Pith. sign in

REVIEW 5 major objections 6 minor 66 references

Transforming Decoder-Only Transformers for Accurate WiFi-Telemetry Based Indoor Localization

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a decoder-only large language model fine-tuned with LoRA can map raw WiFi telemetry tokens to distance, reaching sub-meter RSSI/FTM errors and centimeter-level CSI errors without handcrafted signal processing or…

desk verdict Plausible first use of LoRA-finetuned LLaMA for WiFi telemetry distance regression, but the headline CSI results are leaky and the paper lacks baselines. read the letter →

arxiv 2505.15835 v1 pith:3CNYQ42E submitted 2025-05-16 cs.NI cs.AI

Nayan Sanjay Bhatia , Katia Obraczka This is my paper · ORCID
classification cs.NIcs.AI
keywords indoorlocalizationWiFitelemetrydecoder-onlytransformerLLMregressionCSIFTMRSSILoRAfine-tuning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that a decoder-only large language model, fine-tuned with LoRA on raw WiFi telemetry serialized as text, can estimate transmitter-receiver distances accurately enough for indoor localization. The reported numbers are sub-meter median errors for RSSI and FTM measurements in corridor, office, and lecture-theatre environments, and centimeter-level mean absolute errors for CSI in a multipath hallway. The point is that one model, without handcrafted signal processing, calibration, data scaling, or a fixed input schema, can handle telemetry from different vendors and environments. If true, this makes LLM-based regression a practical alternative to specialized fingerprinting and ranging pipelines, especially on consumer hardware.

What carries the argument

The load-bearing object is the prompt-based next-token regression interface: each WiFi packet is formatted as a JSON-style instruction with special delimiters, the telemetry values are plain tokens, and the model must emit a numeric answer in the form '{answer}m' as the final token. LoRA adapters freeze the base weights and train only a small fraction of parameters, while auxiliary MAE and MSE monitors track convergence without changing the underlying cross-entropy objective. Greedy decoding with temperature zero and a fixed seed makes the output deterministic. This machinery lets a language model treat distance as the next word, removing the need for separate encoders or handcrafted features.

What would settle it

Re-run the CSI experiment with a temporal split: train on the first minutes of each position's recording and test on the last minutes, or hold out entire positions. Centimeter-level errors that survive this split would confirm spatial generalization, while errors jumping to meter scale would show that the current accuracy came from memorization.

Watch

Extended reading notes

Core claim

WiFiGPT is presented as a repurposing of a decoder-only LLM for indoor localization. The central discovery is that a causal language model can be re-tooled as a distance regressor by framing distance as the next token: after supervised fine-tuning with LoRA on prompt-formatted telemetry, the model outputs a single numeric value in meters. On the custom CSI hallway dataset, LLaMA-3.2-3B reaches MAE 0.069 m and MSE 0.168 m; on the public RSSI/FTM dataset, median errors stay near or below 1 m across all three environments and model sizes. The paper also shows that the model performs implicit data imputation, since FTM-only and RSSI-only ablation runs still produce usable estimates, with fused features performing best.

Load-bearing premise

The weakest assumption is that randomly splitting the roughly five-minute recordings taken at each fixed position into train and test sets measures spatial generalization rather than the model recognizing near-identical, temporally adjacent samples from the same spot.

Editorial extensions

If this is right

  • A single prompt-template system can replace separate preprocessing pipelines for RSSI, FTM, and CSI, since all telemetry types are treated as serialized text.
  • Deployment on resource-constrained devices becomes plausible because LoRA updates only a small fraction of weights and inference is a single autoregressive pass.
  • The system degrades gracefully under missing features: ablation results show that RSSI-only and FTM-only inputs still yield useful distance estimates.
  • Practitioners can smooth tail-heavy single-packet errors with a sliding window of 5 to 10 packets at a cost of only a few milliseconds of additional delay.

Reading between the lines

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

  • Editorial inference: the CSI few-shot evaluation is the weakest link, because the random sample-wise split of five-minute recordings at fixed positions likely lets temporally adjacent near-duplicates leak across the train/test boundary, making the centimeter-level numbers an overstatement of true spatial generalization.
  • Editorial inference: a stronger claim would test cross-environment transfer with a single adapter trained on multiple rooms and evaluated on an unseen layout, since the paper currently fine-tunes separately for each environment.
  • Editorial inference: the same serialization trick could extend to other single-antenna sensing modalities such as BLE, UWB, or inertial data, because nothing in the architecture depends on WiFi-specific features.
  • Editorial inference: prompt sensitivity, which the paper lists as a limitation, can be probed directly by paraphrasing the instruction while keeping the telemetry fixed; predicted distances should not change if the model learned geometry rather than surface text.
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

5 major / 6 minor

Summary. The paper proposes WiFiGPT, a system that fine-tunes decoder-only LLaMA-3 models with LoRA adapters to regress transmitter-receiver distance from raw WiFi telemetry (CSI, FTM, RSSI) represented as token sequences. Telemetry arrays are embedded in a prompt template and the model is trained with next-token prediction to output a distance in meters; inference uses greedy decoding for deterministic outputs. Experiments are reported on a custom ESP32 CSI hallway dataset and a public FTM/RSSI dataset covering lecture theater, office, and corridor environments. The headline results are sub-meter MAE for FTM/RSSI (around 0.9-1.3 m) and centimeter-level MAE for CSI (about 6.9 cm), from which the authors conclude that LLM-based localization can match or surpass specialized techniques without handcrafted signal processing or calibration.

Significance. If the reported results held, the paper would support a genuinely unified, schema-less LLM-based approach to WiFi localization across heterogeneous telemetry types and environments. The deterministic greedy-decoding inference, the LoRA-based fine-tuning for efficiency, and the direct use of raw telemetry tokens are sensible design choices, and the use of a public FTM/RSSI dataset is a strength. However, the empirical evidence as presented does not establish the central claims: the CSI evaluation leaks temporally correlated samples between train and test, no baseline algorithms are run on the same data, and the 'without calibration' claim is contradicted by the per-environment fine-tuning protocol. The idea is worth exploring, but the manuscript in its current form does not provide a valid demonstration of the stated contribution.

major comments (5)
  1. [Section V.A, Table III] The CSI evaluation splits a continuous ~5-minute recording made at each fixed 1-meter position randomly by sample (10% train, 10% validation, 80% test). Adjacent CSI packets recorded at the same stationary location are near-duplicates of one another, so the model can effectively memorize input patterns rather than learn a distance mapping. The reported 6.9 cm MAE on 51,500 test samples is therefore not evidence of generalization to unseen positions or environments. The evaluation must be redone with a temporal split (e.g., training on an early segment of each recording and testing on a later segment) or a leave-one-position-out split, with per-position errors reported.
  2. [Section III.C.2 and Abstract] The abstract and conclusion claim the system operates 'without handcrafted signal processing or calibration,' but Section III.C.2 states that the model is trained separately for each dataset and environment. No cross-environment or cross-session experiment is reported, so the system has not been shown to be calibration-free; it is fitted per deployment. Demonstrating the 'no calibration' claim would require evaluating an adapter trained in one environment on another environment or on unseen hardware without additional fine-tuning.
  3. [Section V, Tables III and IV] No baseline algorithm is evaluated on the same data and splits in the experiments. The statement that WiFiGPT 'matches and often surpasses' state-of-the-art methods is therefore unsupported; it appears to rely on comparisons to previously published numbers obtained with different datasets, protocols, and metrics. At minimum, the authors should report KNN, LGBM, and a simple linear or MLP regressor on the same telemetry features and splits, and for CSI should include a classical method such as SpotFi-style processing or a standard fingerprinting approach.
  4. [References [47]-[49], Tables III and IV] The 8B model is labeled 'LLaMA-3.1-8B' in the results tables, but reference [49] is an MLX conversion of DeepHermes-3-Llama-3-8B-Preview, a chat-tuned variant rather than the base LLaMA-3.1-8B model. Comparing this model with LLaMA-3.2-1B and LLaMA-3.2-3B confounds model size with model family and instruction tuning, so the claimed analysis of how accuracy varies with LLaMA model size is not valid as presented.
  5. [Section IV.B, Table III] Misaligned or unparseable model outputs are skipped in the error metrics rather than counted as errors. Table III reports six misalignments for the 3B model and two for the 8B model; if these outputs correspond to degenerate or high-error predictions, the reported MAE/MSE values are optimistically biased. The paper should report the metrics with misaligned outputs included (e.g., counted at the maximum possible error) or otherwise demonstrate that their exclusion does not affect the conclusions.
minor comments (6)
  1. [Section III.C and Section IV.B] The text says in Section III.C that MAE and MSE are used as custom loss metrics for training, but Section IV.B later states they are auxiliary monitoring tools and are not used for gradient updates; this inconsistency should be resolved and the training objective stated precisely.
  2. [References [47]-[49]] The references for the base models point to MLX-community checkpoint conversions rather than the original LLaMA-3 model releases; the paper should cite the original model descriptions and state the exact base checkpoints used, especially for the 8B variant.
  3. [Section V.A] The CSI experiment is called 'Few-Shot Learning,' but 10% of 64,375 samples is roughly 6,400 training samples, which is not a few-shot regime; the labeling should be corrected or the training set size stated explicitly.
  4. [Section VI] The text says 'as shown in Figure 2a/2b' when referring to trilateration geometry, but the cited figure is Figure 3; the cross-reference should be fixed.
  5. [Table IV] The high-percentile column is labeled '100th Iteration' and contains inconsistently formatted entries (e.g., '10.0100' versus '10.05'); the column label and numeric precision should be made uniform.
  6. [Throughout] There are several typos, including 'RToF' in Section II.A, 'course grained' in Section V.C.2, and 'eletrocardiogram' in Section II.B; these should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: WiFiGPT is an empirical fine-tuning study whose claims, even where overclaimed, are not derived from their own inputs.

full rationale

WiFiGPT is an empirical systems paper: a pretrained LLaMA model is LoRA fine-tuned on telemetry-to-distance pairs and evaluated with MAE/MSE; there is no mathematical derivation whose conclusion is assumed in its premises. The nearest candidate—the abstract's 'without handcrafted signal processing or calibration' claim—is contradicted by the paper's own statement 'We train the LLM model separately for each dataset and environment' (Section III.C) and by the limitation 'Our system's performance may degrade in a new environment without any prior training or Dataset of new sensors' (Section VI), but an overclaim is not circular reasoning. The CSI few-shot evaluation's random per-sample split of roughly 5-minute stationary recordings (Section V.A) may leak temporally correlated near-duplicates into the test set and inflate the reported centimeter-level errors, yet this is a statistical validity threat, not an equation reducing to its input. There are no load-bearing self-citations: the dataset and method references are external. Accordingly, no circular step meets the required evidentiary bar; the honest finding is no significant circularity.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central accuracy numbers rest on hand-chosen prompt formatting and LoRA settings, on the untested claim that text-pretrained tokenization preserves spatial information in telemetry, and on the assumption that the random split avoids temporal leakage. No new physical entities or derived constants are introduced.

free parameters (2)
  • LoRA fine-tuning hyperparameters = batch size 2, LoRA layers 16, learning rate 2e-4; LoRA rank and alpha not reported
    Chosen by hand to fit a 32GB M1 Max; they directly affect the reported accuracy and are not derived from data.
  • Prompt template and delimiter tokens = custom JSONL instruction format with <|begin_of_text|>, <|start_header_id|>, and related tokens
    The paper notes LLMs are sensitive to prompt phrasing; this ad hoc template determines the token sequence the model sees and is not justified by any experiment.
assumptions (3)
  • domain assumption Random sample-wise splitting of continuous telemetry recordings gives unbiased test performance.
    Section IV.A.1 collects about 5 minutes of CSI at fixed 1-meter positions; Section V.A splits samples randomly, which can place temporally correlated samples from the same position into both train and test.
  • domain assumption A text-pretrained LLaMA model can learn distance regression from tokenized numeric telemetry via cross-entropy next-token training.
    This is the core method assumption; the paper provides no comparison against a randomly initialized transformer to isolate the contribution of pretraining.
  • domain assumption Per-environment fine-tuning is a fair realization of a 'no calibration' deployment claim.
    Section III.C.2 trains a separate model for each dataset and environment, so the system is calibrated to each environment rather than deployed zero-shot.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Transforming Decoder-Only Transformers for Accurate WiFi-Telemetry Based Indoor Localization." pith.science (2026). https://pith.science/paper/3CNYQ42E

@misc{pith2026250515835,
  author       = {Pith},
  title        = {Pith review of: Transforming Decoder-Only Transformers for Accurate WiFi-Telemetry Based Indoor Localization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3CNYQ42E}},
  note         = {Machine review of arXiv:2505.15835}
}
read the original abstract

Wireless Fidelity (WiFi) based indoor positioning is a widely researched area for determining the position of devices within a wireless network. Accurate indoor location has numerous applications, such as asset tracking and indoor navigation. Despite advances in WiFi localization techniques -- in particular approaches that leverage WiFi telemetry -- their adoption in practice remains limited due to several factors including environmental changes that cause signal fading, multipath effects, interference, which, in turn, impact positioning accuracy. In addition, telemetry data differs depending on the WiFi device vendor, offering distinct features and formats; use case requirements can also vary widely. Currently, there is no unified model to handle all these variations effectively. In this paper, we present WiFiGPT, a Generative Pretrained Transformer (GPT) based system that is able to handle these variations while achieving high localization accuracy. Our experiments with WiFiGPT demonstrate that GPTs, in particular Large Language Models (LLMs), can effectively capture subtle spatial patterns in noisy wireless telemetry, making them reliable regressors. Compared to existing state-of-the-art methods, our method matches and often surpasses conventional approaches for multiple types of telemetry. Achieving sub-meter accuracy for RSSI and FTM and centimeter-level precision for CSI demonstrates the potential of LLM-based localisation to outperform specialized techniques, all without handcrafted signal processing or calibration.

Figures

Figures reproduced from arXiv: 2505.15835 by the authors.

Figure 1
Figure 1. WiFiGPT: System Flow generalize distance predictions across different environments and hardware setups. Importantly, by learning directly from raw telemetry data, our model remains strong and works well with any device. It adapts reliably to different indoor conditions without relying on specific hardware or signal types. B. Data Preparation Layer A token is a chunk of character sets that the model reads or generate… view at source ↗
Figure 2
Figure 2. CDF of Localization Error B. FTM and RSSI The results (Table IV) show a clear trend in which the environmental conditions (LOS and NLOS) and the com￾plexity of the model affect the positioning accuracy. Across all setups, LLMs exhibit strong regression capabilities, with median localization errors typically near or below 1 meter. In the fully NLOS Corridor environment, the 3B model has the highest accuracy, with an … view at source ↗
Figure 3
Figure 3. Trilateration geometry: (a) Ideal Scenario (b) Real Scenario [63] [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

66 extracted references · 34 canonical work pages

  1. [49]

    Deephermes-3-llama-3-8b-preview-bf16,

    ——, “Deephermes-3-llama-3-8b-preview-bf16,” https://huggingface.co/mlx-community/DeepHermes-3-Llama-3- 8B-Preview-bf16, 2025, converted to MLX format from NousResearch/DeepHermes-3-Llama-3-8B-Preview using mlx-lm version 0.21.1. [Online]. Available: https://huggingface.co/mlx- community/DeepHermes-3-Llama-3-8B-Preview-bf16

    work page 2025

  2. [1]

    Machine learning based indoor localization using wi-fi rssi fingerprints: An overview,

    N. Singh, S. Choe, and R. Punmiya, “Machine learning based indoor localization using wi-fi rssi fingerprints: An overview,”IEEE Access, vol. 9, pp. 127 150–127 174, 2021

    work page 2021

  3. [2]

    A survey on indoor positioning systems for iot-based applications,

    P. S. Farahsari, A. Farahzadi, J. Rezazadeh, and A. Bagheri, “A survey on indoor positioning systems for iot-based applications,”IEEE Internet of Things Journal, vol. 9, no. 10, pp. 7680–7699, 2022

    2022

  4. [3]

    A review of pedestrian indoor positioning systems for mass market applications,

    A. Correa, M. Barcelo, A. Morell, and J. L. Vicario, “A review of pedestrian indoor positioning systems for mass market applications,” Sensors, vol. 17, no. 8, p. 1927, 2017

    work page 1927

  5. [4]

    A survey of problems and approaches in wireless-based indoor positioning,

    F. Dwiyasa and M.-H. Lim, “A survey of problems and approaches in wireless-based indoor positioning,” in2016 International conference on indoor positioning and indoor navigation (IPIN). IEEE, 2016, pp. 1–7

    work page 2016

  6. [5]

    A survey of active and passive in- door localisation systems,

    G. Deak, K. Curran, and J. Condell, “A survey of active and passive in- door localisation systems,”Computer Communications, vol. 35, no. 16, pp. 1939–1954, 2012

    work page 1939

  7. [6]

    A survey on ubiquitous wifi-based indoor localization system for smartphone users from implementation per- spectives,

    P. Roy and C. Chowdhury, “A survey on ubiquitous wifi-based indoor localization system for smartphone users from implementation per- spectives,”CCF Transactions on Pervasive Computing and Interaction, vol. 4, no. 3, pp. 298–318, 2022

    work page 2022

  8. [7]

    Wi-fi-based real-time uav localization: A comparative analysis between rssi-based and ftm-based approaches,

    E. Pagliari, L. Davoli, and G. Ferrari, “Wi-fi-based real-time uav localization: A comparative analysis between rssi-based and ftm-based approaches,”IEEE Transactions on Aerospace and Electronic Systems, vol. 60, no. 6, pp. 8757–8778, 2024

    work page 2024

Show all 66 references
  1. [8]

    A survey of indoor localization systems and technologies,

    F. Zafari, A. Gkelias, and K. K. Leung, “A survey of indoor localization systems and technologies,”IEEE Communications Surveys & Tutorials, vol. 21, no. 3, pp. 2568–2599, 2019

    2019

  2. [9]

    Exposing the csi: A systematic investigation of csi-based wi-fi sensing capabilities and limitations,

    M. Cominelli, F. Gringoli, and F. Restuccia, “Exposing the csi: A systematic investigation of csi-based wi-fi sensing capabilities and limitations,” in2023 IEEE International Conference on Pervasive Computing and Communications (PerCom). IEEE, 2023, pp. 81–90

    2023

  3. [10]

    A survey of machine learning for indoor positioning,

    A. Nessa, B. Adhikari, F. Hussain, and X. N. Fernando, “A survey of machine learning for indoor positioning,”IEEE access, vol. 8, pp. 214 945–214 965, 2020

    2020

  4. [11]

    An empirical comparison of techniques for handling incom- plete data using decision trees,

    B. Twala, “An empirical comparison of techniques for handling incom- plete data using decision trees,”Applied Artificial Intelligence, vol. 23, no. 5, pp. 373–405, 2009

    2009

  5. [12]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. u. Kaiser, and I. Polosukhin, “Attention is all you need,” in Advances in Neural Information Processing Systems, I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garne...

    2017

  6. [13]

    How multilingual is multilin- gual bert?

    T. Pires, E. Schlinger, and D. Garrette, “How multilingual is multilin- gual bert?”arXiv preprint arXiv:1906.01502, 2019

    1906 arXiv

  7. [14]

    Dolphingemma: How google ai is helping decode dolphin communica- tion,

    Google DeepMind, “Dolphingemma: How google ai is helping decode dolphin communica- tion,” https://blog.google/technology/ai/dolphingemma/, Febru- ary 2024, accessed: 2025-05-03. [Online]. Available: https://blog.google/technology/ai/dolphingemma/

    2024

  8. [15]

    How can time series analysis bene- fit from multiple modalities? a survey and outlook,

    H. Liu, H. Kamarthi, Z. Zhao, S. Xu, S. Wang, Q. Wen, T. Hartvigsen, F. Wang, and B. A. Prakash, “How can time series analysis bene- fit from multiple modalities? a survey and outlook,”arXiv preprint arXiv:2503.11835, 2025

    2025

  9. [16]

    Does chatgpt generate accessible code? investigating accessibility challenges in llm-generated source code,

    W. Aljedaani, A. Habib, A. Aljohani, M. Eler, and Y . Feng, “Does chatgpt generate accessible code? investigating accessibility challenges in llm-generated source code,” inProceedings of the 21st International Web for All Conference, ser. W4A ’24. New York, NY , USA: Associati...

    2024

  10. [17]

    Indoor ranging and localisation algorithm based on received signal strength indicator using statistic parameters for wireless sensor networks,

    S. Pagano, S. Peirani, and M. Valle, “Indoor ranging and localisation algorithm based on received signal strength indicator using statistic parameters for wireless sensor networks,”IET Wireless Sensor Systems, vol. 5, no. 5, pp. 243–249, 2015

    2015

  11. [18]

    Rssi based localization scheme in wireless sensor networks: A survey,

    H. P. Mistry and N. H. Mistry, “Rssi based localization scheme in wireless sensor networks: A survey,” in2015 Fifth International Conference on Advanced Computing & Communication Technologies. IEEE, 2015, pp. 647–652

    2015

  12. [19]

    Wireless rssi fingerprinting localization,

    S. Yiu, M. Dashti, H. Claussen, and F. Perez-Cruz, “Wireless rssi fingerprinting localization,”Signal Processing, vol. 131, pp. 235–244, 2017

    2017

  13. [20]

    Machine learning based indoor localization using wi-fi rssi fingerprints: An overview,

    N. Singh, S. Choe, and R. Punmiya, “Machine learning based indoor localization using wi-fi rssi fingerprints: An overview,”IEEE access, vol. 9, pp. 127 150–127 174, 2021

    2021

  14. [21]

    The microsoft indoor localization competition: Experiences and lessons learned,

    D. Lymberopoulos and J. Liu, “The microsoft indoor localization competition: Experiences and lessons learned,”IEEE Signal Processing Magazine, vol. 34, no. 5, pp. 125–140, 2017

    2017

  15. [22]

    Fine time measurement for the internet of things: A practical approach using esp32,

    V . Barral Vales, O. C. Fern´andez, T. Dom´ınguez-Bola˜no, C. J. Escudero, and J. A. Garc ´ıa-Naya, “Fine time measurement for the internet of things: A practical approach using esp32,”IEEE Internet of Things Journal, vol. 9, no. 19, pp. 18 305–18 318, 2022

    2022

  16. [23]

    Verification: Accuracy evaluation of wifi fine time measurements on an open platform,

    M. Ibrahim, H. Liu, M. Jawahar, V . Nguyen, M. Gruteser, R. Howard, B. Yu, and F. Bai, “Verification: Accuracy evaluation of wifi fine time measurements on an open platform,” inProceedings of the 24th annual international conference on mobile computing and networking, 2018, pp...

    2018

  17. [24]

    Comparison of 2.4 ghz wifi ftm-and rssi-based indoor posi- tioning methods in realistic scenarios,

    M. Bullmann, T. Fetzer, F. Ebner, M. Ebner, F. Deinzer, and M. Grze- gorzek, “Comparison of 2.4 ghz wifi ftm-and rssi-based indoor posi- tioning methods in realistic scenarios,”Sensors, vol. 20, no. 16, p. 4515, 2020

    2020

  18. [25]

    Accurate indoor localization with zero start-up cost,

    S. Kumar, S. Gil, D. Katabi, and D. Rus, “Accurate indoor localization with zero start-up cost,” inProceedings of the 20th annual international conference on Mobile computing and networking, 2014, pp. 483–494

    2014

  19. [26]

    {ArrayTrack}: A{Fine-Grained}indoor location system,

    J. Xiong and K. Jamieson, “{ArrayTrack}: A{Fine-Grained}indoor location system,” in10th USENIX Symposium on Networked Systems Design and Implementation (NSDI 13), 2013, pp. 71–84

    2013

  20. [27]

    Spotfi: Decimeter level localization using wifi,

    M. Kotaru, K. Joshi, D. Bharadia, and S. Katti, “Spotfi: Decimeter level localization using wifi,” inProceedings of the 2015 ACM Conference on Special Interest Group on Data Communication, ser. SIGCOMM ’15. New York, NY , USA: Association for Computing Machinery, 2015, p. 269–...

    2015

  21. [28]

    Enabling phased array signal processing for mobile wifi devices,

    K. Qian, C. Wu, Z. Yang, Z. Zhou, X. Wang, and Y . Liu, “Enabling phased array signal processing for mobile wifi devices,”IEEE Trans- actions on Mobile Computing, vol. 17, no. 8, pp. 1820–1833, 2017

    2017

  22. [29]

    Estimation of signal parameters via rotational invariance techniques-esprit,

    A. Paulraj, R. Roy, and T. Kailath, “Estimation of signal parameters via rotational invariance techniques-esprit,” inNineteeth Asilomar Confer- ence on Circuits, Systems and Computers, 1985.IEEE, 1985, pp. 83–89

    1985

  23. [30]

    Esp32 series datasheet,

    E. Systems, “Esp32 series datasheet,” https://www.espressif.com/en/products/socs/esp32, 2023, accessed: 2025-05-14

    2023

  24. [31]

    Raspberry pi documentation,

    R. P. Foundation, “Raspberry pi documentation,” https://www.raspberrypi.com/documentation/, 2023, accessed: 2025- 05-14

    2023

  25. [32]

    Are decoder-only language models better than encoder-only language models in understanding word meaning?

    M. Qorib, G. Moon, and H. T. Ng, “Are decoder-only language models better than encoder-only language models in understanding word meaning?” inFindings of the Association for Computational Linguistics ACL 2024, 2024, pp. 16 339–16 347

    2024

  26. [33]

    Bert: a review of applications in natural language processing and understanding,

    M. V . Koroteev, “Bert: a review of applications in natural language processing and understanding,”arXiv preprint arXiv:2103.11943, 2021

    2021 arXiv

  27. [34]

    Mining limited data sufficiently: A bert-inspired approach for csi time series application in wireless communication and sensing,

    Z. Zhao, F. Meng, H. Li, X. Li, and G. Zhu, “Mining limited data sufficiently: A bert-inspired approach for csi time series application in wireless communication and sensing,” 2024. [Online]. Available: https://arxiv.org/abs/2412.06861

    2024

  28. [35]

    Bert (bidirectional encoder representations from transformers) for missing data imputation in solar irradiance time series,

    L. B. Cesar, M.- ´A. Manso-Callejo, and C.-I. Cira, “Bert (bidirectional encoder representations from transformers) for missing data imputation in solar irradiance time series,”Engineering Proceedings, vol. 39, no. 1, p. 26, 2023

    2023

  29. [36]

    Gpt-4 technical report,

    O. AI, “Gpt-4 technical report,”arXiv preprint arXiv:2303.08774, 2023

    2023 arXiv

  30. [37]

    Llama: Open and efficient foundation language models,

    H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozi `ere, N. Goyal, E. Hambro, F. Azharet al., “Llama: Open and efficient foundation language models,”arXiv preprint arXiv:2302.13971, 2023

    2023 arXiv

  31. [38]

    Netllm: Adapting large language models for networking,

    D. Wu, X. Wang, Y . Qiao, Z. Wang, J. Jiang, S. Cui, and F. Wang, “Netllm: Adapting large language models for networking,” inProceedings of the ACM SIGCOMM 2024 Conference, ser. ACM SIGCOMM ’24. ACM, Aug. 2024, p. 661–678. [Online]. Available: http://dx.doi.org/10.1145/3651890.3672268

    2024

  32. [39]

    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,” 2024. [Online]. Available: https://arxiv.org/abs/2405.17053

    2024 arXiv

  33. [40]

    Large language models can be zero-shot anomaly detectors for time series?

    S. Alnegheimish, L. Nguyen, L. Berti-Equille, and K. Veeramachaneni, “Large language models can be zero-shot anomaly detectors for time series?” 2024. [Online]. Available: https://arxiv.org/abs/2405.14755

    2024 arXiv

  34. [41]

    Finding the missing data: A bert-inspired approach against package loss in wireless sensing,

    Z. Zhao, T. Chen, F. Meng, H. Li, X. Li, and G. Zhu, “Finding the missing data: A bert-inspired approach against package loss in wireless sensing,” inIEEE INFOCOM 2024 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS). IEEE, May 2024, p. 1–6. [Online]. Av...

    2024

  35. [42]

    Understanding localization by a tailored gpt,

    X. Zhao, G. Wang, Z. An, Q. Pan, and L. Yang, “Understanding localization by a tailored gpt,” inProceedings of the 22nd Annual International Conference on Mobile Systems, Applications and Services, ser. MOBISYS ’24. New York, NY , USA: Association for Computing Machinery, 2024...

    2024

  36. [43]

    From words to numbers: Your large language model is secretly a capable regressor when given in-context examples,

    R. Vacareanu, V .-A. Negru, V . Suciu, and M. Surdeanu, “From words to numbers: Your large language model is secretly a capable regressor when given in-context examples,”arXiv preprint arXiv:2404.07544, 2024

    2024 arXiv

  37. [44]

    Understanding llm embeddings for regression,

    E. Tang, B. Yang, and X. Song, “Understanding llm embeddings for regression,”arXiv preprint arXiv:2411.14708, 2024

    2024 arXiv

  38. [45]

    Think inside the json: Rein- forcement strategy for strict llm schema adherence,

    B. Agarwal, I. Joshi, and V . Rojkova, “Think inside the json: Rein- forcement strategy for strict llm schema adherence,”arXiv preprint arXiv:2502.14905, 2025

    2025 arXiv

  39. [46]

    Large language model alignment: A survey,

    T. Shen, R. Jin, Y . Huang, C. Liu, W. Dong, Z. Guo, X. Wu, Y . Liu, and D. Xiong, “Large language model alignment: A survey,”arXiv preprint arXiv:2309.15025, 2023

    2023 arXiv

  40. [47]

    Llama-3.2-1b-instruct-bf16,

    M. Community, “Llama-3.2-1b-instruct-bf16,” https://huggingface.co/mlx-community/Llama-3.2-1B-Instruct- MLXTuned, 2024, converted to MLX format from meta- llama/Llama-3.2-1B-Instruct using mlx-lm version 0.17.1. [Online]. Available: https://huggingface.co/mlx-community/Llama-3...

    2024

  41. [48]

    Llama-3.2-3b-bf16,

    ——, “Llama-3.2-3b-bf16,” https://huggingface.co/mlx- community/Llama-3.2-3B-bf16, 2024, converted to MLX format from meta-llama/Llama-3.2-3B using mlx-lm version 0.17.1. [Online]. Available: https://huggingface.co/mlx-community/Llama-3.2-3B-bf16

    2024

  42. [50]

    Llama beyond english: An empirical study on language capability transfer,

    J. Zhao, Z. Zhang, L. Gao, Q. Zhang, T. Gui, and X. Huang, “Llama beyond english: An empirical study on language capability transfer,” arXiv preprint arXiv:2401.01055, 2024

    2024 arXiv

  43. [51]

    How abilities in large language models are affected by supervised fine-tuning data composition,

    G. Dong, H. Yuan, K. Lu, C. Li, M. Xue, D. Liu, W. Wang, Z. Yuan, C. Zhou, and J. Zhou, “How abilities in large language models are affected by supervised fine-tuning data composition,”arXiv preprint arXiv:2310.05492, 2023

    2023 arXiv

  44. [52]

    Lora: Low-rank adaptation of large language models

    E. J. Hu, Y . Shen, P. Wallis, Z. Allen-Zhu, Y . Li, S. Wang, L. Wang, W. Chenet al., “Lora: Low-rank adaptation of large language models.” ICLR, vol. 1, no. 2, p. 3, 2022

    2022

  45. [53]

    MLX: Efficient and flexible machine learning on apple silicon,

    A. Hannun, J. Digani, A. Katharopoulos, and R. Collobert, “MLX: Efficient and flexible machine learning on apple silicon,” 2023. [Online]. Available: https://github.com/ml-explore

    2023

  46. [54]

    When scaling meets llm finetuning: The effect of data, model and finetuning method,

    B. Zhang, Z. Liu, C. Cherry, and O. Firat, “When scaling meets llm finetuning: The effect of data, model and finetuning method,”arXiv preprint arXiv:2402.17193, 2024

    2024 arXiv

  47. [55]

    Fine-tuning guide,

    U. Team, “Fine-tuning guide,” https://docs.unsloth.ai/get-started/fine- tuning-guide, 2025, accessed: 2025-05-11

    2025

  48. [56]

    Training compute-optimal large language models,

    J. Hoffmann, S. Borgeaud, A. Mensch, E. Buchatskaya, T. Cai, E. Rutherford, D. de Las Casas, L. A. Hendricks, J. Welbl, A. Clark, T. Hennigan, E. Noland, K. Millican, G. van den Driessche, B. Damoc, A. Guy, S. Osindero, K. Simonyan, E. Elsen, J. W. Rae, O. Vinyals, and L. Sifr...

    2022 arXiv

  49. [57]

    WiFi RSS & RTT dataset with different LOS conditions for indoor positioning,

    X. Feng, K. A. Nguyen, and Z. Luo, “WiFi RSS & RTT dataset with different LOS conditions for indoor positioning,” Zenodo, Jun. 2024, accessed: May 11, 2025. [Online]. Available: https://zenodo.org/doi/10.5281/zenodo.11558791

    2024 doi

  50. [58]

    A universal method to combat multipaths for rfid sensing,

    G. Wang, C. Qian, K. Cui, X. Shi, H. Ding, W. Xi, J. Zhao, and J. Han, “A universal method to combat multipaths for rfid sensing,” inIEEE INFOCOM 2020 - IEEE Conference on Computer Communications, 2020, pp. 277–286

    2020

  51. [59]

    A wifi rss-rtt indoor positioning system using dynamic model switching algorithm,

    X. Feng, K. A. Nguyen, and Z. Luo, “A wifi rss-rtt indoor positioning system using dynamic model switching algorithm,”IEEE Journal of Indoor and Seamless Positioning and Navigation, 2024

    2024

  52. [60]

    Spotfi: Decimeter level localization using wifi,

    M. Kotaru, K. Joshi, D. Bharadia, and S. Katti, “Spotfi: Decimeter level localization using wifi,”SIGCOMM Comput. Commun. Rev., vol. 45, no. 4, p. 269–282, Aug. 2015. [Online]. Available: https://doi.org/10.1145/2829988.2787487

    2015

  53. [61]

    Tokenizer choice for llm training: Negligible or crucial?

    M. Ali, M. Fromm, K. Thellmann, R. Rutmann, M. L ¨ubbering, J. Lev- eling, K. Klug, J. Ebert, N. Doll, J. Buschhoffet al., “Tokenizer choice for llm training: Negligible or crucial?” inFindings of the Association for Computational Linguistics: NAACL 2024, 2024, pp. 3907–3924

    2024

  54. [62]

    How susceptible are llms to influence in prompts?

    S. Anagnostidis and J. Bulian, “How susceptible are llms to influence in prompts?” 2024. [Online]. Available: https://arxiv.org/abs/2408.11865

    2024 arXiv

  55. [63]

    A geometric interpretation of trilateration for rss-based localization,

    H. M. Le, J.-P. Rossi, and D. Slock, “A geometric interpretation of trilateration for rss-based localization,” in2020 28th European Signal Processing Conference (EUSIPCO), 2021, pp. 1797–1801

    2021

  56. [64]

    Towards optimizing the costs of llm usage,

    S. Shekhar, T. Dubey, K. Mukherjee, A. Saxena, A. Tyagi, and N. Kotla, “Towards optimizing the costs of llm usage,”arXiv preprint arXiv:2402.01742, 2024

    2024 arXiv

  57. [65]

    The era of 1-bit llms: All large language models are in 1.58 bits,

    S. Ma, H. Wang, L. Ma, L. Wang, W. Wang, S. Huang, L. Dong, R. Wang, J. Xue, and F. Wei, “The era of 1-bit llms: All large language models are in 1.58 bits,”arXiv preprint arXiv:2402.17764, vol. 1, 2024

    2024 arXiv

  58. [66]

    Bitnet b1.58 2b4t technical report,

    S. Ma, H. Wang, S. Huang, X. Zhang, Y . Hu, T. Song, Y . Xia, and F. Wei, “Bitnet b1.58 2b4t technical report,” 2025. [Online]. Available: https://arxiv.org/abs/2504.12285

    2025 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.