen, L. Zhou, M. Zhang, and H. Qu, “Substructure aware graph neural networks,” in Proceedings of the AAAI conference on artificial intelligence , vol. 37, no. 9, 2023, pp. 11 129–11 137

    2023

  22. [30]

    A new model for learning in graph domains,

    M. Gori, G. Monfardini, and F. Scarselli, “A new model for learning in graph domains,” in Proceedings. 2005 IEEE international joint conference on neural networks, 2005., vol. 2. IEEE, 2005, pp. 729–734

    2005

  23. [31]

    The graph neural network model,

    F. Scarselli, M. Gori, A. C. Tsoi, M. Hagenbuchner, and G. Monfar- dini, “The graph neural network model,” IEEE transactions on neural networks, vol. 20, no. 1, pp. 61–80, 2008

    2008

  24. [32]

    A simple graph neural network via layer sniffer,

    D. Zeng, L. Zhou, W. Liu, H. Qu, and W. Chen, “A simple graph neural network via layer sniffer,” in ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2022, pp. 5687–5691

    2022

  25. [33]

    Rethinking random walk in graph representation learning,

    D. Zeng, W. Chen, W. Liu, L. Zhou, and H. Qu, “Rethinking random walk in graph representation learning,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2023, pp. 1–5

    2023

  26. [35]

    Deep graph clustering via dual correlation reduction,

    Y . Liu, W. Tu, S. Zhou, X. Liu, L. Song, X. Yang, and E. Zhu, “Deep graph clustering via dual correlation reduction,” in Proceedings of the AAAI conference on artificial intelligence, vol. 36, no. 7, 2022, pp. 7603– 7611

    2022

  27. [36]

    A survey of deep graph clustering: Taxonomy, challenge, and application,

    Y . Liu, J. Xia, S. Zhou, S. Wang, X. Guo, X. Yang, K. Liang, W. Tu, S. Z. Li, and X. Liu, “A survey of deep graph clustering: Taxonomy, challenge, and application,” CoRR, vol. abs/2211.12875, 2022

    2022 arXiv

  28. [37]

    Simple contrastive graph clustering,

    Y . Liu, X. Yang, S. Zhou, and X. Liu, “Simple contrastive graph clustering,” arXiv preprint arXiv:2205.07865 , 2022

    2022 arXiv

  29. [38]

    Hard sample aware network for contrastive deep graph clustering,

    Y . Liu, X. Yang, S. Zhou, X. Liu, Z. Wang, K. Liang, W. Tu, L. Li, J. Duan, and C. Chen, “Hard sample aware network for contrastive deep graph clustering,” in Proceedings of the AAAI conference on artificial intelligence, vol. 37, no. 7, 2023, pp. 8914–8922

    2023

  30. [39]

    Attribute and structure preserving graph con- trastive learning,

    J. Chen and G. Kou, “Attribute and structure preserving graph con- trastive learning,” in Proceedings of the AAAI conference on artificial intelligence, vol. 37, no. 6, 2023, pp. 7024–7032

    2023

  31. [40]

    Semi-supervised classification with graph convolutional networks,

    T. N. Kipf and M. Welling, “Semi-supervised classification with graph convolutional networks,” in 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Con- ference Track Proceedings. OpenReview.net, 2017

    2017

  32. [41]

    Awq: activationaware weight quantization for llm compression and accel- eration. corr, abs/2306.00978, 2023. doi: 10.48550,

    J. Lin, J. Tang, H. Tang, S. Yang, X. Dang, and S. Han, “Awq: activationaware weight quantization for llm compression and accel- eration. corr, abs/2306.00978, 2023. doi: 10.48550,” arXiv preprint ARXIV .2306.00978

    2023 arXiv

  33. [42]

    Categorical reparameterization with gumbel-softmax,

    E. Jang, S. Gu, and B. Poole, “Categorical reparameterization with gumbel-softmax,” in 5th International Conference on Learning Repre- sentations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings. OpenReview.net, 2017

    2017

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.