Pith. sign in

REVIEW 4 major objections 5 minor 59 references

LightLLM: A Versatile Large Language Model for Predictive Light Sensing

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A frozen LLM, fitted with task-specific encoders and knowledge prompts, beats specialized light-sensing models by 4.4x in unseen rooms.

desk verdict Clever LLM-for-sensors framework with a thorough ablation, but the headline 4.4x generalization claim rests on an output-head mechanism the paper never specifies. read the letter →

arxiv 2411.15211 v1 pith:GHGO6O3L submitted 2024-11-20 cs.LG cs.AIcs.CVeess.SP

Jiawei Hu , Hong Jia , Mahbub Hassan , Lina Yao , Brano Kusy , Wen Hu This is my paper · ORCID
classification cs.LGcs.AIcs.CVeess.SP
keywords predictivelightsensinglargelanguagemodelsLoRAindoorlocalizationsolarforecastingestimationlatentfusionlayerunseen-environmentgeneralization
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 tries to establish that a pre-trained large language model, kept completely frozen and adapted only through lightweight trainable components, can serve as a general engine for predictive light sensing. LightLLM pairs each task with a sensor encoder, a textual knowledge prompt, and a latent fusion layer, and the authors report that this design generalizes to environments never seen during training: median localization error drops 4.4x relative to a specialized spectral-localization system, and indoor solar estimation error drops 3.4x when room layout and lighting are changed. The reason a reader should care is that, if true, it offers a single architecture that can be repurposed across sensing tasks without retraining a full model for every new room or lighting condition.

What carries the argument

The load-bearing mechanism is the Latent Fusion Layer (LFL), a multi-head attention module in which the task-specific encoder's features serve as queries ($Q = W_Q F_{\mathrm{enc}}$) while the embedded knowledge prompt supplies keys and values ($K = W_K P_{\mathrm{embed}}$, $V = W_V P_{\mathrm{embed}}$); learnable scalars $\alpha$ and $\beta$ reweight the two streams before scaled dot-product attention. Around this sit LoRA low-rank matrices $A$ and $B$ that adapt the frozen LLM's weights without changing the base parameters, a task-specific output head (softmax classification over candidate locations for localization), and, for localization, a knowledge graph whose edges encode field-of-view overlap and obstacle-free sightlines between sensors and light sources. The argument is that this combination lets the LLM apply general knowledge about light, geometry, and environment to new sensor signatures without retraining the base model.

What would settle it

Concretely, inspect the localization experiment: if the classification head is rebuilt or fine-tuned on the test room's location labels before evaluation, then the result is transfer learning, not deployment in a truly unseen environment. A decisive check is to train on the apartment, then test in a third room with a different coordinate grid without touching the output head; if median error jumps to several meters, the 4.4x figure depends on the shared-layout assumption rather than on the LLM's generalization.

Watch

Extended reading notes

Core claim

LightLLM's central claim is that a frozen LLM can outperform task-specific deep models on light-sensing tasks provided the sensor data are encoded properly and grounded through environmental prompts. The evidence the paper puts forward is cross-environment: trained in one room and tested in another, LightLLM keeps median localization error at 0.98 m and 1.19 m in office and apartment, while the Iris baseline degrades to 3.93 m and 4.35 m, a 4.4x improvement; with changed lighting and furniture, LightLLM's indoor solar estimation error is 26.33% MAPE against the best conventional model's 83.02%, which the abstract summarizes as 3.4x. The paper also argues that direct prompting of frontier LLMs with chain-of-thought remains far less accurate, showing that the task-specific encoder, knowledge prompt, and fusion layer, not the prompt alone, carry the generalization.

Load-bearing premise

The load-bearing premise is that the localization output head can produce predictions for a test room whose candidate positions were never seen during training; Section 2.7 describes a softmax classifier over possible locations, but Section 3.1 does not state whether that head is retrained for the new room, so the 4.4x generalization claim rests on an unspecified adaptation protocol.

Editorial extensions

If this is right

  • The same frozen LLM backbone can be reused across different light-sensing tasks by swapping only the encoder, prompt, fusion layer, and output head.
  • Specialized per-environment training is not the only route to accurate localization; a single model can transfer across rooms with shared coordinates.
  • Encoding and fusing sensor data explicitly beats turning sensor readings into text for frontier LLMs, so prompt-only LLM pipelines are likely insufficient for this class of tasks.
  • Using a larger base LLM (LLaMA-7B over GPT-2) improves all three tasks, so scaling the frozen backbone should further improve the framework.
  • Each component (knowledge graph, task-specific encoder, LoRA, latent fusion layer) measurably contributes to unseen-environment performance in the ablations, giving a clear recipe for adapting the design.

Reading between the lines

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

  • Editorial extension: If the cross-room localization result holds under a clean protocol, the same encoder-plus-prompt-plus-fusion pattern could be applied to other context-rich sensor modalities, such as RF-based activity recognition or acoustic room identification, which face the same 'new room each deployment' problem.
  • Editorial extension: The 4.4x claim depends on an unspecified detail in the unseen-room experiment: the localization head is a softmax classifier over candidate locations, and the paper does not say whether that head is rebuilt or how the coordinate grid is shared between apartment and office. A reader should check this before treating the gain as zero-retraining generalization.
  • Editorial extension: A direct experimental comparison against domain-adversarial or test-time-adaptation baselines on the same datasets would separate the LLM's world knowledge from its role as a high-capacity adaptable feature extractor.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. LightLLM is a framework that adapts a frozen pre-trained LLM to light-based sensing tasks via task-specific encoders, textual knowledge prompts, a latent fusion layer, and LoRA fine-tuning, with task-specific output heads. The paper evaluates the method on three tasks: light-spectral indoor localization, outdoor solar forecasting, and indoor solar estimation, using real-world datasets. The central claim is that LightLLM generalizes to unseen environments better than specialized baselines, highlighted by a 4.4x localization accuracy improvement in rooms not seen during training and a 3.4x improvement in unseen indoor solar estimation. The paper also includes ablations showing the contributions of each component and comparisons against direct ChatGPT-4 prompting.

Significance. If the results hold, LightLLM would demonstrate that a frozen LLM with task-specific front ends and LoRA can serve as a general-purpose engine for sensor-based prediction tasks, achieving strong cross-environment generalization with low tuning overhead. The work covers three distinct tasks with real deployments, and it compares against relevant baselines including TimeLLM, SkyGPT, and Iris. The systematic ablations (LFL, KG, LoRA, TSE, and the LLM itself) are valuable for understanding which components matter. However, the significance is currently limited by two load-bearing problems: an unspecified output-layer mechanism in the unseen-localization protocol, and the absence of any error bars or significance tests across all experimental tables.

major comments (4)
  1. [3.1, 2.7] The unseen-environment localization evaluation does not specify how the classification head described in Section 2.7 is applied to the test room's location set. Since the office (108 m^2, 27 sensors) and apartment (25 m^2, 17 sensors) have different candidate positions, a softmax head trained on one room cannot produce probabilities for the other's locations. The paper leaves open whether (a) the head is retrained on target-room labels (which would make the 'unseen' claim a transfer-learning claim and would compare against an Iris baseline that is not given the same adaptation opportunity), (b) both rooms share a common discretized coordinate grid (not stated and geometrically implausible given the different dimensions), or (c) the head is replaced by a coordinate-output regression (contradicting Section 2.7). The headline 4.4x improvement in Section 3.1 depends on this mechanism and must be specified.
  2. [Section 3 (all tables)] All reported metrics are point estimates without error bars, confidence intervals, or significance tests. The datasets are small (e.g., 1,035 samples in Section 3.3; two rooms in Section 3.1), so the claimed margins, such as the 4.4x localization improvement or the CRPS differences in Table 1, could be within noise. Please provide repeated-run statistics or statistical tests for the main comparisons, or at minimum report the number of independent trials and the variance across runs.
  3. [3.1 (seen environments)] The text first states that LightLLM shows 'comparable or slightly better' performance than Iris, but in the office environment it reports LightLLM 90th-percentile error of 0.89 m versus Iris 0.60 m, which is worse, and then says Iris 'performs slightly worse at 0.60m'. This is internally contradictory and undermines the credibility of the surrounding comparison. Please correct the description and, if the office result is indeed worse, discuss why.
  4. [2.4, Figure 4] The knowledge prompt for solar forecasting inserts {min_values} and {max_values} of the past period. The paper does not state whether these statistics are computed from the training split or from each test window. If computed from the source distribution only, the prompt in the unseen temporal-split evaluation (Section 3.2) could carry statistics that are inappropriate for the test period; if computed from test data, it would be a data-leakage concern. Please clarify the computation and provenance of these prompt statistics.
minor comments (5)
  1. [Abstract] The phrase 'fine tunes' should be 'fine-tunes' for consistency with standard usage.
  2. [Figure 4 caption] The word 'genralization' is misspelled; it should be 'generalization'.
  3. [Section 2.5] The phrase 'an multi-head attention mechanism' should be 'a multi-head attention mechanism'.
  4. [Section 3.1] The article 'a' is missing before 'apartment' in the sentence describing the two environments; also 'evalute' should be 'evaluate'.
  5. [General] No code or data availability statement is included. Given the real-world datasets, a statement about releasing the code and raw data would aid reproducibility.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: LightLLM's headline results are empirical evaluations on held-out data, and the only self-citation (the Iris baseline) is not load-bearing.

full rationale

LightLLM is an empirical systems paper rather than a derivation. Each headline result (4.4x localization improvement, 3.4x indoor solar improvement, CRPS/FS gains for forecasting) is obtained by training the proposed encoder/prompt/fusion/LoRA stack on one split or environment and testing on a held-out split or environment; no equation in the paper defines the output in terms of those target values. The one analytic element, Eq. (5) for photocurrent, is used explicitly as a failing baseline (average MAPE 46.83%) rather than as the source of LightLLM's predictions. The task-specific prompts contain descriptive input statistics (e.g., min/max sunlight values) and the fusion/LoRA parameters are learned, so predictions are not equal to these inputs by construction. The cross-room localization protocol in Section 3.1 is underspecified relative to the softmax classification head described in Section 2.7, since the apartment and office have different location label sets; however, this is a reproducibility gap, not a circular reduction, because the paper does not derive the 0.98 m and 1.19 m errors from the head definition. The Iris baseline [18] is prior work by overlapping authors, but it is used as an empirical comparison point alongside external methods (SkyGPT, TimeLLM, ConvLSTM) and an external dataset (SKIPP'D), so the central claims do not reduce to self-citation. Overall, the derivation chain is self-contained with respect to its inputs; score 1 reflects only a minor, non-load-bearing self-citation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on four domain assumptions and three free parameter groups. The most consequential are the LLM-knowledge-transfer premise and the unspecified KG thresholds, because the headline generalization results depend on both. No invented entities appear.

free parameters (3)
  • KG construction thresholds
    Algorithm 1 requires distance_threshold, vertical_threshold, FOV angle, and detection range to decide graph edges. These values are never stated, yet the KG ablation shows the graph is responsible for a 50% localization error increase when removed (Section 3.4.4).
  • LoRA rank and scaling = r=64, alpha=64, dropout=0.1
    Section 2.6 states these were chosen empirically. Since Table 5 shows LoRA contributes to performance, the specific choice affects the reported gains.
  • Prompt statistics (min/max) = computed from training data
    The task prompt in Figure 4 inserts {min_values} and {max_values} of the historical PV output. These data-derived values carry distribution information into the fused representation and are not described as standard model inputs.
assumptions (4)
  • domain assumption Pre-trained LLMs carry transferable knowledge about spatial layouts, lighting, and environmental patterns that can be activated by short text prompts.
    Section 2.4: 'the pre-trained LLM already possesses generalized knowledge about spatial relationships and environmental patterns' is the core premise for why prompting helps cross-environment generalization.
  • domain assumption The Knowledge Graph built from sensor coordinates, orientations, FOVs, and obstacle maps correctly represents the physical relationships that determine light readings.
    Algorithm 1 constructs edges from geometry; if this geometry is wrong or incomplete, the GNN encoder's graph is wrong, and Section 3.4.4 shows localization degrades 50% without the KG.
  • domain assumption Eighteen-channel spectral sensor readings contain enough information for localization, solar forecasting, and solar estimation.
    All three tasks use the AS7265x sensor or PV series; the paper does not analyze which wavelengths matter or whether 18 channels are sufficient.
  • domain assumption LoRA fine-tuning preserves the pre-trained model's general knowledge while adapting to the task.
    Section 2.6 assumes the frozen model plus low-rank updates keeps pre-trained knowledge intact; this is standard in the literature, but the paper does not verify it beyond the ablation in Table 5.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LightLLM: A Versatile Large Language Model for Predictive Light Sensing." pith.science (2026). https://pith.science/paper/GHGO6O3L

@misc{pith2026241115211,
  author       = {Pith},
  title        = {Pith review of: LightLLM: A Versatile Large Language Model for Predictive Light Sensing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GHGO6O3L}},
  note         = {Machine review of arXiv:2411.15211}
}
read the original abstract

We propose LightLLM, a model that fine tunes pre-trained large language models (LLMs) for light-based sensing tasks. It integrates a sensor data encoder to extract key features, a contextual prompt to provide environmental information, and a fusion layer to combine these inputs into a unified representation. This combined input is then processed by the pre-trained LLM, which remains frozen while being fine-tuned through the addition of lightweight, trainable components, allowing the model to adapt to new tasks without altering its original parameters. This approach enables flexible adaptation of LLM to specialized light sensing tasks with minimal computational overhead and retraining effort. We have implemented LightLLM for three light sensing tasks: light-based localization, outdoor solar forecasting, and indoor solar estimation. Using real-world experimental datasets, we demonstrate that LightLLM significantly outperforms state-of-the-art methods, achieving 4.4x improvement in localization accuracy and 3.4x improvement in indoor solar estimation when tested in previously unseen environments. We further demonstrate that LightLLM outperforms ChatGPT-4 with direct prompting, highlighting the advantages of LightLLM's specialized architecture for sensor data fusion with textual prompts.

Figures

Figures reproduced from arXiv: 2411.15211 by the authors.

Figure 1
Figure 1. Illustration of PLS tasks in the smart build [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Architecture of LightLLM. LightLLM in terms of accuracy and genralization compared to state-of-the-art methods. • We provide an extensive ablation study to evaluate the impact of each component within LightLLM, of￾fering insights into its adaptability and performance across different tasks. 2 Methodology 2.1 System Overview Our proposed framework, LightLLM, integrates advanced encoders for various data modalities wi… view at source ↗
Figure 3
Figure 3. Example of a KG illustrating sensor and light [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Task-Specific Knowledge Prompt Example: Solar Energy Forecasting Prompt. 2.5 Adaptive Latent Fusion Layer The LFL in LightLLM is a key component designed to in￾tegrate encoded features from task-specific encoders with prompt embeddings generated from task-specific know…
Figure 5
Figure 5. Figure 5: Overview of Latent Fusion Layer. key (𝐾) assesses relevance, while the value (𝑉 ) provides the information to enhance the final representation. Each head in the multi-head attention mechanism works with different learned weight matrices 𝑊𝑄 ,𝑊𝐾 and 𝑊𝑉 . It splits the ma…
Figure 6
Figure 6. Figure 6: Localization error under seen scenarios. outputs a probability distribution over possible locations, al￾lowing the model to predict the most likely position based on the input data. The classification head uses a softmax activa￾tion function to convert the model’s logi…
Figure 8
Figure 8. Figure 8: , LightLLM maintains its edge with a CRPS of 2.52 kW and a FS improvement of 31.4%, outperforming SkyGPT (CRPS: 2.81 kW, FS: 23.4%) and TimeLLM (CRPS: 2.67 kW, FS [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 10
Figure 10. Figure 10: Experiment results show the difference be [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 11
Figure 11. Figure 11: Prompt example for LSI-based localization. [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 13
Figure 13. Figure 13: Performance comparison of three ablation [PITH_FULL_IMAGE:figures/full_fig_p011_13.png]
Figure 14
Figure 14. Figure 14: Performance comparison between using dif [PITH_FULL_IMAGE:figures/full_fig_p012_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 35 canonical work pages

  1. [1]

    Original Samsung Smart TV Remote (Solar Cell Remote Control - BN59-01385B Genuine)

    2023. Original Samsung Smart TV Remote (Solar Cell Remote Control - BN59-01385B Genuine). https://remotecontrolwarehouse.com.au/ products/

    work page 2023

  2. [2]

    Monica Agrawal, Stefan Hegselmann, Hunter Lang, Yoon Kim, and David Sontag. 2022. Large language models are few-shot clinical information extractors. arXiv preprint arXiv:2205.12689 (2022)

    arXiv 2022

  3. [3]

    Mistral AI. 2024. Mistral. https://mistral.ai/

    work page 2024

  4. [4]

    ams OSRAM. 2022. AS7265x Smart Spectral Sensor. https://ams.com/ en/as7265x

    work page 2022

  5. [5]

    Bharathan Balaji, Arka Bhattacharya, Gabriel Fierro, Jingkun Gao, Joshua Gluck, Dezhi Hong, Aslak Johansen, Jason Koh, Joern Ploen- nigs, Yuvraj Agarwal, et al. 2016. Brick: Towards a unified metadata schema for buildings. In Proceedings of the 3rd ACM International Conference on Systems for Energy-Efficient Built Environments . 41–50

    work page 2016

  6. [6]

    Kurt Benke and Bruce Tomkins. 2017. Future food-production systems: vertical farming and controlled-environment agriculture. Sustainabil- ity: Science, Practice and Policy 13, 1 (2017), 13–26

    work page 2017

  7. [7]

    Swarup Biswas and Hyeok Kim. 2020. Solar cells for indoor applica- tions: Progress and development. Polymers 12, 6 (2020), 1338

    work page 2020

  8. [8]

    Kathrin Blagec, Adriano Barbosa-Silva, Simon Ott, and Matthias Samwald. 2022. A curated, ontology-based, large-scale knowledge graph of artificial intelligence tasks and benchmarks. Scientific Data 9, 1 (2022), 322

    work page 2022

Show all 59 references
  1. [9]

    Aldo Vieira Da Rosa and Juan Carlos Ordonez. 2021. Fundamentals of renewable energy processes. Academic Press

    2021

  2. [10]

    Jacob Devlin. 2018. Bert: Pre-training of deep bidirectional trans- formers for language understanding. arXiv preprint arXiv:1810.04805 (2018)

    2018 arXiv

  3. [11]

    Danny Driess, Fei Xia, Mehdi SM Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, et al . 2023. Palm-e: An embodied multimodal language model. arXiv preprint arXiv:2303.03378 (2023)

    2023 arXiv

  4. [12]

    AA El-Sebaii, FS Al-Hazmi, AA Al-Ghamdi, and Saud Jameel Yagh- mour. 2010. Global, direct and diffuse solar radiation on horizontal and tilted surfaces in Jeddah, Saudi Arabia. Applied energy 87, 2 (2010), 568–576

    2010

  5. [13]

    Tilmann Gneiting and Matthias Katzfuss. 2014. Probabilistic fore- casting. Annual Review of Statistics and Its Application 1, 1 (2014), 125–151

    2014

  6. [14]

    Ian Goodfellow. 2016. Deep learning

    2016

  7. [15]

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2020. Generative adversarial networks. Commun. ACM 63, 11 (2020), 139– 144

    2020

  8. [16]

    Vincent Le Guen and Nicolas Thome. 2020. Disentangling physical dynamics from unknown factors for unsupervised video prediction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11474–11484

    2020

  9. [17]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 (2021)

    2021 arXiv

  10. [18]

    Jiawei Hu, Yanxiang Wang, Hong Jia, Wen Hu, Mahbub Hassan, Brano Kusy, Ashraf Uddin, and Moustafa Youssef. 2023. Iris: Passive Visible Light Positioning Using Light Spectral Information. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 7, 3 (...

    2023

  11. [19]

    Jiawei Hu, Yanxiang Wang, Hong Jia, Cheng Jiang, Mahbub Hassan, Brano Kusy, and Wen Hu. 2024. LiDARSpectra: Synthetic Indoor Spectral Mapping with Low-cost LiDARs. In 2024 23rd ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN). IEEE, 75–87

    2024

  12. [20]

    Rich H Inman, Hugo TC Pedro, and Carlos FM Coimbra. 2013. Solar forecasting methods for renewable energy integration. Progress in energy and combustion science 39, 6 (2013), 535–576

    2013

  13. [21]

    Arturo Fajardo Jaimes and Fernando Rangel de Sousa. 2017. Simple modeling of photovoltaic solar cells for indoor harvesting applications. Solar Energy 157 (2017), 792–802

    2017

  14. [22]

    Sijie Ji, Xinzhe Zheng, and Chenshu Wu. 2024. HARGPT: Are LLMs Zero-Shot Human Activity Recognizers? arXiv preprint arXiv:2403.02727 (2024)

    2024 arXiv

  15. [23]

    Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, et al. 2023. Time-llm: Time series forecasting by reprogramming large language models. arXiv preprint arXiv:2310.01728 (2023). Hu, et al

    2023 arXiv

  16. [24]

    Daniel Konings, Nathaniel Faulkner, Fakhrul Alam, Edmund M-K Lai, and Serge Demidenko. 2019. FieldLight: Device-free indoor human lo- calization using passive visible light positioning and artificial potential fields. IEEE Sensors Journal 20, 2 (2019), 1054–1066

    2019

  17. [25]

    Colin Lea, Michael D Flynn, Rene Vidal, Austin Reiter, and Gregory D Hager. 2017. Temporal convolutional networks for action segmenta- tion and detection. In proceedings of the IEEE Conference on Computer Vision and Pattern Recognition . 156–165

    2017

  18. [26]

    Xiang Lisa Li and Percy Liang. 2021. Prefix-tuning: Optimizing contin- uous prompts for generation. arXiv preprint arXiv:2101.00190 (2021)

    2021 arXiv

  19. [27]

    Zewen Li, Fan Liu, Wenjie Yang, Shouheng Peng, and Jun Zhou. 2021. A survey of convolutional neural networks: analysis, applications, and prospects. IEEE transactions on neural networks and learning systems 33, 12 (2021), 6999–7019

    2021

  20. [28]

    Pengfei Liu, Weizhe Yuan, Jinlan Fu, Zhengbao Jiang, Hiroaki Hayashi, and Graham Neubig. 2023. Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing.Comput. Surveys 55, 9 (2023), 1–35

    2023

  21. [29]

    Dong Ma, Guohao Lan, Mahbub Hassan, Wen Hu, Mushfika B Upama, Ashraf Uddin, and Moustafa Youssef. 2019. Solargest: Ubiquitous and battery-free gesture recognition using solar cells. In The 25th annual international conference on mobile computing and networking . 1–15

    2019

  22. [30]

    Yogesh Kumar Meena, Krishna Seunarine, Deepak Ranjan Sahoo, Si- mon Robinson, Jennifer Pearson, Chi Zhang, Matt Carnie, Adam Pock- ett, Andrew Prescott, Suzanne K Thomas, et al. 2020. PV-tiles: towards closely-coupled photovoltaic and digital materials for useful, beautiful an...

    2020

  23. [31]

    Yuhao Nie, Xiatong Li, Andea Scott, Yuchi Sun, Vignesh Venugopal, and Adam Brandt. 2023. SKIPP’D: A SKy Images and Photovoltaic Power Generation Dataset for short-term solar forecasting. Solar Energy 255 (2023), 171–179

    2023

  24. [32]

    Yuhao Nie, Eric Zelikman, Andea Scott, Quentin Paletta, and Adam Brandt. 2024. SkyGPT: Probabilistic ultra-short-term solar forecast- ing using synthetic sky images from physics-constrained VideoGPT. Advances in Applied Energy 14 (2024), 100172

    2024

  25. [33]

    OpenAI. 2023. GPT-4 Technical Report. https://openai.com/research/ gpt-4

    2023

  26. [34]

    Quentin Paletta, Guillaume Arbod, and Joan Lasenby. 2021. Bench- marking of deep learning irradiance forecasting models from sky images–An in-depth analysis. Solar Energy 224 (2021), 855–867

    2021

  27. [35]

    Antonio Parretta, Angelo Sarno, and Luciano RM Vicari. 1998. Effects of solar irradiation conditions on the outdoor performance of photo- voltaic modules. Optics Communications 153, 1-3 (1998), 153–163

    1998

  28. [36]

    Joaquin Quiñonero-Candela, Masashi Sugiyama, Anton Schwaighofer, and Neil D Lawrence. 2022. Dataset shift in machine learning . Mit Press

    2022

  29. [37]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al . 2021. Learning transferable visual mod- els from natural language supervision. In International conference on machine lea...

    2021

  30. [38]

    Machel Reid, Nikolay Savinov, Denis Teplyashin, Dmitry Lepikhin, Timothy Lillicrap, Jean-baptiste Alayrac, Radu Soricut, Angeliki Lazari- dou, Orhan Firat, Julian Schrittwieser, et al. 2024. Gemini 1.5: Unlock- ing multimodal understanding across millions of tokens of context....

    2024 arXiv

  31. [39]

    Xingjian Shi, Zhourong Chen, Hao Wang, Dit-Yan Yeung, Wai-Kin Wong, and Wang-chun Woo. 2015. Convolutional LSTM network: A machine learning approach for precipitation nowcasting. Advances in neural information processing systems 28 (2015)

    2015

  32. [40]

    Karan Singhal, Tao Tu, Juraj Gottweis, Rory Sayres, Ellery Wulczyn, Le Hou, Kevin Clark, Stephen Pfohl, Heather Cole-Lewis, Darlene Neal, et al. 2023. Towards expert-level medical question answering with large language models. arXiv preprint arXiv:2305.09617 (2023)

    2023 arXiv

  33. [41]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie- Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971 (2023)

    2023 arXiv

  34. [42]

    A Vaswani. 2017. Attention is all you need. Advances in Neural Information Processing Systems (2017)

    2017

  35. [43]

    Yanxiang Wang, Jiawei Hu, Hong Jia, Wen Hu, Mahbub Hassan, Ashraf Uddin, Brano Kusy, and Moustafa Youssef. 2023. Spectral-Loc: Indoor Localization using Light Spectral Information. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 7, 1 (2023), 1–26

    2023

  36. [44]

    Zeyu Wang and Ravi S Srinivasan. 2017. A review of artificial in- telligence based building energy use prediction: Contrasting the ca- pabilities of single and ensemble prediction models. Renewable and Sustainable Energy Reviews 75 (2017), 796–808

    2017

  37. [45]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompt- ing elicits reasoning in large language models. Advances in neural information processing systems 35 (2022), 24824–24837

    2022

  38. [46]

    Karl Weiss, Taghi M Khoshgoftaar, and DingDing Wang. 2016. A survey of transfer learning. Journal of Big data 3 (2016), 1–40

    2016

  39. [47]

    Yuxuan Weng, Guoquan Wu, Tianyue Zheng, Yanbing Yang, and Jun Luo. 2024. Large Model for Small Data: Foundation Model for Cross- Modal RF Human Activity Recognition. In Proceedings of the 22nd ACM Conference on Embedded Networked Sensor Systems . 436–449

    2024

  40. [48]

    Matthew Wright and Ashraf Uddin. 2012. Organic—inorganic hybrid solar cells: A comparative review. Solar energy materials and solar cells 107 (2012), 87–111

    2012

  41. [49]

    Duo Wu, Xianda Wang, Yaqi Qiao, Zhi Wang, Junchen Jiang, Shuguang Cui, and Fangxin Wang. 2024. NetLLM: Adapting Large Language Models for Networking. In Proceedings of the ACM SIGCOMM 2024 Conference. 661–678

    2024

  42. [50]

    Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. 2020. A comprehensive survey on graph neural networks. IEEE transactions on neural networks and learning systems 32, 1 (2020), 4–24

    2020

  43. [51]

    Xuhai Xu, Bingsheng Yao, Yuanzhe Dong, Saadia Gabriel, Hong Yu, James Hendler, Marzyeh Ghassemi, Anind K Dey, and Dakuo Wang

  44. [52]

    Huanqi Yang, Sijie Ji, Rucheng Wu, and Weitao Xu. 2024. Are You Being Tracked? Discover the Power of Zero-Shot Trajectory Tracing with LLMs! arXiv preprint arXiv:2403.06201 (2024)

    2024 arXiv

  45. [53]

    Jonathan K Yohanna, Isaac N Itodo, and Victor I Umogbai. 2011. A model for determining the global solar radiation for Makurdi, Nigeria. Renewable Energy 36, 7 (2011), 1989–1992

    2011

  46. [54]

    Jinsong Zhang, Rodrigo Verschae, Shohei Nobuhara, and Jean-François Lalonde. 2018. Deep photovoltaic nowcasting. Solar Energy 176 (2018), 267–276

    2018

  47. [55]

    Zhun Zhong, Liang Zheng, Guoliang Kang, Shaozi Li, and Yi Yang

  48. [56]

    Xiao Xiang Zhu, Devis Tuia, Lichao Mou, Gui-Song Xia, Liangpei Zhang, Feng Xu, and Friedrich Fraundorfer. 2017. Deep learning in remote sensing: A comprehensive review and list of resources. IEEE geoscience and remote sensing magazine 5, 4 (2017), 8–36. LightLLM: A Versatile L...

    2017

  49. [57]

    Fuzhen Zhuang, Zhiyuan Qi, Keyu Duan, Dongbo Xi, Yongchun Zhu, Hengshu Zhu, Hui Xiong, and Qing He. 2020. A comprehensive survey on transfer learning. Proc. IEEE 109, 1 (2020), 43–76

    2020

  50. [2020]

    In Proceedings of the AAAI conference on artificial intelligence , Vol

    Random erasing data augmentation. In Proceedings of the AAAI conference on artificial intelligence , Vol. 34. 13001–13008

  51. [2024]

    Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 8, 1 (2024), 1–32

    Mental-llm: Leveraging large language models for mental health prediction via online text data. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 8, 1 (2024), 1–32

    2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.