Pith. sign in

REVIEW 4 major objections 6 minor 37 references

JPPO++: Joint Power and Denoising-inspired Prompt Optimization for Mobile LLM Services

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Jointly tuning prompt length and transmit power cuts mobile LLM latency by up to 46.5%.

desk verdict Relevant joint-optimization problem and a clear framework, but the evaluation does not back the central 'maintains quality' claim; the paper's own LongBench results show a real quality cost. read the letter →

arxiv 2412.03621 v4 pith:SQ4BHPHD submitted 2024-12-04 cs.NI

Feiran You , Hongyang Du , Kaibin Huang , Abbas Jamalipour This is my paper · ORCID
classification cs.NI
keywords largelanguagemodelssmallpromptengineeringpowerallocationjointoptimizationcompressiondeepreinforcementlearningdenoising-inspired
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

Mobile LLM requests are slowed twice by long prompts: uploading them over wireless links costs time, and the cloud model spends more time processing more tokens. This paper claims that both costs can be cut together by compressing the prompt at the edge with a small language model and, at the same time, choosing the transmit power with a deep reinforcement learning agent. The central move is to compress iteratively, treating the long original prompt as a noisy image and the compressed prompt as the clean image, so that four steps, each halving the prompt, preserve more information than one step that cuts the prompt to a sixteenth of its length. The paper reports that this joint design cuts total service time by 17% relative to no compression while keeping its measured output quality, and by 46.5% when the aggressive one-sixteenth compression ratio is used. A careful reader should care because prompt lengths are growing rapidly with in-context learning and document uploads, making the compression-versus-power trade-off a practical lever for low-latency AI services.

What carries the argument

The load-bearing object is the composite fidelity metric $f = \phi_1 f_1 + \phi_2 f_2 + \phi_3 f_3$, with fixed weights $(0.4, 0.3, 0.3)$, where $f_1$ is the token overlap between original and compressed prompts, $f_2$ is the compression ratio scaled by $(1 - \mathrm{BEP}(\mathrm{SNR}))$, and $f_3$ is the token overlap between the LLM's answer on the compressed and uncompressed prompts. This single scalar appears both in the optimization objective and as the deep-reinforcement-learning reward, so every reported gain in service time is measured against it. The second mechanism is the denoising-inspired compression schedule $\alpha(t) = \psi_\kappa^{-\sigma(t)}$ with $\sigma(t)$ chosen as linear, cosine, or quadratic; this identity converts one large, risky compression step into $M$ controlled smaller steps, letting the small language model preserve more information at the same final ratio.

What would settle it

Run the same long-document tasks with no compression versus sixteen-fold iterative compression and measure answer quality with a metric independent of token overlap, such as exact-match or human ratings; if output quality falls sharply while the paper's fidelity score stays high, the central quality claim is false. Alternatively, measure the full pipeline on real mobile hardware and a real fading link: if the 17% service-time reduction at maintained quality does not reproduce end-to-end, then the delay model, not the compression method, is the weak link.

Watch

Extended reading notes

Core claim

The paper's central claim is that prompt compression ratio and wireless transmit power are coupled decision variables in mobile LLM serving, and that a learned policy over both outperforms any fixed choice of either. It formalizes the objective as maximizing a composite fidelity score under energy, power, latency, and minimum-fidelity constraints, then solves it with a double deep Q-network whose reward is exactly that score. The discovery that carries the paper is the compression-ratio scheduling idea: instead of compressing a sixteen-fold target in one pass, the framework applies a schedule that removes a little at each step, and the paper finds that four iterative halving steps outperform one direct sixteenth-size compression on fidelity, reward, and total time. It also finds that the shape of the schedule matters, with cosine and quadratic schedules outperforming linear ones on long meeting transcripts, and that the best operating point sits at moderate compression and moderate power, confirming that neither lever can be tuned alone.

Load-bearing premise

The entire optimization and every quality claim run through one hand-weighted fidelity score built from token overlap and transmission error rate; if that score does not track how well the model actually answers the user's question, then the reported latency savings may be accompanied by an unmeasured drop in output quality.

Editorial extensions

If this is right

  • If the claims hold, service providers can cut mobile LLM latency by roughly 17% without any quality loss on the paper's fidelity measure, just by adding edge prompt compression and learned power control.
  • At aggressive compression, iterative refinement dominates one-shot compression: at sixteen-fold compression, the iterative method improves the service-time reduction from 42.3% to 46.5% while also scoring higher on fidelity and the learned reward.
  • Because the optimal reward occurs at moderate compression and moderate power, static heuristics that fix one of the two variables will leave latency and quality on the table.
  • The choice of compression schedule is a real degree of freedom: cosine and quadratic schedules outperform linear on long multi-speaker documents, so prompt structure should inform how compression steps are spaced.
  • The DRL policy converges and adapts to changing channel conditions in simulation, suggesting the same joint state-action design could be retargeted to new networks by swapping the reward and action definitions.

Reading between the lines

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

  • A natural extension the paper leaves open is to make the schedule shape and the number of iterations decision variables rather than fixed hyperparameters; its own results showing that quadratic and cosine schedules beat linear on structured documents suggest the gains would grow.
  • The reported quality preservation is only as strong as the token-overlap fidelity measure; a testable check would re-run the same experiments using an external semantic metric or human judgment on the compressed prompts.
  • The energy model relies on thermal design power and estimated GPU times, so the end-to-end latency and energy trade-off should be validated on real edge hardware before deployment claims are generalized.
  • The 17%-at-fixed-quality and 46.5%-at-sixteen-fold numbers come from a mix of two datasets and a simulation of the wireless link; a dedicated end-to-end prototype with live radio conditions would show how much of the gain survives outside simulation.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes JPPO and JPPO++, frameworks that jointly optimize SLM-based prompt compression and wireless transmission power for mobile LLM services. A DRL agent selects discrete compression and power levels to maximize a composite fidelity metric subject to energy and latency constraints. The authors report that JPPO++ reduces service time by 17% versus no compression, and that 16x compression reduces service time by 42.3% (JPPO) and 46.5% (JPPO++) while maintaining output quality. The evaluation uses a weighted fidelity score as both the DRL reward and the main effectiveness measure, supplemented by a LongBench task-quality comparison and simulated latency numbers.

Significance. If the reported results were reliable, the paper would address a timely and practical problem: cutting both transmission and inference costs for mobile LLM services by coupling prompt compression with power control. The denoising-inspired iterative compression idea is potentially interesting, and the paper gives a clearly stated optimization problem with reproducible seeds and an explicit Double DQN formulation. However, the central evidence is not reliable: the quality metric is circular, the only independent task-level evaluation contradicts the headline quality-preservation claim, and the reported service-time reductions are internally inconsistent. The paper does not establish its main quantitative claims as submitted.

major comments (4)
  1. [Section 4.2.1, Section 5.2.1, Fig. 4, Eqs. (4)-(7), Table 2] The composite fidelity f = 0.4 f1 + 0.3 f2 + 0.3 f3 is used both as the DRL reward (Section 4.2.1, Reward) and as the effectiveness metric in Fig. 4 and Section 5.2.1. The agent optimizes exactly the function on which it is later evaluated, so the reported fidelity and reward improvements partly reflect the policy maximizing its own objective rather than an independent measure of output quality. Moreover, f1 (Eq. (4)) and f3 (Eq. (6)) are token-overlap rates, while f2 (Eq. (5)) is alpha times (1 - BEP), so one fidelity component is by construction proportional to the inverse compression ratio. None of these components is validated against semantic or task-level quality. The only independent task-level evidence, Fig. 7, shows no compression outperforming JPPO++ on Multi-news (QA F1 0.23 vs 0.19, ROUGE-L 0.25 vs 0.16) and on GovReport (F1 0.37 vs 0.26/0.31, ROUGE-L 0.33 vs 0.18/0.21). The abstract's unconditional claim of maintaining output quality is therefore unsupported and contradicted by the paper's own data.
  2. [Eq. (19d), Section 5.2.1] The constraint f > f_th in Eq. (19d) is never assigned a numerical value, and Section 5.2.1 invokes an unexplained 'tolerating up to a 30% fidelity drop.' Without a concrete f_th, the 'acceptable loss' claim is unfalsifiable, and the optimization problem is missing a stated feasibility bound. The 30% figure is a drop in the internal composite f, not in any task-level metric, so it cannot justify the conclusion that output quality is maintained under aggressive compression.
  3. [Abstract, Section 5.2.1, Fig. 6] The service-time reduction numbers are mutually inconsistent. The text reports that single-round 16x compression reduces service time by 42.3% and iterative compression by 46.5% relative to no compression, while Fig. 6(b) reports that the iterative method reduces service time by 39-52% relative to the single-round baseline. If both statements were true, the iterative-vs-no-compression reduction would be roughly 65-72% (using the Fig. 6 values) or the iterative-vs-single-round reduction would be about 7% (using 46.5% versus 42.3%), not 39-52%. The abstract's separate 17% reduction figure also does not appear in the body. These inconsistencies undermine the paper's central quantitative claims.
  4. [Section 5.2, related work [16], [27]] The experimental evaluation compares JPPO and JPPO++ only against no compression and one-shot compression; it does not compare against the prompt-compression or power-allocation baselines discussed in the related work, such as LLMLingua [16] or PCRL [27]. Because the paper motivates JPPO++ by the limitations of those methods, the absence of any comparison leaves the claimed practical advantage over prior approaches unestablished.
minor comments (6)
  1. [Table 2 and Eq. (7)] The weights are labeled alpha_1, alpha_2, alpha_3 in Table 2 but phi_1, phi_2, phi_3 in Eq. (7); the notation should be made consistent.
  2. [Section 4.2.1 and Algorithm 1] The state vector is written as [f_n(eta_n), gamma_n, ] with an empty entry, and the BEP mentioned in the text is missing from the vector; Algorithm 1 also lists 'target privacy parameters (epsilon, delta)' that are not used in the problem.
  3. [Section 4.2.1 and Tables 3-6] The compression action is described as a discrete level from 0 to 4 and the power level from 0 to 9, but Tables 3-6 use rows and columns numbered 1 to 10; the mapping between action levels and table indices should be clarified.
  4. [Section 5.1, 5.2.3, Fig. 5] Table 2 reports 10,000 episodes per test run, but Section 5.2.3 and Fig. 5 report reward convergence over 100 episodes; the relationship between these two numbers should be stated explicitly.
  5. [Eq. (20) and Eq. (22)] Eq. (20) is the standard Q-learning update rather than the Double DQN target; the text should clarify that Eq. (22) is the actual Double DQN update used.
  6. [Fig. 7] The LongBench results in Fig. 7 are reported without error bars or statistical significance; given the small differences in the Multi-news category, some measure of variance is needed to support the claim of minimal degradation.

Circularity Check

1 steps flagged · score 6.0 of 10

Quality-preservation claim is self-referential: the fidelity metric f in Eqs. (4)-(7) doubles as the DRL reward and as the paper's evidence of preserving output quality, so reported high f partly reflects the policy optimizing its own objective.

  1. self definitional [Eq. (7) in Sec. 3.2; Sec. 4.2.1 Reward definition; Sec. 5.2.1 Effectiveness analysis]
    "The overall fidelity metric f can be defined as a weighted sum of these components as: f = φ1 f1 + φ2 f2 + φ3 f3. ... The reward maximizes fidelity while minimizing penalties related to BEP and power usage. ... Based on the fidelity metric f, the iterative strategy consistently outperforms the single-step counterpart across all three components, demonstrating better semantic preservation and communication robustness."

    The paper defines its quality measure f in Eqs. (4)-(7), then uses exactly the same f as the objective in Eq. (19) and as the DRL reward in Sec. 4.2.1. The effectiveness evaluation in Sec. 5.2.1 then cites this same f to claim 'better semantic preservation and communication robustness' and, together with Fig. 6, to assert that JPPO++ 'preserves task quality'. Because the DRL policy is rewarded for maximizing f, reporting higher f for the trained policy is the agent optimizing its own objective; it is not an external demonstration that task-level output quality is maintained. The weights (0.4, 0.3, 0.3) are hand-set rather than fitted, so this is not a fitted-input case, but the objective/evaluation identity is a self-definitional reduction of the quality half of the central claim.

full rationale

The service-time reductions are arithmetic consequences of the compression and transmission model and are not circular: shortening prompts reduces transmission and LLM inference time in Eqs. (15)-(17). The self-citation to the authors' ICC version [1] is not load-bearing; no derivation or uniqueness claim depends on it. The circularity is concentrated in the quality half of the central claim: the fidelity metric f is defined as the paper's quality measure, used as the optimization objective and DRL reward, and then reused as evidence of semantic preservation and task-quality preservation. A policy trained to maximize f predictably scores higher on f, which is optimization rather than external validation. The paper does attempt an independent check in Fig. 7 using LongBench QA F1 and ROUGE-L, but that evidence shows no compression is best on every reported score (e.g., Multi-news QA F1 0.23 vs 0.20/0.19; GovReport ROUGE-L 0.33 vs 0.18/0.21), so it does not support the 'maintaining output quality' claim. The threshold f_th in Eq. (19d) is never specified, leaving 'acceptable loss' undefined. Because the latency claim is independent and an external benchmark is present, the circularity is partial rather than total, giving score 6.

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

The paper introduces no new physical entities such as particles or forces. Its load-bearing additions are hand-set parameters and unvalidated modeling assumptions: the fidelity weights, the compression schedules, the token-overlap fidelity metric, and the simulation-based time model.

free parameters (4)
  • Fidelity weights phi1, phi2, phi3 = 0.4, 0.3, 0.3
    Set by hand in Table 2, not learned or justified; they define the objective f in Eq. (7) and therefore all reported reward and fidelity comparisons.
  • Compression schedule sigma(t) = linear, cosine, quadratic
    Three arbitrary schedules in Eqs. (10)-(12); the performance ranking across schedules depends on this choice, and the paper does not optimize it.
  • Target compression ratio and iteration count = 16x, 4 iterations of 2x
    Chosen as the experimental setting in Section 5; no procedure is given for selecting them for a given prompt or task.
  • Discrete action levels for compression and power = compression levels 0-4, power levels 0-9
    Arbitrary discretization in Section 4.2.1; the reported reward tables may depend on this granularity.
assumptions (5)
  • ad hoc to paper Token overlap is a valid measure of semantic and task fidelity
    Eqs. (4)-(6) define f1 and f3 as token overlaps; no correlation with human judgment or standard task metrics is established.
  • domain assumption AWGN Rayleigh fading model with the given BEP formula applies to compressed text transmission
    Eqs. (16)-(18) assume a standard wireless model from the literature; this is acceptable background but is not validated for compressed natural language prompts.
  • ad hoc to paper The DDPM denoising analogy transfers to text compression
    Section 3.3.1 argues by analogy that gradual compression reduces information loss; no formal equivalence or proof is provided.
  • domain assumption Energy and delay models from LLMCarbon and GPU TDP accurately capture SLM and LLM costs
    Eqs. (13)-(15) rely on citation [31] and on linear scaling assumptions that are not verified in this paper.
  • domain assumption DRL with the defined state, action, and reward converges to a policy that reflects real system performance
    Section 4.2.1 defines the DQN and Section 5.2.3 shows convergence in reward, but there is no proof or external validation that the learned policy matches real-world behavior.

how reviews work

0 comments
Cite this review

Pith. "Pith review of JPPO++: Joint Power and Denoising-inspired Prompt Optimization for Mobile LLM Services." pith.science (2026). https://pith.science/paper/SQ4BHPHD

@misc{pith2026241203621,
  author       = {Pith},
  title        = {Pith review of: JPPO++: Joint Power and Denoising-inspired Prompt Optimization for Mobile LLM Services},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SQ4BHPHD}},
  note         = {Machine review of arXiv:2412.03621}
}
read the original abstract

Large Language Models (LLMs) are increasingly integrated into mobile services over wireless networks to support complex user requests. This trend has led to longer prompts, which improve LLMs' performance but increase data transmission costs and require more processing time, thereby reducing overall system efficiency and negatively impacting user experience. To address these challenges, we propose Joint Prompt and Power Optimization (JPPO), a framework that jointly optimizes prompt compression and wireless transmission power for mobile LLM services. JPPO leverages a Small Language Model (SLM) deployed at edge devices to perform lightweight prompt compression, reducing communication load before transmission to the cloud-based LLM. A Deep Reinforcement Learning (DRL) agent dynamically adjusts both the compression ratio and transmission power based on network conditions and service constraints, aiming to minimize service time while preserving response fidelity. We further extend the framework to JPPO++, which introduces a denoising-inspired compression scheme. This design performs iterative prompt refinement by progressively removing less informative tokens, allowing for more aggressive yet controlled compression. Experimental results show that JPPO++ reduces service time by 17% compared to the no-compression baseline while maintaining output quality. Under compression-prioritized settings, a reduction of up to 16x in prompt length can be achieved with an acceptable loss in accuracy. Specifically, JPPO with a 16x ratio reduces total service time by approximately 42.3%, and JPPO++ further improves this reduction to 46.5%.

Figures

Figures reproduced from arXiv: 2412.03621 by the authors.

Figure 1
Figure 1. Time consumption comparison of first token gen [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. System model of wireless network-aided mobile LLM [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The motivation of our proposed denoising-inspired prompt compression method. Part A illustrates the process of [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: The example illustrates wireless network-aided LLM services with SLM-based prompt compression, with a [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: The convergence performance of reward over 100 [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: The illustration of transmission time and service [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: The score performance when running with the [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 30 canonical work pages

  1. [16]

    LLMLingua: Compressing prompts for accelerated inference of large language models,

    H. Jiang, Q. Wu, C.-Y. Lin, Y. Yang, and L. Qiu, “LLMLingua: Compressing prompts for accelerated inference of large language models,” arXiv preprint arXiv:2310.05736, 2023

    arXiv 2023

  2. [27]

    Discrete prompt compression with rein- forcement learning,

    H. Jung and K.-J. Kim, “Discrete prompt compression with rein- forcement learning,”IEEE Access, vol. 12, pp. 72 578–72 587, 2024

    work page 2024

  3. [1]

    JPPO: Joint power and prompt optimization for accelerated large language model services,

    F. You, H. Du, K. Huang, and A. Jamalipour, “JPPO: Joint power and prompt optimization for accelerated large language model services,” inIEEE International Conference on Communications, to appear, 2025

    work page 2025

  4. [2]

    Large language models (LLMs) inference offloading and resource allocation in cloud-edge computing: An active inference approach,

    Y. He, J. Fang, F. R. Yu, and V . C. Leung, “Large language models (LLMs) inference offloading and resource allocation in cloud-edge computing: An active inference approach,”IEEE Transactions on Mobile Computing, 2024

    work page 2024

  5. [3]

    A survey on large language models for communication, network, and service management: Application insights, challenges, and future directions,

    G. O. Boateng, H. Sami, A. Alagha, H. Elmekki, A. Hammoud, R. Mizouni, A. Mourad, H. Otrok, J. Bentahar, S. Muhaidatet al., “A survey on large language models for communication, network, and service management: Application insights, challenges, and future directions,”IEEE Communications Surveys & Tutorials, 2025

    2025

  6. [4]

    EdgeMoE: Empowering sparse large language models on mobile devices,

    R. Yi, L. Guo, S. Wei, A. Zhou, S. Wang, and M. Xu, “EdgeMoE: Empowering sparse large language models on mobile devices,” IEEE Transactions on Mobile Computing, 2025

    work page 2025

  7. [5]

    Mobile edge intelligence for large language models: A contemporary survey,

    G. Qu, Q. Chen, W. Wei, Z. Lin, X. Chen, and K. Huang, “Mobile edge intelligence for large language models: A contemporary survey,”IEEE Communications Surveys & Tutorials, 2025

    2025

  8. [6]

    Indus- trial internet of things with large language models (LLMs): an intelligence-based reinforcement learning approach,

    Y. Ren, H. Zhang, F. R. Yu, W. Li, P . Zhao, and Y. He, “Indus- trial internet of things with large language models (LLMs): an intelligence-based reinforcement learning approach,”IEEE Trans- actions on Mobile Computing, 2024

    work page 2024

Show all 37 references
  1. [7]

    LLM-based edge intelligence: A com- prehensive survey on architectures, applications, security and trustworthiness,

    O. Friha, M. Amine Ferrag, B. Kantarci, B. Cakmak, A. Ozgun, and N. Ghoualmi-Zine, “LLM-based edge intelligence: A com- prehensive survey on architectures, applications, security and trustworthiness,”IEEE Open Journal of the Communications Society, vol. 5, pp. 5799–5856, 2024. 13

    2024

  2. [8]

    Recent advances in natural language processing via large pre-trained language models: A survey,

    B. Min, H. Ross, E. Sulem, A. P . B. Veyseh, T. H. Nguyen, O. Sainz, E. Agirre, I. Heintz, and D. Roth, “Recent advances in natural language processing via large pre-trained language models: A survey,”ACM Computing Surveys, vol. 56, no. 2, pp. 1–40, 2023

    2023

  3. [9]

    Large language model enhanced multi-agent systems for 6G communications,

    F. Jiang, Y. Peng, L. Dong, K. Wang, K. Yang, C. Pan, D. Niyato, and O. A. Dobre, “Large language model enhanced multi-agent systems for 6G communications,”IEEE Wireless Communications, 2024

    2024

  4. [10]

    A review of current trends, techniques, and challenges in large language models (LLMs),

    R. Patil and V . Gudivada, “A review of current trends, techniques, and challenges in large language models (LLMs),”Applied Sciences, vol. 14, no. 5, p. 2074, 2024

    2024

  5. [11]

    What makes for good tokenizers in vision transformer?

    S. Qian, Y. Zhu, W. Li, M. Li, and J. Jia, “What makes for good tokenizers in vision transformer?”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 11, pp. 13 011–13 023, 2023

    2023

  6. [12]

    Long-context LLMs struggle with long in-context learning,

    T. Li, G. Zhang, Q. D. Do, X. Yue, and W. Chen, “Long-context LLMs struggle with long in-context learning,” arXiv preprint arXiv:2404.02060, 2024

    2024 arXiv

  7. [13]

    Chain-of-thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, F. Xia, E. Chi, Q. V . Le, D. Zhouet al., “Chain-of-thought prompting elicits reasoning in large language models,”Advances in Neural Information Processing System, vol. 35, pp. 24 824–24 837, 2022

    2022

  8. [14]

    Efficient prompting for LLM-based generative internet of things,

    B. Xiao, B. Kantarci, J. Kang, D. Niyato, and M. Guizani, “Efficient prompting for LLM-based generative internet of things,”IEEE Internet of Things Journal, pp. 1–1, 2024

    2024

  9. [15]

    To repeat or not to repeat: Insights from scaling LLM under token-crisis,

    F. Xue, Y. Fu, W. Zhou, Z. Zheng, and Y. You, “To repeat or not to repeat: Insights from scaling LLM under token-crisis,”Advances in Neural Information Processing Systems, vol. 36, 2024

    2024

  10. [17]

    LLM-Slice: Dedicated wireless network slicing for large language models,

    B. Liu, J. Tong, and J. Zhang, “LLM-Slice: Dedicated wireless network slicing for large language models,” inProceedings of the ACM Conference on Embedded Networked Sensor Systems, 2024, pp. 853–854

    2024

  11. [18]

    Deeploy: Enabling energy- efficient deployment of small language models on heterogeneous microcontrollers,

    M. Scherer, L. Macan, V . J. B. Jung, P . Wiese, L. Bompani, A. Burrello, F. Conti, and L. Benini, “Deeploy: Enabling energy- efficient deployment of small language models on heterogeneous microcontrollers,”IEEE Transactions on Computer-Aided Design of Integrated Circuits and ...

    2024

  12. [19]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P . Abbeel, “Denoising diffusion probabilistic models,”Advances in neural information processing systems, vol. 33, pp. 6840–6851, 2020

    2020

  13. [20]

    Wire- lessLLM: Empowering large language models towards wireless intelligence,

    J. Shao, J. Tong, Q. Wu, W. Guo, Z. Li, Z. Lin, and J. Zhang, “Wire- lessLLM: Empowering large language models towards wireless intelligence,”Journal of Communications and Information Networks, vol. 9, no. 2, pp. 99–112, 2024

    2024

  14. [21]

    Edge intelligence optimization for large language model inference with batching and quantization,

    X. Zhang, J. Liu, Z. Xiong, Y. Huang, G. Xie, and R. Zhang, “Edge intelligence optimization for large language model inference with batching and quantization,” in2024 IEEE Wireless Communications and Networking Conference (WCNC), 2024, pp. 1–6

    2024

  15. [22]

    Beyond the cloud: Edge inference for generative large language models in wireless networks,

    X. Zhang, J. Nie, Y. Huang, G. Xie, Z. Xiong, J. Liu, D. Niyato, and X. S. Shen, “Beyond the cloud: Edge inference for generative large language models in wireless networks,”IEEE Transactions on Wireless Communications, pp. 1–1, 2024

    2024

  16. [23]

    Large multi-modal models (LMMs) as universal foundation models for AI-native wireless systems,

    S. Xu, C. Kurisummoottil Thomas, O. Hashash, N. Muralidhar, W. Saad, and N. Ramakrishnan, “Large multi-modal models (LMMs) as universal foundation models for AI-native wireless systems,”IEEE Network, vol. 38, no. 5, pp. 10–20, 2024

    2024

  17. [24]

    Adapting LLMs for efficient context processing through soft prompt compression,

    C. Wang, Y. Yang, R. Li, D. Sun, R. Cai, Y. Zhang, and C. Fu, “Adapting LLMs for efficient context processing through soft prompt compression,” inProceedings of the International Conference on Modeling, Natural Language Processing and Machine Learning, 2024, pp. 91–97

    2024

  18. [25]

    Prompt-assisted semantic interference cancelation on moderate interference chan- nels,

    Z. Meng, Q. Li, A. Pandharipande, and X. Ge, “Prompt-assisted semantic interference cancelation on moderate interference chan- nels,”IEEE Wireless Communications Letters, vol. 13, no. 10, pp. 2847–2851, 2024

    2024

  19. [26]

    Cross modal compression with variable rate prompt,

    J. Gao, J. Li, C. Jia, S. Wang, S. Ma, and W. Gao, “Cross modal compression with variable rate prompt,”IEEE Transactions on Multimedia, vol. 26, pp. 3444–3456, 2024

    2024

  20. [28]

    Intelligent cloud-edge collaborations for energy-efficient user association and power allocation in space-air-ground integrated networks,

    Z. Wang, L. Zhang, D. Feng, G. Wu, and L. Yang, “Intelligent cloud-edge collaborations for energy-efficient user association and power allocation in space-air-ground integrated networks,”IEEE Journal on Selected Areas in Communications, vol. 42, no. 12, pp. 3659–3673, 2024

    2024

  21. [29]

    Joint resource allocations for energy consumption optimization in HAPS-aided MEC-NOMA systems,

    X. Yu, X. Zhang, Y. Rui, K. Wang, X. Dang, and M. Guizani, “Joint resource allocations for energy consumption optimization in HAPS-aided MEC-NOMA systems,”IEEE Journal on Selected Areas in Communications, vol. 42, no. 12, pp. 3632–3646, 2024

    2024

  22. [30]

    Graph neural networks approach for joint wireless power control and spectrum allocation,

    M. Marwani and G. Kaddoum, “Graph neural networks approach for joint wireless power control and spectrum allocation,”IEEE Transactions on Machine Learning in Communications and Networking, vol. 2, pp. 717–732, 2024

    2024

  23. [31]

    LLMCarbon: Modeling the end-to-end carbon footprint of large language models,

    A. Faiz, S. Kaneda, R. Wang, R. Osi, P . Sharma, F. Chen, and L. Jiang, “LLMCarbon: Modeling the end-to-end carbon footprint of large language models,” inProceedings of the International Con- ference on Learning Representations. ICLR, 2024

    2024

  24. [32]

    Tse and P

    D. Tse and P . Viswanath,Fundamentals of wireless communication. Cambridge university press, 2005

    2005

  25. [33]

    I. S. Gradshteyn and I. M. Ryzhik,Table of Integrals, Series, and Products, 7th ed. Academic Press, 2007

    2007

  26. [34]

    Deep reinforcement learn- ing with double q-learning,

    H. Van Hasselt, A. Guez, and D. Silver, “Deep reinforcement learn- ing with double q-learning,” inProceedings of the AAAI conference on artificial intelligence, vol. 30, no. 1, 2016

    2016

  27. [35]

    D. Rothman,Transformers for Natural Language Processing: Build, train, and fine-tune deep neural network architectures for NLP with Python, Hugging Face, and OpenAI’s GPT-3, ChatGPT, and GPT-4. Packt Publishing Ltd, 2022

    2022

  28. [36]

    Meetingbank: A benchmark dataset for meeting summarization,

    Y. Hu, T. Ganter, H. Deilamsalehy, F. Dernoncourt, H. Foroosh, and F. Liu, “Meetingbank: A benchmark dataset for meeting summarization,” inProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL). Toronto, Canada: Association for Computatio...

    2023

  29. [37]

    Longbench: A bilingual, multi- task benchmark for long context understanding,

    Y. Bai, X. Lv, J. Zhang, H. Lyu, J. Tang, Z. Huang, Z. Du, X. Liu, A. Zeng, L. Houet al., “Longbench: A bilingual, multi- task benchmark for long context understanding,”arXiv preprint arXiv:2308.14508, 2023

    2023 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.