Pith. sign in

REVIEW 3 major objections 5 minor 49 references

EpiLLM: Unlocking the Potential of Large Language Models in Epidemic Forecasting

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read The paper claims that a frozen LLM, fed infection cases and human mobility as token sequences, outperforms dedicated spatio-temporal epidemic models on real-world COVID-19 forecasting.

desk verdict A solid frozen-LLM epidemic forecaster whose main claimed mechanism—mobility co-prediction—is never directly validated, and whose significance claims lack statistical backing. read the letter →

arxiv 2505.12738 v1 pith:BKXLIDGB submitted 2025-05-19 cs.LG cs.AIcs.SI

classification cs.LGcs.AIcs.SI
keywords largelanguagemodelsepidemicforecastingspatio-temporalmodelingautoregressivepromptlearninghumanmobilityCOVID-19next-tokenprediction
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 argues that epidemic forecasting can be treated as a language-model task: convert regional infection counts and inter-region human mobility into token sequences, and let a frozen large language model predict the next tokens. The proposed framework, EpiLLM, uses a dual-branch projector to align both signals into the LLM's embedding space, then performs autoregressive next-token prediction with learnable spatio-temporal prompts. On real-world COVID-19 datasets from England, France, Italy, and Spain, the paper reports that the best EpiLLM variant achieves the lowest RMSE and MAE for 3- and 7-day direct forecasts, with RMSE improvements over the strongest baseline up to 30.38%. If correct, this would mean a frozen general-purpose LLM, with only lightweight adapters trained, can outperform specialized epidemic models and improve further as the backbone scales.

What carries the argument

The load-bearing mechanism is dual-branch token alignment combined with autoregressive next-token prediction. A graph neural network maps infection-case features to tokens $Z_{1:T}$, an MLP maps mobility matrices to tokens $H_{1:T}$, frozen LLM layers predict the next patch from previous tokens, and decoupled adapters map predictions back to the original feature space. Around this, spatio-temporal prompt learning adds direction-aware forward and backward edges and learnable time-gating weights $\gamma_k$ over a token window, allowing the model to weight recent time steps more heavily. The mobility branch's predicted adjacency matrix is then fed into the epidemic branch for multi-step forecasting.

What would settle it

Run EpiLLM with ground-truth future mobility fed into the epidemic branch instead of predicted mobility; if 14-day RMSE does not worsen compared with the predicted-mobility version, the mobility branch is not contributing as claimed. Alternatively, compute RMSE and MAE of the predicted mobility matrices against actual future mobility at 3- and 7-day horizons.

Watch

Extended reading notes

Core claim

The central claim is that spatio-temporal epidemic forecasting can be reformulated as next-token prediction over tokens derived from infection cases and human mobility, and that this reformulation lets a frozen LLM outperform dedicated epidemic forecasting models. The framework jointly models an epidemic branch and a mobility branch; at inference it first predicts future mobility structure, then uses those predicted mobility graphs to forecast future cases iteratively for arbitrary horizons. The paper reports direct-forecast superiority on all four datasets, successful 6- and 14-day multi-step forecasts, and scaling behavior in which larger LLM backbones generally yield lower RMSE while the fraction of trainable parameters falls to as low as 0.03%.

Load-bearing premise

The load-bearing premise is that the mobility branch can forecast future human movement accurately enough to help infection forecasting; the paper asserts this but never measures mobility prediction error against actual mobility data.

Editorial extensions

If this is right

  • On all four COVID-19 datasets, the best EpiLLM variant records the lowest RMSE and MAE for 3- and 7-day direct forecasts; on Spain the RMSE improvement over the strongest baseline is 30.38%.
  • Because the framework generates mobility jointly with cases, it can produce 6- and 14-day multi-step forecasts, which the comparison models cannot do without future mobility input.
  • Forecast quality improves with LLM parameter scale within the GPT2 and GEMMA3 families, while the trainable-parameter ratio drops to between 0.03% and 0.58%, so larger frozen backbones give better forecasts at modest added training cost.
  • Learned prompt weights rise toward the current time step and forward edges dominate backward edges, consistent with the temporal-directionality prior used to initialize them.
  • Replacing the LLM backbone with trainable MLP, RNN, or Transformer blocks degrades performance, indicating that the pre-trained autoregressive next-token mechanism carries part of the forecasting ability.

Reading between the lines

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

  • If the mobility branch forecasts future movement accurately, the same dual-branch recipe could transfer to other mobility-driven contagions by retraining only the lightweight projectors and adapters on new regional data.
  • The reported scaling trend implies a testable corollary: on longer epidemic records, larger frozen backbones should show growing relative gains over small ones, which a reader could verify before adopting the method.
  • A comparison against an oracle variant that receives ground-truth future mobility instead of predicted mobility would isolate how much of the multi-step gain comes from mobility prediction; the paper does not report this comparison.
  • The learned prompt weights suggest the model discovers a recency prior, so transferring trained prompt parameters to a new outbreak region may accelerate adaptation; that transferability is not demonstrated in the paper.
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

3 major / 5 minor

Summary. The paper introduces EpiLLM, a framework that adapts frozen large language models to spatio-temporal epidemic forecasting. It uses a dual-branch design: an epidemic branch tokenizes infection-case histories with a GNN and a mobility branch tokenizes human-mobility matrices with an MLP, then an autoregressive next-token-prediction objective aligns both with an LLM. Spatio-temporal prompt learning adds learnable direction-aware edges and time-gating weights. The method is evaluated on COVID-19 datasets for England, France, Italy, and Spain for direct horizons of 3 and 7 days and multi-step horizons of 6 and 14 days. The authors report consistent direct-forecasting improvements over fourteen baselines, with EpiLLM-GEMMA3 achieving the best RMSE/MAE in most settings and up to 30.38% RMSE improvement on Spain, plus ablations, scaling experiments, prompt visualizations, and parameter-efficiency statistics.

Significance. If the results hold, the paper would demonstrate that a frozen LLM with light trainable projectors and adapters can outperform dedicated spatio-temporal epidemic models, which is a meaningful and practically attractive contribution. The code release, the explicit autoregressive reformulation, and the prompt-visualization analysis are strengths. I found no evidence of circularity: evaluation is on held-out future days and the hand-set prompt initializations do not encode test labels. However, the central mechanism and the multi-step claims currently rest on unverified assumptions about the mobility branch and on absent baseline comparisons, so the contribution is promising but not yet fully established.

major comments (3)
  1. [Appendix D, Eq. (9)-(10), Table 5] The mobility branch is load-bearing but its forecast accuracy is never evaluated against ground truth. The paper states in Appendix D that "Human mobility prediction constitutes the core component of our framework," and Eqs. (9)-(10) feed predicted mobility matrices into the epidemic branch during inference. Yet no table or figure reports held-out mobility-prediction error. Table 5 only shows that removing or approximating mobility input hurts performance; it neither establishes that the predicted mobility is accurate nor isolates the mobility branch's contribution from the frozen LLM's autoregressive prior. Please add a direct evaluation of mobility forecasts (e.g., MAE/RMSE of mobility matrices on held-out test periods) and, ideally, an oracle-mobility variant that uses ground-truth mobility during multi-step inference.
  2. [Table 2, Section 5.2] The multi-step forecasting results contain no baselines, only the three EpiLLM backbones. The text claims that "Other baselines fail to perform multi-step forecasting due to the absence of future mobility prediction," but this is an assertion rather than a demonstrated result. Standard baselines such as ATMGNN, MPNN+LSTM, and MGNN can be run in recursive or direct multi-step modes, and an LLM-based case-only variant could be included as a control. Without such comparisons, the multi-step capability claim is unsupported.
  3. [Table 1 caption, Section 5.1] The claim that "Experimental results have passed the statistical significance tests" is not substantiated. No test name, test statistic, p-value, confidence interval, or standard deviation is reported anywhere. Since the paper averages over 10 runs, error bars and a paired significance test across runs or regions should be reported; otherwise the word "significantly" in the abstract and Section 5.2 is not supported.
minor comments (5)
  1. [Table 5] The England 14-day RMSE entry "44.887" appears to have an extra digit compared with the other entries; please make the decimal precision consistent.
  2. [Figure 2] The ablation figure is difficult to read at the current resolution; a numerical table of the ablation results would make the comparisons verifiable.
  3. [Section 5.2, Table 2] The statement that EpiLLM-GEMMA3 demonstrates superior multi-step generation capability is contradicted by Table 2, where EpiLLM-DeepSeekR1 has lower RMSE on England 6-day and Spain 6-day; please qualify or correct the claim.
  4. [Eq. (4)] The notation H_{1:T} = MLP(M_{1:T}) with M_t in R^{N x N} should clarify how the MLP handles the matrix-valued input and whether spatial structure is flattened or treated graph-wise.
  5. [Abstract and Introduction] There are minor typos, e.g., "futher forecasting" in the Introduction and "countires" in Appendix A; a proofread would be helpful.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: EpiLLM is trained on historical windows and evaluated on held-out future days; the unvalidated mobility branch is an evidence gap, not a circular reduction.

full rationale

The paper's central claim (Section 5.2, Table 1) is an empirical comparison on held-out future days. Appendix A specifies a temporally ordered split: the last {3, 6, 7, 14} days are reserved as the test set, the immediately preceding {3, 7} days serve as validation, and the remaining data are used for training. The training losses in Eqs. (7) and (8), L_Epi and L_Mob, supervise reconstruction of historical windows from previous historical tokens; no test-label information enters the training objective. At inference, Eqs. (9) and (10) roll out the mobility branch's own predictions into the epidemic branch, which is a standard autoregressive generation procedure rather than a tautology. The spatio-temporal prompts are initialized to fixed values (forward edge 1, backward edge 0.5, time gates 1; Appendix E, Table 6) and then optimized on historical data, so their learned values are not fitted to the forecast targets. The paper does cite its own prior work [20] (AutoCAs) as a design precedent for autoregressive LLM forecasting and for freezing the LLM while tuning projectors and adapters, but that citation is not load-bearing for the epidemic benchmark numbers: the COVID-19 results are computed by the present model on external data, not imported from the citation. The most serious weakness is that Appendix D asserts that human mobility prediction 'constitutes the core component of our framework,' yet the paper never evaluates mobility predictions against ground-truth mobility. This is an unresolved evidential gap about an auxiliary branch, not a circular derivation: the mobility predictor is trained on historical mobility and its quality is not defined in terms of the epidemic forecast target. Because no equation reduces the claimed prediction to its inputs and no fitted parameter is renamed as a prediction, the appropriate circularity score is 0.

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

The central claim rests on standard ML practice (train/test splits, frozen LLM adaptation) plus domain assumptions about mobility-driven spread and data quality. The free parameters are the unreported loss weight, the hand-set prompt initialization, and the chosen token window size.

free parameters (3)
  • Lambda (mobility loss weight)
    Appears in Eq. 8 as the weight on the mobility-branch loss; no value is reported, so the exact training objective is under-specified.
  • Prompt initialization values = forward edge 1, backward edge 0.5, time gates 1
    Set by hand in Appendix E to reflect a temporal directionality prior; these are trainable but the initialization is a modeling choice.
  • Token window length w = 3 or 7
    Chosen to match assumed incubation periods; it determines the prompt window and number of gating parameters, but is not swept or justified by data.
assumptions (4)
  • domain assumption Human mobility is a primary driver of epidemic spread, and transmission occurs only where population flow is nonzero.
    Used in the problem formulation (Eq. 1) and motivates the dual-branch architecture; it is a modeling assumption, not derived from the data.
  • domain assumption The reused COVID-19 case and Facebook mobility data are accurate, and the exclusions (regions without cases or unlinkable mobility records) do not introduce bias.
    Appendix A describes dataset construction based on third-party preprocessing; the paper does not independently validate data quality.
  • domain assumption Autoregressive next-token prediction with a frozen LLM is an effective objective for spatio-temporal epidemic forecasting.
    This is the central premise of the method; the paper supports it only through the reported experiments, with no theoretical guarantee.
  • domain assumption The temporally ordered split (last 3/7/14 days as test, preceding 3/7 days as validation) is a valid evaluation protocol for these ~60-day series.
    The short test windows may not represent later epidemic phases; the protocol follows prior work but is not stress-tested.

how reviews work

0 comments
Cite this review

Pith. "Pith review of EpiLLM: Unlocking the Potential of Large Language Models in Epidemic Forecasting." pith.science (2026). https://pith.science/paper/BKXLIDGB

@misc{pith2026250512738,
  author       = {Pith},
  title        = {Pith review of: EpiLLM: Unlocking the Potential of Large Language Models in Epidemic Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BKXLIDGB}},
  note         = {Machine review of arXiv:2505.12738}
}
read the original abstract

Advanced epidemic forecasting is critical for enabling precision containment strategies, highlighting its strategic importance for public health security. While recent advances in Large Language Models (LLMs) have demonstrated effectiveness as foundation models for domain-specific tasks, their potential for epidemic forecasting remains largely unexplored. In this paper, we introduce EpiLLM, a novel LLM-based framework tailored for spatio-temporal epidemic forecasting. Considering the key factors in real-world epidemic transmission: infection cases and human mobility, we introduce a dual-branch architecture to achieve fine-grained token-level alignment between such complex epidemic patterns and language tokens for LLM adaptation. To unleash the multi-step forecasting and generalization potential of LLM architectures, we propose an autoregressive modeling paradigm that reformulates the epidemic forecasting task into next-token prediction. To further enhance LLM perception of epidemics, we introduce spatio-temporal prompt learning techniques, which strengthen forecasting capabilities from a data-driven perspective. Extensive experiments show that EpiLLM significantly outperforms existing baselines on real-world COVID-19 datasets and exhibits scaling behavior characteristic of LLMs.

Figures

Figures reproduced from arXiv: 2505.12738 by the authors.

Figure 1
Figure 1. The overall framework of EpiLLM consists of three modules: (1) dual-branch token [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Ablation study of EpiLLM for epidemic forecasting. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Scaling behavior of EpiLLM on France and Italy datasets. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Prompt visualization of EpiLLM [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Case study of France (part regions) COVID-19 progression during May 10-12, 2020. Areas [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 6
Figure 6. Figure 6: Case study of Spain (part regions) COVID-19 progression during May 10-12, 2020. Areas [PITH_FULL_IMAGE:figures/full_fig_p018_6.png]
Figure 7
Figure 7. Figure 7: Case study of Italy (part regions) COVID-19 progression during April 22-24, 2020. Areas [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 29 canonical work pages

  1. [1]

    The covid-19 pandemic

    Marco Ciotti, Massimo Ciccozzi, Alessandro Terrinoni, Wen-Can Jiang, Cheng-Bin Wang, and Sergio Bernardini. The covid-19 pandemic. Critical reviews in clinical laboratory sciences, 57(6):365–388, 2020

  2. [2]

    Covid- 19 epidemic prediction and the impact of public health interventions: A review of covid-19 epidemic models

    Yue Xiang, Yonghong Jia, Linlin Chen, Lei Guo, Bizhen Shu, and Enshen Long. Covid- 19 epidemic prediction and the impact of public health interventions: A review of covid-19 epidemic models. Infectious Disease Modelling, 6:324–342, 2021

  3. [3]

    An enhanced seir model for prediction of covid-19 with vaccination effect

    Ramesh Chandra Poonia, Abdul Khader Jilani Saudagar, Abdullah Altameem, Mohammed Alkhathami, Muhammad Badruddin Khan, and Mozaherul Hoque Abul Hasanat. An enhanced seir model for prediction of covid-19 with vaccination effect. Life, 12(5):647, 2022

  4. [4]

    Fitting and forecasting the trend of covid-19 by seir (+ caq) dynamic model

    YY Wei, ZZ Lu, ZC Du, ZJ Zhang, Yang Zhao, SP Shen, Bo Wang, YT Hao, and Feng Chen. Fitting and forecasting the trend of covid-19 by seir (+ caq) dynamic model. Zhonghua liu xing bing xue za zhi= Zhonghua liuxingbingxue zazhi, 41(4):470–475, 2020

  5. [5]

    Mathematical modelling to inform new zealand’s covid-19 response

    Shaun Hendy, Nicholas Steyn, Alex James, Michael J Plank, Kate Hannah, Rachelle N Binny, and Audrey Lustig. Mathematical modelling to inform new zealand’s covid-19 response. Journal of the Royal Society of New Zealand, 51(sup1):S86–S106, 2021. 10

  6. [6]

    Bangladesh covid-19 daily cases time series analysis using facebook prophet model

    Sakib Mahmud. Bangladesh covid-19 daily cases time series analysis using facebook prophet model. Available at SSRN 3660368, 2020

  7. [7]

    Arima-based forecasting of the dynamics of confirmed covid-19 cases for selected european countries

    Tadeusz Kufel. Arima-based forecasting of the dynamics of confirmed covid-19 cases for selected european countries. Equilibrium. Quarterly Journal of Economics and Economic Policy, 15(2):181–204, 2020

  8. [8]

    Why is it difficult to accurately predict the covid-19 epidemic? Infectious disease modelling, 5:271–281, 2020

    Weston C Roda, Marie B Varughese, Donglin Han, and Michael Y Li. Why is it difficult to accurately predict the covid-19 epidemic? Infectious disease modelling, 5:271–281, 2020

Show all 49 references
  1. [9]

    Deep learning for epidemiolog- ical predictions

    Yuexin Wu, Yiming Yang, Hiroshi Nishiura, and Masaya Saitoh. Deep learning for epidemiolog- ical predictions. In The 41st international ACM SIGIR conference on research & development in information retrieval, pages 1085–1088, 2018

  2. [10]

    Prediction of epidemic trends in covid-19 with logistic model and machine learning technics

    Peipei Wang, Xinqi Zheng, Jiayang Li, and Bangren Zhu. Prediction of epidemic trends in covid-19 with logistic model and machine learning technics. Chaos, Solitons & Fractals , 139:110058, 2020

  3. [11]

    Forecasting prediction of covid-19 outbreak using linear regression

    Gurleen Kaur, Parminder Kaur, Navinderjit Kaur, and Prabhpreet Kaur. Forecasting prediction of covid-19 outbreak using linear regression. In Data Intelligence and Cognitive Informatics: Proceedings of ICDICI 2022, pages 195–221. Springer, 2022

  4. [12]

    A random forest model for forecasting regional covid-19 cases utilizing reproduction number estimates and demographic data

    Joseph Galasso, Duy M Cao, and Robert Hochberg. A random forest model for forecasting regional covid-19 cases utilizing reproduction number estimates and demographic data. Chaos, Solitons & Fractals, 156:111779, 2022

  5. [13]

    Application of a data- driven xgboost model for the prediction of covid-19 in the usa: a time-series study

    Zheng-gang Fang, Shu-qin Yang, Cai-xia Lv, Shu-yi An, and Wei Wu. Application of a data- driven xgboost model for the prediction of covid-19 in the usa: a time-series study. BMJ open, 12(7):e056685, 2022

  6. [14]

    Time series forecasting of covid-19 transmission in canada using lstm networks

    Vinay Kumar Reddy Chimmula and Lei Zhang. Time series forecasting of covid-19 transmission in canada using lstm networks. Chaos, solitons & fractals, 135:109864, 2020

  7. [15]

    Transfer graph neural networks for pandemic forecasting

    George Panagopoulos, Giannis Nikolentzos, and Michalis Vazirgiannis. Transfer graph neural networks for pandemic forecasting. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 4838–4845, 2021

  8. [16]

    Spatio-temporal graph neural networks: A survey

    Zahraa Al Sahili and Mariette Awad. Spatio-temporal graph neural networks: A survey. arXiv preprint arXiv:2301.10569, 2023

  9. [17]

    Spatio-temporal graph learning for epidemic prediction

    Shuo Yu, Feng Xia, Shihao Li, Mingliang Hou, and Quan Z Sheng. Spatio-temporal graph learning for epidemic prediction. ACM Transactions on Intelligent Systems and Technology, 14(2):1–25, 2023

  10. [18]

    A review of graph neural networks in epidemic modeling

    Zewen Liu, Guancheng Wan, B Aditya Prakash, Max SY Lau, and Wei Jin. A review of graph neural networks in epidemic modeling. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 6577–6587, 2024

  11. [19]

    Revolutionizing finance with llms: An overview of applications and insights

    Huaqin Zhao, Zhengliang Liu, Zihao Wu, Yiwei Li, Tianze Yang, Peng Shu, Shaochen Xu, Haixing Dai, Lin Zhao, Gengchen Mai, et al. Revolutionizing finance with llms: An overview of applications and insights. arXiv preprint arXiv:2401.11641, 2024

  12. [20]

    Autocas: Autoregressive cascade predictor in social networks via large language models

    Yuhao Zheng, Chenghua Gong, Rui Sun, Juyuan Zhang, Liming Pan, and Linyuan Lv. Autocas: Autoregressive cascade predictor in social networks via large language models. arXiv preprint arXiv:2502.18040, 2025

  13. [21]

    Llm multimodal traffic accident forecasting

    Irene de Zarzà, Joachim de Curtò, Gemma Roig, and Carlos T Calafate. Llm multimodal traffic accident forecasting. Sensors, 23(22):9225, 2023

  14. [22]

    Autotimes: Au- toregressive time series forecasters via large language models

    Yong Liu, Guo Qin, Xiangdong Huang, Jianmin Wang, and Mingsheng Long. Autotimes: Au- toregressive time series forecasters via large language models. Advances in Neural Information Processing Systems, 37:122154–122184, 2024. 11

  15. [23]

    Time-llm: Time series forecasting by reprogramming large language models

    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. Time-llm: Time series forecasting by reprogramming large language models. arXiv preprint arXiv:2310.01728, 2023

  16. [24]

    Real-time epidemic forecasting: challenges and opportunities

    Angel N Desai, Moritz UG Kraemer, Sangeeta Bhatia, Anne Cori, Pierre Nouvellet, Mark Herringer, Emily L Cohn, Malwina Carrion, John S Brownstein, Lawrence C Madoff, et al. Real-time epidemic forecasting: challenges and opportunities. Health security, 17(4):268–275, 2019

  17. [25]

    Advancing real-time pandemic forecasting using large language models: A covid-19 case study

    Hongru Du, Jianan Zhao, Yang Zhao, Shaochong Xu, Xihong Lin, Yiran Chen, Lauren M Gardner, and Hao Frank Yang. Advancing real-time pandemic forecasting using large language models: A covid-19 case study. arXiv preprint arXiv:2404.06962, 2024

  18. [26]

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

    Biao Zhang, Zhongtao Liu, Colin Cherry, and Orhan Firat. When scaling meets llm finetuning: The effect of data, model and finetuning method. arXiv preprint arXiv:2402.17193, 2024

  19. [27]

    Temporal multires- olution graph neural networks for epidemic prediction

    Truong Son Hy, Viet Bach Nguyen, Long Tran-Thanh, and Risi Kondor. Temporal multires- olution graph neural networks for epidemic prediction. In Workshop on Healthcare AI and COVID-19, pages 21–32. PMLR, 2022

  20. [28]

    What language model architecture and pretraining objective works best for zero-shot generalization? In International Conference on Machine Learning, pages 22964–22984

    Thomas Wang, Adam Roberts, Daniel Hesslow, Teven Le Scao, Hyung Won Chung, Iz Beltagy, Julien Launay, and Colin Raffel. What language model architecture and pretraining objective works best for zero-shot generalization? In International Conference on Machine Learning, pages 22...

  21. [29]

    An empirical study of autoregressive pre-training from videos

    Jathushan Rajasegaran, Ilija Radosavovic, Rahul Ravishankar, Yossi Gandelsman, Christoph Feichtenhofer, and Jitendra Malik. An empirical study of autoregressive pre-training from videos. arXiv preprint arXiv:2501.05453, 2025

  22. [30]

    Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing

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

  23. [31]

    Graph prompt learning: A comprehensive survey and beyond

    Xiangguo Sun, Jiawen Zhang, Xixi Wu, Hong Cheng, Yun Xiong, and Jia Li. Graph prompt learning: A comprehensive survey and beyond. arXiv preprint arXiv:2311.16534, 2023

  24. [32]

    A survey of time series foundation models: Generalizing time series representation with large language model

    Jiexia Ye, Weiqi Zhang, Ke Yi, Yongzi Yu, Ziyue Li, Jia Li, and Fugee Tsung. A survey of time series foundation models: Generalizing time series representation with large language model. arXiv preprint arXiv:2405.02358, 2024

  25. [33]

    Enhanced gaussian process regression-based forecasting model for covid-19 outbreak and significance of iot for its detection

    Shwet Ketu and Pramod Kumar Mishra. Enhanced gaussian process regression-based forecasting model for covid-19 outbreak and significance of iot for its detection. Applied Intelligence, 51(3):1492–1512, 2021

  26. [34]

    Exploring graph structure in graph neural networks for epidemic forecasting

    Ching-Hao Fan, Sai Supriya Varugunda, and Lijing Wang. Exploring graph structure in graph neural networks for epidemic forecasting. In Temporal Graph Learning Workshop@ NeurIPS 2023, 2023

  27. [35]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

  28. [36]

    Multiresolution equivariant graph variational autoencoder

    Truong Son Hy and Risi Kondor. Multiresolution equivariant graph variational autoencoder. Machine Learning: Science and Technology, 4(1):015031, 2023

  29. [37]

    Foundation models for time series analysis: A tutorial and survey

    Yuxuan Liang, Haomin Wen, Yuqi Nie, Yushan Jiang, Ming Jin, Dongjin Song, Shirui Pan, and Qingsong Wen. Foundation models for time series analysis: A tutorial and survey. In Proceedings of the 30th ACM SIGKDD conference on knowledge discovery and data mining, pages 6555–6565, 2024

  30. [38]

    Agents with foundation models: advance and vision

    Chenghua Gong and Xiang Li. Agents with foundation models: advance and vision. Frontiers of Computer Science, 19(4):194330, 2025. 12

  31. [39]

    Infectious disease forecasting in india using llm’s and deep learning

    Chaitya Shah, Kashish Gandhi, Javal Shah, Kreena Shah, Nilesh Patil, and Kiran Bhowmick. Infectious disease forecasting in india using llm’s and deep learning. arXiv preprint arXiv:2410.20168, 2024

  32. [40]

    Do we really need foundation models for multi-step-ahead epidemic forecasting? In NeurIPS Workshop on Time Series in the Age of Large Models, 2024

    Mrinmoy Dey, Aprameyo Chakrabartty, Dhruv Sarkar, and Tanujit Chakraborty. Do we really need foundation models for multi-step-ahead epidemic forecasting? In NeurIPS Workshop on Time Series in the Age of Large Models, 2024

  33. [41]

    Chronos: Learning the language of time series

    Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen, Xiyuan Zhang, Pedro Mercado, Huibin Shen, Oleksandr Shchur, Syama Sundar Rangapuram, Sebastian Pineda Arango, Shubham Kapoor, et al. Chronos: Learning the language of time series. arXiv preprint arXiv:2403.07815, 2024

  34. [42]

    Only the curve shape matters: Training foundation models for zero-shot multivariate time series forecasting through next curve shape prediction

    Cheng Feng, Long Huang, and Denis Krompass. Only the curve shape matters: Training foundation models for zero-shot multivariate time series forecasting through next curve shape prediction. arXiv preprint arXiv:2402.07570, 2024

  35. [43]

    Predicting covid-19 pandemic by spatio-temporal graph neural networks: A new zealand’s study

    Viet Bach Nguyen, Truong Son Hy, Long Tran-Thanh, and Nhung Nghiem. Predicting covid-19 pandemic by spatio-temporal graph neural networks: A new zealand’s study. arXiv preprint arXiv:2305.07731, 2023

  36. [44]

    Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting

    Bing Yu, Haoteng Yin, and Zhanxing Zhu. Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting. arXiv preprint arXiv:1709.04875, 2017

  37. [45]

    Language models are unsupervised multitask learners

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. Language models are unsupervised multitask learners. OpenAI blog, 1(8):9, 2019

  38. [46]

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948, 2025

  39. [47]

    Gemma 3 technical report

    Gemma Team, Aishwarya Kamath, Johan Ferret, Shreya Pathak, Nino Vieillard, Ramona Merhej, Sarah Perrin, Tatiana Matejovicova, Alexandre Ramé, Morgane Rivière, et al. Gemma 3 technical report. arXiv preprint arXiv:2503.19786, 2025

  40. [48]

    Adam: A method for stochastic optimization

    Diederik P Kingma. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

  41. [49]

    Neural message passing for quantum chemistry

    Justin Gilmer, Samuel S Schoenholz, Patrick F Riley, Oriol Vinyals, and George E Dahl. Neural message passing for quantum chemistry. In International conference on machine learning , pages 1263–1272. PMLR, 2017. 13 A Further Details of Datasets Table 4: The statistics informat...

Pith tools

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