Pith. sign in

REVIEW 2 major objections 5 minor 42 references

Split Fine-Tuning for Large Language Models in Wireless Networks

T0 review · 2 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper claims that split fine-tuning, combining parallel split learning with LoRA and three-stage activation compression, can fine-tune large language models on memory-constrained mobile devices in wireless networks, cutting…

desk verdict A plausible wireless split-fine-tuning system with a genuine integration contribution, but the headline delay numbers rest on a serial/parallel inconsistency in the delay model. read the letter →

arxiv 2501.09237 v1 pith:NTYYTOIT submitted 2025-01-16 cs.DC

Songge Zhang , Guoliang Cheng , Xinyu Huang , Zuguang Li , Wen Wu , Lingyang Song , Xuemin Shen This is my paper · ORCID
classification cs.DC
keywords splitlearningLoRAlargelanguagemodelfine-tuningactivationcompressionwirelessedgecomputingresourceallocationmixed-integeroptimization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proposes Split Fine-Tuning (SFT), a scheme for fine-tuning large language models on mobile devices that cannot hold the full model in memory. The model is cut at a chosen transformer block; the device runs the lower blocks with local LoRA adapters, the edge server runs the upper blocks on one shared model, and compressed activations travel over the wireless link. The paper claims this combination of parallel split learning, a three-stage compression pipeline (sparsification, stochastic quantization, lossless encoding), and two-timescale resource allocation cuts fine-tuning delay by up to 80.2% and communication overhead by 93.6% relative to federated and split-learning baselines, while keeping device memory and accuracy within constraints. If true, personalized fine-tuning of LLMs becomes feasible on memory-constrained mobile hardware in wireless networks.

What carries the argument

The load-bearing mechanism is the transformer cut: the $L$ transformer blocks are split at layer $l$, giving each device a small device-side model (embedding plus $l$ blocks with LoRA) and leaving the remaining $L-l$ blocks on the server, so devices send only compressed intermediate activations instead of model updates. The round-delay model $\tau^t = \max_{n \in \mathcal{N}} \tau_n^t$ sums seven phases per device (distribution, device compute, activation upload, server compute, gradient download, device update, LoRA upload) and then takes the maximum over devices, which is what the two-timescale optimizer minimizes. The optimizer separates variables by timescale: the augmented Lagrangian method chooses sparsification rate $\rho$, quantization level $E$, and cut layer $l$ in the large timescale, while sequential quadratic programming allocates bandwidth $b_n$ in the small timescale. A third-order fitted accuracy function $A(\rho, E)$ acts as the constraint that ties compression choices to fine-tuning quality.

What would settle it

In the reported eight-device setup, log the server's GPU busy time per fine-tuning round. If the server processes devices sequentially, the total server time will grow linearly with the number of devices; recomputing the round delay with server delay summed over devices rather than maximized would show whether the 80.2% reduction still holds.

Watch

Extended reading notes

Core claim

The central claim is that fine-tuning an LLM in a wireless network does not require the full model on any device: splitting the transformer at layer $l$, keeping only $l$ blocks plus LoRA adapters on the device, and letting all devices share one server-side model reduces device memory by about 58.2% relative to federated LoRA training while preserving accuracy. The paper further claims that compressing the intermediate activations with Top-K sparsification, stochastic quantization, and run-length lossless encoding keeps accuracy loss within 2% at up to 80% sparsity and 3-bit quantization, while shrinking transmitted data to roughly 6.8% of the original. The delay minimization chooses sparsification rate, quantization level, cut layer, and per-device bandwidth; the reported result is that this joint optimization reduces fine-tuning delay by up to 80.2% and communication overhead by 93.6% compared with the baselines.

Load-bearing premise

The delay model takes the maximum over devices, which assumes the server can process all devices' tasks at the same time, but the paper also states the server runs those tasks in the order activations arrive; if the server-side work is serial rather than parallel, the reported delay reductions would be smaller.

Editorial extensions

If this is right

  • Devices with roughly 4-8 GB of memory can participate in LLM fine-tuning by holding only a few transformer blocks and their LoRA adapters, while the server holds the rest.
  • The three-stage compression scheme can cut activation uploads to about 6.8% of their original size with accuracy loss kept within 2%.
  • Allocating bandwidth and choosing the cut layer jointly at different timescales reduces per-round communication delay by up to 53.1% even at a 5 MHz system bandwidth.
  • Total fine-tuning delay drops by up to 80.2% and communication overhead by 93.6% versus federated fine-tuning and split-learning baselines.

Reading between the lines

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

  • A natural extension is to apply the same parallel-split structure to decoder-only LLMs, where the compressed hidden states are token sequences rather than patch embeddings; the delay model would need an autoregressive generation factor.
  • The offline third-order accuracy fit could be replaced by an online accuracy estimator, allowing the optimizer to handle datasets and tasks not seen during fitting.
  • The downlink gradient transmission could be compressed with the same sparsification-quantization-encoding pipeline, which would shrink the gradient-transmission term in the round-delay sum beyond what the paper reports.
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

2 major / 5 minor

Summary. The paper proposes Split Fine-Tuning (SFT), a scheme for fine-tuning large language models in wireless networks by splitting the model between mobile devices and an edge server, using LoRA adapters, activation compression (sparsification, stochastic quantization, and lossless encoding), and a two-timescale resource-allocation algorithm that jointly optimizes compression rates, transformer-block allocation, and bandwidth. The authors claim that SFT reduces fine-tuning delay by up to 80.2% and communication overhead by up to 93.6% compared with federated-learning and split-learning baselines while satisfying device memory and accuracy constraints. The manuscript includes a system model, a delay/memory/communication analysis, a formulated mixed-integer optimization problem, an augmented-Lagrangian/SQP solution approach, and simulations on CIFAR-100 and Tiny-ImageNet under IID and non-IID settings.

Significance. If the central claims hold, the paper would make a useful contribution to edge-side LLM fine-tuning by addressing two practical bottlenecks: device memory and communication cost. The combination of split learning, LoRA, and joint compression/resource optimization is sensible and the evaluation covers several relevant baselines and data distributions. The paper gives credit for a fairly complete system model and a detailed memory-consumption analysis, and it ships a concrete algorithmic framework. However, the main delay-reduction claim is currently supported by a delay model that appears internally inconsistent with the described sequential server execution, and the accuracy feasibility claim rests on fitted surrogate models. These issues are load-bearing for the headline results, so the manuscript needs substantive revision before the claims can be accepted.

major comments (2)
  1. [§IV-A and §V-A, Eq. (19)] The round-delay model in Eq. (19) takes τ^t = max_n τ^t_n, which treats the server-side computation for all devices as fully overlapped. This contradicts Section IV-A, which states that the server 'will perform the LoRA FP sequentially according to the order in which the immediate activation is received,' and Algorithm 1 (lines 5–18) shows a per-device loop in which the server executes FP/BP before moving to the next device. With a single shared server-side model, the server-side work must be serialized, so the round delay should include a sum over devices of τ_SC(n) (or at least the serialized server component), not a pure maximum. Because the reported 80.2% delay reduction in Section VIII is computed from the max model, this inconsistency directly affects the central quantitative claim. The authors should either justify how the server-side computation is actually parallelized (e.g., separate server replicas or pipelining) or revise the delay model and rerun the simulations.
  2. [§IV-C and §VI, Eq. (27b)] The accuracy constraint in Eq. (27b) uses A(ρ, E), which is a third-order polynomial fitted to offline data in Section IV-C (Fig. 3), and the compression-strategy mapping in Section IV-B is a piecewise linear function fitted to offline data. Thus the 'optimal' compression rates and the feasibility of the accuracy constraint are determined by fitted surrogate models, not by actual measured fine-tuning accuracy. If the delay results in Section VIII enforce Eq. (27b) through these surrogates, then the claim 'while satisfying ... model accuracy constraints' is not established for the true model. The authors should validate the optimized (ρ*, E*, l*) on actual fine-tuning runs and provide sensitivity analysis with respect to the surrogate fitting error; reporting the fit MSE is not sufficient.
minor comments (5)
  1. [§V-A, Eq. (12)] Eq. (12) writes τ^t_CC(l) on the left-hand side but the right-hand side depends on device n via f_n, C^u_n, and D^u_n; the device index should appear in the left-hand side's notation.
  2. [§V-A, Eq. (14)] Eq. (14) uses the notation f^s_n for the server-side frequency, but the server is a single shared resource; please clarify whether the server frequency is device-dependent or whether f^s_n denotes something else, and update the notation consistently.
  3. [§IV-B and §V-A, Eqs. (10), (13), (15)] The delay expressions in Eqs. (13) and (15) use β as the compression ratio, while the sparsification rate is defined as ρ in Eq. (10); the relationship between β, ρ, the quantization level E, and the lossless encoding should be stated explicitly to avoid ambiguity.
  4. [Algorithm 2] The constraint enumeration contains typos: 'for q = 1' should be 'p = 1', and the lower-bound constraint on l is listed twice; please correct the indexing and remove the duplicate.
  5. [§VIII-A and Table II] Table II lists the system bandwidth B as 30 MHz, but Section VIII-A states that 'We consider a system bandwidth of 5 MHz' and Figure 9 discusses a 'constrained 5 MHz bandwidth'; please align these values and clarify which bandwidth is used for the reported results.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the headline delay/communication claims are simulation results against external baselines, and the data-fitted accuracy/compression surrogates are transparently empirical rather than load-bearing self-derivations.

full rationale

The paper's derivation chain is self-contained. The delay model (Eqs. 11-20), memory model (Eqs. 21-26), and communication/workload analysis (Section V-C) are analytical expressions derived from stated system assumptions, and the reported 80.2% delay reduction and 93.6% communication reduction are simulation results benchmarked against FL-based FT, SL-based FT, and SFT-without-compression (Section VIII), rather than outputs of the optimization's own surrogate. The two data-fitted elements - the piecewise-linear compression-strategy mapping in Section IV-B and the third-order accuracy function A(rho,E) in Section IV-C - are explicitly presented as data-driven approximations ('we approached the problem from a data-driven perspective'), used as constraints/surrogates in P1. They are not first-principles derivations, but neither is A defined in terms of the delay objective nor is the delay objective defined in terms of A; the optimization could in principle be re-run with any other accuracy model. The paper does not relabel these fits as external predictions, and Fig. 5 validates actual convergence behavior. Self-citations (e.g., [26], [27], [37]) are contextual related-work/support citations and are not load-bearing; no uniqueness theorem is imported. The reader-noted tension between the max round-delay in Eq. (19) and the stated sequential server-side LoRA FP in Section IV-A is an internal modeling-consistency concern about whether server work truly overlaps; it does not make any quantity equal to its own input by construction, so it is outside the circularity definition and does not affect the score.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced; the only invented components are architectural (the split fine-tuning scheme) and modeling surrogates (the fitted accuracy and compression prediction functions). The fitted functions are the main free parameters the central optimization depends on.

free parameters (2)
  • Third-order accuracy model A(rho,E) = coefficients not reported
    Section IV-C fits observed accuracy to a third-order function (MSE < 0.26%); this fitted curve is used as the accuracy constraint in Eq. (27b), so the optimal compression choices depend on a fit rather than a derived law.
  • Piecewise linear compression prediction function = not reported
    Section IV-B says the server constructs a piecewise linear function from a minimal set of public training data to predict the best {rho,E} for a target compression ratio beta; this fitted mapping drives the compression rates in the delay optimization.
assumptions (5)
  • domain assumption LoRA adapters maintain task accuracy while updating only a small fraction of parameters
    Invoked throughout Sections III-B and IV; performance of SFT depends on LoRA being sufficient for fine-tuning, which is standard but not proven for the tested tasks.
  • domain assumption Top-K sparsification plus stochastic quantization plus Golomb encoding preserves accuracy within a bounded degradation for the tested models
    Section IV-B builds the compression scheme; Section VIII verifies empirically for ViT-base but no formal bound is provided.
  • ad hoc to paper The fitted third-order accuracy function A(rho,E) is a valid surrogate for true accuracy over the optimization domain
    Section IV-C and Fig. 3 use this fitted function; the accuracy constraint (27b) is exactly this fit, so errors in the fit directly change the claimed feasible region.
  • ad hoc to paper The server-side model can run all devices' LoRA forward/backward passes in parallel, so round delay is the max over devices
    Eq. (19) takes max over per-device delays, but Section IV-A states the server performs LoRA FP sequentially with one server-side model; if server work is serial, delay should be a sum.
  • domain assumption Pre-trained model weights are frozen and only LoRA adapters are updated
    Section III-B defines the LoRA update Theta + DeltaTheta; this is the standard PEFT assumption, but it is not verified against full fine-tuning in the experiments.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Split Fine-Tuning for Large Language Models in Wireless Networks." pith.science (2026). https://pith.science/paper/NTYYTOIT

@misc{pith2026250109237,
  author       = {Pith},
  title        = {Pith review of: Split Fine-Tuning for Large Language Models in Wireless Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NTYYTOIT}},
  note         = {Machine review of arXiv:2501.09237}
}
read the original abstract

Fine-tuning is the process of adapting the pre-trained large language models (LLMs) for downstream tasks. Due to substantial parameters, fine-tuning LLMs on mobile devices demands considerable memory resources, and suffers from high communication overhead and long fine-tuning delay. In this paper, we propose an efficient LLM fine-tuning scheme in wireless networks, named Split Fine-Tuning (SFT), which can accommodate LLM fine-tuning on mobile devices. Specifically, an LLM is split into a server-side part on the edge server and a device-side part on the mobile device to satisfy the device-side memory constraint. All devices share a server-side model and perform parallel fine-tuning to reduce fine-tuning delay. In addition, to reduce significant communication overhead incurred by data exchange between devices and the edge server, we propose a data compression scheme by jointly leveraging sparsification, stochastic quantization, and lossless encoding methods. Furthermore, we formulate a fine-tuning delay minimization problem under accuracy and memory constraints, taking device heterogeneity and channel dynamics into account. To solve the problem, the nonlinear mixed-integer problem is decoupled into two subproblems in different timescales. The two-timescale resource management algorithm is proposed to jointly optimize the compression rate and transformer block allocation in the large timescale using the augmented Lagrangian method, and determine spectrum resource allocation in the small timescale via sequential quadratic programming. Extensive simulation results demonstrate that the proposed scheme can reduce the fine-tuning delay by up to 80.2% and communication overhead by 93.6% compared to state-of-the-art benchmarks, while satisfying device-side memory and model accuracy constraints.

Figures

Figures reproduced from arXiv: 2501.09237 by the authors.

Figure 1
Figure 1. (a) In the SFT frame, devices are trained parallelly with a shared server-side pre-trained model and multiple LoRAs [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Transmission compression scheme. and device-side LoRA adapters using the FedAvg method. The aggregation process can be mathematically represented as ∆Θs n (t + 1, 1) = X N n=1 Dn D ∆Θs n (t, K), ∀n ∈ N , (7) and ∆Θu n (t + 1, 1) = X N n=1 Dn D ∆Θu n (t, K), ∀n ∈ N . (8) The aggregated process will involve a new fine-tuning round until the target accuracy is reached. B. Compression Scheme In the LLMs, the transmissio… view at source ↗
Figure 4
Figure 4. Fine-tuning delay in each round. where fn is the GPU frequency of device n, C U n is the number of cores of the GPU at device n, and DU n represents the number of FLOPs executed in a single core cycle of the GPU. 3) Immediate activation transmission (IT) latency: Each device transmits the immediate activation from the l-th trans￾former block to the server. Let ΨA denote the size of the output, which matches the size… view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: Fine-tuning performance comparison among different schemes. [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Memory consumption on a device with respect to [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Impact of sparsity and bit width on accuracy. [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Communication overhead in different fine-tuning [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 10
Figure 10. Figure 10: Delay performance of different fine-tuning schemes. [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 36 canonical work pages

  1. [1]

    Efficient federated learning for modern NLP,

    D. Cai, Y . Wu, S. Wang, F. X. Lin, and M. Xu, “Efficient federated learning for modern NLP,” in Proc. ACM Mobicom , 2023, pp. 1–16

    work page 2023

  2. [23]

    Efficient federated learning for modern NLP,

    D. Cai, Y . Wu, S. Wang, F. X. Lin, and M. Xu, “Efficient federated learning for modern NLP,” in Proc. ACM MobiCom , 2023, pp. 1–16

    work page 2023

  3. [2]

    DeViT: Decomposing vision transformers for collaborative inference in edge devices,

    G. Xu, Z. Hao, Y . Luo, H. Hu, J. An, and S. Mao, “DeViT: Decomposing vision transformers for collaborative inference in edge devices,” IEEE Trans. Mobile Comput. , vol. 23, no. 5, pp. 5917–5932, 2024

    work page 2024

  4. [3]

    Dual vision transformer,

    T. Yao, Y . Li, Y . Pan, Y . Wang, X.-P. Zhang, and T. Mei, “Dual vision transformer,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 45, no. 9, pp. 10 870–10 882, 2023

    work page 2023

  5. [4]

    Federatedscope-LLM: A comprehensive package for fine-tuning large language models in federated learning,

    W. Kuang, B. Qian, Z. Li, D. Chen, D. Gao, X. Pan, Y . Xie, Y . Li, B. Ding, and J. Zhou, “Federatedscope-LLM: A comprehensive package for fine-tuning large language models in federated learning,” in Proc. ACM SIGKDD, 2024, pp. 5260–5271

    work page 2024

  6. [5]

    Holistic network virtualization and pervasive network intelligence for 6G,

    X. Shen, J. Gao, W. Wu, M. Li, C. Zhou, and W. Zhuang, “Holistic network virtualization and pervasive network intelligence for 6G,” IEEE Commun. Surveys Tuts., vol. 24, no. 1, pp. 1–30, 2022

    work page 2022

  7. [6]

    WirelessLLM: Empowering large language models towards wireless intelligence,

    J. Shao, J. Tong, Q. Wu, W. Guo, Z. Li, Z. Lin, and J. Zhang, “WirelessLLM: Empowering large language models towards wireless intelligence,” IEEE J. Commun. Netw. , vol. 9, no. 2, pp. 99–112, 2024

    work page 2024

  8. [7]

    LLM-powered synthetic environments for self- driving scenarios,

    O. A. M. Adekanye, “LLM-powered synthetic environments for self- driving scenarios,” in Proc. AAAI, 2024, pp. 23 721–23 723

    work page 2024

Show all 42 references
  1. [8]

    Prefix-tuning: Optimizing continuous prompts for generation,

    X. L. Li and P. Liang, “Prefix-tuning: Optimizing continuous prompts for generation,” in Proc. ACL IJCNLP, 2021, pp. 4582–4597

    2021

  2. [9]

    AI-assisted network-slicing based next-generation wireless networks,

    X. Shen, J. Gao, W. Wu, K. Lyu, M. Li, W. Zhuang, X. Li, and J. Rao, “AI-assisted network-slicing based next-generation wireless networks,” IEEE Open J. Veh. Technol , vol. 1, pp. 45–66, 2020

    2020

  3. [10]

    Efficient and privacy-preserving feature importance-based vertical federated learning,

    A. Li, J. Huang, J. Jia, H. Peng, L. Zhang, L. A. Tuan, H. Yu, and X.-Y . Li, “Efficient and privacy-preserving feature importance-based vertical federated learning,” IEEE Trans. Mobile Comput. , vol. 23, no. 6, pp. 7238–7255, 2024

    2024

  4. [11]

    FL-TAC: Enhanced fine-tuning in federated learning via low-rank, task-specific adapter clustering,

    S. Ping, Y . Mao, Y . Liu, X. Zhang, and W. Ding, “FL-TAC: Enhanced fine-tuning in federated learning via low-rank, task-specific adapter clustering,” in Proc. ACM ICLR , 2024, pp. 1–6

    2024

  5. [12]

    AI-native network slicing for 6G networks,

    W. Wu, C. Zhou, M. Li, H. Wu, H. Zhou, N. Zhang, X. Shen, and W. Zhuang, “AI-native network slicing for 6G networks,” IEEE Wireless Commun., vol. 29, no. 1, pp. 96–103, 2022

    2022

  6. [13]

    Unstructured pruning and low rank factori- sation of self-supervised pre-trained speech models,

    H. Wang and W.-Q. Zhang, “Unstructured pruning and low rank factori- sation of self-supervised pre-trained speech models,” IEEE J. Sel. Topics Signal Process., pp. 1–14, DOI: 10.1109/JSTSP.2024.3433616, 2024

    2024

  7. [14]

    Digital twin based user-centric resource management for multicast short video streaming,

    X. Huang, W. Wu, S. Hu, M. Li, C. Zhou, and X. Shen, “Digital twin based user-centric resource management for multicast short video streaming,” IEEE J. Sel. Topics Signal Process. , vol. 18, no. 1, pp. 50– 65, 2024

    2024

  8. [15]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

    2010 arXiv

  9. [16]

    Parameter-efficient transfer learning for NLP,

    N. Houlsby, A. Giurgiu, S. Jastrzebski, B. Morrone, Q. De Laroussilhe, A. Gesmundo, M. Attariyan, and S. Gelly, “Parameter-efficient transfer learning for NLP,” in Proc. PMLR, 2019, pp. 2790–2799

    2019

  10. [17]

    The power of scale for parameter-efficient prompt tuning,

    B. Lester, R. Al-Rfou, and N. Constant, “The power of scale for parameter-efficient prompt tuning,” in Proc. ACL EMNLP , 2021, pp. 3045–3059

    2021

  11. [18]

    Towards a unified view of parameter-efficient transfer learning,

    J. He, C. Zhou, X. Ma, T. Berg-Kirkpatrick, and G. Neubig, “Towards a unified view of parameter-efficient transfer learning,” in Proc. ACM ICLR, 2022, pp. 1–15

    2022

  12. [19]

    FedMes: Speeding up federated learning with multiple edge servers,

    D. Han, M. Choi, J. Park, and J. Moon, “FedMes: Speeding up federated learning with multiple edge servers,” IEEE J. Sel. Areas Commun. , vol. 39, no. 12, pp. 3870–3885, 2021

    2021

  13. [20]

    DetFed: Dynamic resource scheduling for deterministic federated learning over time-sensitive networks,

    D. Yang, W. Zhang, Q. Ye, C. Zhang, N. Zhang, C. Huang, H. Zhang, and X. Shen, “DetFed: Dynamic resource scheduling for deterministic federated learning over time-sensitive networks,” IEEE Transactions on Mobile Computing, vol. 23, no. 5, pp. 5162–5178, 2024

    2024

  14. [21]

    PromptFL: Let federated participants cooperatively learn prompts instead of models – federated learning in age of foundation model,

    T. Guo, S. Guo, J. Wang, X. Tang, and W. Xu, “PromptFL: Let federated participants cooperatively learn prompts instead of models – federated learning in age of foundation model,” IEEE Trans. Mobile Comput. , vol. 23, no. 5, pp. 5179–5194, 2024

    2024

  15. [22]

    Low-parameter federated learning with large language models,

    J. Jiang, H. Jiang, Y . Ma, X. Liu, and C. Fan, “Low-parameter federated learning with large language models,” in Proc. Springer WISA, 2024, pp. 319–330

    2024

  16. [24]

    En- semble distillation based adaptive quantization for supporting federated learning in wireless networks,

    Y .-J. Liu, G. Feng, D. Niyato, S. Qin, J. Zhou, X. Li, and X. Xu, “En- semble distillation based adaptive quantization for supporting federated learning in wireless networks,” IEEE Trans. Wireless Commun., vol. 22, no. 6, pp. 4013–4027, 2023

    2023

  17. [25]

    Sparse training for federated learning with regularized error correction,

    R. Greidi and K. Cohen, “Sparse training for federated learning with regularized error correction,” IEEE J. Sel. Topics Signal Process. , pp. 1–16, 2024

    2024

  18. [26]

    Split federated learning: Speed up model training in resource-limited wireless networks,

    S. Zhang, W. Wu, P. Hu, S. Li, and N. Zhang, “Split federated learning: Speed up model training in resource-limited wireless networks,” in Proc. IEEE ICDCS, 2023, pp. 985–986

    2023

  19. [27]

    Split learning over wireless networks: Parallel design and resource management,

    W. Wu, M. Li, K. Qu, C. Zhou, X. Shen, W. Zhuang, X. Li, and W. Shi, “Split learning over wireless networks: Parallel design and resource management,” IEEE J. Sel. Areas Commun. , vol. 41, no. 4, pp. 1051– 1066, 2023

    2023

  20. [28]

    Accelerating split federated learning over wireless communication networks,

    C. Xu, J. Li, Y . Liu, Y . Ling, and M. Wen, “Accelerating split federated learning over wireless communication networks,” IEEE Trans. Wireless Commun., vol. 23, no. 6, pp. 5587–5599, 2024

    2024

  21. [29]

    Accelerating federated learning with data and model parallelism in edge computing,

    Y . Liao, Y . Xu, H. Xu, Z. Yao, L. Wang, and C. Qiao, “Accelerating federated learning with data and model parallelism in edge computing,” IEEE/ACM Trans. Netw., vol. 32, no. 1, pp. 904–918, 2024

    2024

  22. [30]

    ParallelSFL: A novel split federated learning framework tackling heterogeneity issues,

    Y . Liao, Y . Xu, H. Xu, Z. Yao, L. Huang, and C. Qiao, “ParallelSFL: A novel split federated learning framework tackling heterogeneity issues,” in Proc. ACM MobiCom , 2024, pp. 845–860

    2024

  23. [31]

    Device-edge cooperative fine-tuning of foundation models as a 6G service,

    H. Wu, X. Chen, and K. Huang, “Device-edge cooperative fine-tuning of foundation models as a 6G service,” IEEE Wireless Commun. , vol. 31, no. 3, pp. 60–67, 2024

    2024

  24. [32]

    Pluto and Charon: A time and memory efficient collaborative edge AI framework for personal LLMs fine-tuning,

    B. Ouyang, S. Ye, L. Zeng, T. Qian, J. Li, and X. Chen, “Pluto and Charon: A time and memory efficient collaborative edge AI framework for personal LLMs fine-tuning,” in Proc. ACM ICPP, 2024, pp. 762–771

    2024

  25. [33]

    Improving LoRA in privacy- preserving federated learning,

    Y . Sun, Z. Li, Y . Li, and B. Ding, “Improving LoRA in privacy- preserving federated learning,” arXiv preprint arXiv:2403.12313 , 2024

    2024 arXiv

  26. [34]

    Fast: Fidelity-adjustable semantic transmission over heterogeneous wireless networks,

    P. Li, G. Cheng, J. Kang, R. Yu, L. Qian, Y . Wu, and D. Niyato, “Fast: Fidelity-adjustable semantic transmission over heterogeneous wireless networks,” in Proc. IEEE ICC , 2023, pp. 4689–4694

    2023

  27. [35]

    Robust and communication-efficient federated learning from non-IID data,

    F. Sattler, S. Wiedemann, K.-R. M ¨uller, and W. Samek, “Robust and communication-efficient federated learning from non-IID data,” IEEE Trans. Neural Netw. Learn. Syst. , vol. 31, no. 9, pp. 3400–3413, 2019

    2019

  28. [36]

    Run-length encodings (corresp.),

    S. Golomb, “Run-length encodings (corresp.),” IEEE Trans. Inf. Theory, vol. 12, no. 3, pp. 399–401, 1966

    1966

  29. [37]

    Adap- tive digital twin-assisted 3C management for QoE-driven MSVS: A GAI-based DRL approach,

    X. Huang, X. Qin, M. Li, C. Huang, and X. Shen, “Adap- tive digital twin-assisted 3C management for QoE-driven MSVS: A GAI-based DRL approach,” IEEE Trans. Cogn. Commun. Netw , DOI: 10.1109/TCCN.2024.3516046, 2024

    2024

  30. [38]

    Resource allocation for stable LLM training in mobile edge computing,

    C. Liu and J. Zhao, “Resource allocation for stable LLM training in mobile edge computing,” in Proc. ACM MobiHoc , 2024, pp. 81–90

    2024

  31. [39]

    Learning multiple layers of features from tiny images,

    A. Krizhevsky, G. Hinton, and H. Geoffrey, “Learning multiple layers of features from tiny images,” 2009

    2009

  32. [40]

    FedGKD: Toward heterogeneous federated learning via global knowledge distillation,

    D. Yao, W. Pan, Y . Dai, Y . Wan, X. Ding, C. Yu, H. Jin, Z. Xu, and L. Sun, “FedGKD: Toward heterogeneous federated learning via global knowledge distillation,” IEEE Trans. Comput., vol. 73, no. 1, pp. 3–17, 2024

    2024

  33. [41]

    Communication-efficient learning of deep networks from decentralized data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” in Proc. AAAI, 2017, pp. 1273–1282

    2017

  34. [42]

    Distributed learning of deep neural network over multiple agents,

    O. Gupta and R. Raskar, “Distributed learning of deep neural network over multiple agents,” J. Net. Comp. Appl. , vol. 116, pp. 1–8, 2018

    2018

Pith tools

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