Pith. sign in

REVIEW 5 major objections 6 minor 21 references

Deep Learning for GWP Prediction: A Framework Using PCA, Quantile Transformation, and Ensemble Modeling

T0 review · 5 major / 6 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 This paper claims that a fully connected neural network ensemble, trained on RDKit molecular descriptors with PCA and quantile transformation, predicts the 100-year global warming potential of refrigerants with a test RMSE of 481.9 and an…

desk verdict Small benchmark, plausible result, but unresolved preprocessing leakage and missing reproducibility make the generalization claim premature. read the letter →

arxiv 2411.19124 v1 pith:VZC6KZSW submitted 2024-11-28 cs.LG cond-mat.mtrl-sciphysics.chem-ph

Navin Rajapriya , Kotaro Kawajiri This is my paper · ORCID
classification cs.LGcond-mat.mtrl-sciphysics.chem-ph
keywords GWPpredictionrefrigerantsmoleculardescriptorsensembleneuralnetworkprincipalcomponentanalysisquantiletransformationvirtualscreeningdeeplearning
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 sets out to show that a deep learning framework can predict the 100-year global warming potential (GWP100) of single-component refrigerants accurately enough to be useful for virtual screening. Using 207 compounds from the IPCC AR6 dataset and three molecular descriptor packages, the authors find that RDKit descriptors combined with PCA, quantile transformation, and ensembling the top three models give a test-set RMSE of 481.9 and an R2 of 0.918. That result matters because high-GWP refrigerants are a major climate contributor, and a fast, descriptor-only predictor could let chemists screen candidate low-GWP molecules before synthesis. The paper also interprets which molecular features drive GWP, pointing to molecular weight, lipophilicity, nitriles, and allylic oxides.

What carries the argument

The load-bearing mechanism is the preprocessing-plus-ensemble pipeline: principal component analysis collapses the RDKit descriptors into 48 components while retaining 99% variance, quantile transformation maps the skewed GWP100 target to a uniform distribution for stable training, and the top three of up to ten auto-tuned fully connected neural networks are averaged to make the final prediction. The counterintuitive result is that the simplest descriptor set (RDKit, 2D) wins over Mordred (1826 features) and alvaDesc (5666 features), because the richer sets overfit the small 207-compound dataset.

What would settle it

Re-run the pipeline with PCA and quantile transformation estimated inside each cross-validation fold on the training portion only, keeping the same network architecture search; if the held-out RMSE moves substantially from 481.9 or the R2 falls below 0.9, the reported generalization claim is an artifact of preprocessing leakage.

Watch

Extended reading notes

Core claim

The central discovery, stated on the paper's own terms, is that simpler 2D molecular descriptors generalize better than richer 3D descriptor sets for GWP prediction: the RDKit-based ensemble reaches a test RMSE of 481.9 and an R2 of 0.918, whereas the Mordred and alvaDesc ensembles drop to R2 scores of 0.641 and 0.560. The authors attribute this drop to overfitting caused by the high complexity of 3D descriptors on a small dataset, and they argue that their full pipeline—PCA retaining 99% variance, quantile transformation of the skewed GWP target, and averaging the three best neural networks—produces a stable, scalable model for screening low-GWP refrigerants.

Load-bearing premise

The reported test accuracy is unbiased only if the PCA and quantile transformation were fitted on the training split alone and then applied to the test set; the paper does not explicitly say this.

Editorial extensions

If this is right

  • If the reported accuracy holds, chemists can virtually screen large libraries of candidate refrigerants and rank them by predicted GWP100 before expensive synthesis and measurement.
  • The framework can be extended to other environmental properties such as atmospheric lifetime and toxicity, as the paper itself suggests.
  • For small datasets, simpler 2D molecular descriptors should be preferred over complex 3D descriptor sets to avoid overfitting.
  • The factor analysis offers design rules: avoid features like high molecular weight and allylic oxide groups that raise GWP, and favor features like nitrile groups that may lower it.

Reading between the lines

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

  • The paper never states that PCA and quantile transformation were fitted only on the training split; if they were fitted on all 207 compounds, the reported test metrics are optimistic, and the cleanest way to check is nested cross-validation.
  • Because the dataset is small and uses a single random split, the large gap in R2 between descriptor packages could be an artifact of that split; repeated stratified cross-validation would tell whether the RDKit advantage is stable.
  • A direct head-to-head comparison against the traditional log-transform baseline on the same test set would isolate how much of the gain comes from quantile transformation rather than from the neural network architecture or ensemble.
  • The factor analysis is based on permutation sensitivity of one ensemble; checking whether the same PCs dominate across the three individual ensemble members would show whether the chemical interpretation is robust.
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

5 major / 6 minor

Summary. The manuscript proposes a pipeline for predicting 100-year global warming potential (GWP100) of single-component refrigerants using molecular descriptors (RDKit, Mordred, alvaDesc), PCA (retaining 99% variance), quantile transformation, and an ensemble of fully connected neural networks trained on the Multi-Sigma platform. The authors report that the RDKit-based ensemble achieves an RMSE of 481.9 and an R2 of 0.918 on a common held-out test set and claim 'high predictive accuracy and generalizability.' They also apply factor analysis to identify principal components linked to molecular weight, lipophilicity, and functional groups. The paper's core contribution is an empirical benchmark demonstrating that descriptor-based ensemble neural networks can be used for low-GWP refrigerant screening.

Significance. If the reported results are unbiased, the framework is a plausible candidate for virtual screening, with practical relevance for Kigali Amendment compliance. The paper has several strengths: it compares three descriptor packages on a common test set, provides a hyperparameter table for all top models, uses an explicit preprocessing rationale (PCA vs. autoencoders), and offers an interpretability analysis linking PCs to molecular descriptors. The main threats are possible leakage in preprocessing, the lack of uncertainty quantification, and the absence of baselines; these must be addressed before the generalizability claim is credible. The empirical nature and small dataset (207 compounds) mean the contribution is incremental but potentially useful if made reproducible.

major comments (5)
  1. [Section 2.2] The manuscript never states whether the PCA projection and the quantile transformer were fitted on the training split only or on the full 207-compound dataset. Because the test set is held out to measure generalization, fitting these transforms on all data would leak test information into the model and make the reported test RMSE 481.9 and R2 0.918 (Section 3.1.2) optimistically biased. This must be clarified and, if the transforms were fitted on the full dataset, the evaluation must be redone on a properly nested split.
  2. [Section 3.1.2] The test evaluation consists of a single 10% split (~21 compounds) with no confidence intervals, no repeated runs, and no alternative seeds. The claim that the RDKit ensemble 'indicates high predictive accuracy and generalizability' is too strong for one small split, especially since Table 1 shows Mordred Model 1 with a validation RMSE of 310.57 and R2 of 1.0, yet the Mordred ensemble test RMSE is 1011.28. The authors should report error bars, e.g., from bootstrap or multiple random splits, and show that the RDKit advantage is stable.
  3. [Overall] The paper states that the RDKit-based ensemble achieved 'superior' performance, but no baseline (e.g., group-contribution methods, random forest, linear regression, or a log-transform neural network) is evaluated on the same test split. Without a comparison, the framework's added value over standard QSPR methods is not established. Add at least one standard baseline trained and evaluated under the same splitting protocol.
  4. [Section 3.2] The factor analysis is purely descriptive. The text itself uses 'We speculate' to connect PC10/PC3/PC4 loadings to atmospheric reactivity, and the loadings in Table 2 are not statistically validated (no confidence intervals, no permutation test). These interpretative claims should be explicitly labeled as hypotheses, not as validated mechanisms.
  5. [Reproducibility] No code, data, or clear reproducibility statement is provided. Given that Multi-Sigma is a proprietary platform and the preprocessing order is critical to the central claim, the authors should share the dataset (SMILES and GWP values), the exact train/validation/test splits, and the preprocessing scripts, or at least a detailed pseudocode of the fit/transform steps.
minor comments (6)
  1. [Section 3.1 heading] The heading 'Quantile T ransformation' contains an extra space; fix to 'Quantile Transformation.'
  2. [Throughout] The descriptor package is sometimes capitalized as 'AlvaDesc' (e.g., Section 3.1) and sometimes written as 'alvaDesc'; unify the naming.
  3. [Section 3.1.1] The phrase 'exploratory variables' should be 'explanatory variables' when referring to the PCA-transformed inputs.
  4. [Table 1] The R2(Orig)=1.0 entries for Mordred Model 1 and alvaDesc Model 1 are suspicious; with RMSE values of 310.57 and 168.15, an R2 of exactly 1.0 likely indicates rounding or an inconsistent computation. Please verify and report these metrics to a consistent number of digits.
  5. [Figure 4] The caption says the figure shows 'the predictions made in the testing phase using the ensemble models, and individual model validation during the modeling phase,' which is ambiguous. It should clearly label which panels are training/validation predictions and which are test predictions.
  6. [Section 2.1] Please cite or describe how the 207 GWP100 values were extracted from the IPCC AR6 dataset; the current reference [13] is the full report and does not provide a machine-readable data source.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the supervised ML pipeline is fit on training data and evaluated on a held-out test set, with no prediction reducing to its inputs by construction.

full rationale

Walking the claimed derivation chain—molecular descriptors, PCA, quantile transformation, ensemble neural network, test-set evaluation, and post hoc factor analysis—I find no step where an output is defined in terms of an input, no fitted parameter is relabeled as a prediction, and no load-bearing conclusion rests on a self-citation. GWP labels come from the external IPCC AR6 dataset; descriptors are computed from molecular structure; the quantile transform is standard supervised-learning preprocessing; and the ensemble is selected on validation RMSE then evaluated on a common test set. The factor analysis interprets learned PC loadings after training rather than supplying an input to training. The main weakness is an ambiguity in Section 2.2: the paper does not explicitly state that PCA and the quantile transformer were fitted on the training split only, so if they were fitted on all 207 compounds the reported test RMSE of 481.9 and R2 of 0.918 could be optimistically biased. That is a data-leakage and reproducibility concern, not a circularity of the kind defined here, because it would require an unstated implementation choice rather than a reduction visible in the paper's own equations. No circular step is therefore reported.

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

The central claim depends on three unproven inputs: the reliability and representativeness of IPCC AR6 GWP labels, the sufficiency of molecular descriptors as proxies for atmospheric chemistry, and the unstated assumption that PCA and quantile transforms were fit without test-set information. The model also depends on the chosen PCA threshold, auto-tuned network hyperparameters, and the rule for ensembling three models; these are fit or chosen by hand, not derived from theory. No new physical entities are introduced.

free parameters (3)
  • PCA variance retention threshold = 99%
    Chosen by hand in Section 2.2; it determines 48, 73, and 99 principal components for the three descriptor packages, and this choice directly shapes model inputs.
  • Neural network hyperparameters = Varies by model; examples in Table 1, such as RDKit Model 1 with 3 layers, 82 neurons, batch 75, 2674 epochs, tanh…
    Multi-Sigma auto-tuning selects layers, neurons, activation, batch size, and epochs by validation RMSE; these are fit to the data and differ across descriptor packages.
  • Top-3 ensemble selection rule = 3 models per descriptor package with lowest validation RMSE
    The choice to keep three models and average them is discretionary; changing this rule changes reported performance.
assumptions (3)
  • domain assumption GWP100 values from IPCC AR6 are accurate and complete for the selected 207 compounds.
    Used as ground truth throughout Sections 2.1 and 3; label errors or curation bias propagate directly to model metrics.
  • domain assumption Molecular descriptors computed by RDKit, Mordred, and alvaDesc capture the structure-property relationships that determine GWP.
    The entire predictive framework assumes descriptors are sufficient proxies for radiative forcing and atmospheric lifetime; invoked in Sections 2.1 and 3.2.
  • ad hoc to paper PCA and quantile transformation are fitted on the training split only.
    Section 2.2 applies these transforms without stating test-set exclusion; the central test metrics are unbiased only under this unstated assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning for GWP Prediction: A Framework Using PCA, Quantile Transformation, and Ensemble Modeling." pith.science (2026). https://pith.science/paper/VZC6KZSW

@misc{pith2026241119124,
  author       = {Pith},
  title        = {Pith review of: Deep Learning for GWP Prediction: A Framework Using PCA, Quantile Transformation, and Ensemble Modeling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VZC6KZSW}},
  note         = {Machine review of arXiv:2411.19124}
}
read the original abstract

Developing environmentally sustainable refrigerants is critical for mitigating the impact of anthropogenic greenhouse gases on global warming. This study presents a predictive modeling framework to estimate the 100-year global warming potential (GWP 100) of single-component refrigerants using a fully connected neural network implemented on the Multi-Sigma platform. Molecular descriptors from RDKit, Mordred, and alvaDesc were utilized to capture various chemical features. The RDKit-based model achieved the best performance, with a Root Mean Square Error (RMSE) of 481.9 and an R2 score of 0.918, demonstrating superior predictive accuracy and generalizability. Dimensionality reduction through Principal Component Analysis (PCA) and quantile transformation were applied to address the high-dimensional and skewed nature of the dataset,enhancing model stability and performance. Factor analysis identified vital molecular features, including molecular weight, lipophilicity, and functional groups, such as nitriles and allylic oxides, as significant contributors to GWP values. These insights provide actionable guidance for designing environmentally sustainable refrigerants. Integrating RDKit descriptors with Multi-Sigma's framework, which includes PCA, quantile transformation, and neural networks, provides a scalable solution for the rapid virtual screening of low-GWP refrigerants. This approach can potentially accelerate the identification of eco-friendly alternatives, directly contributing to climate mitigation by enabling the design of next-generation refrigerants aligned with global sustainability objectives.

Figures

Figures reproduced from arXiv: 2411.19124 by the authors.

Figure 1
Figure 1. Framework for predicting GWP values using molecular descriptors, PCA, quantile transformation, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Distribution of GWP100 values before and after quantile transformation: (a) Original scale (b) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Cumulative explained variance as a function of the number of PCs for each molecular descriptor [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Predicted vs. True GWP 100 values for the top three models and ensemble predictions for each [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Contribution of principal components (PCs) to GWP predictions based on factor analysis of [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 20 canonical work pages

  1. [1]

    Experimental analysis of R-450A and R-513A as replacements of R-134a and R-507A in a medium temperature commercial refrigeration system,

    R. Llopis, D. Sánchez, R. Cabello, J. Catalán-Gil, and L. Nebot-Andrés, “Experimental analysis of R-450A and R-513A as replacements of R-134a and R-507A in a medium temperature commercial refrigeration system,” Int. J. Refrig., vol. 84, pp. 52–66, Dec. 2017, doi: 10.1016/j.ijrefrig.2017.08.022

    work page doi:10.1016/j.ijrefrig.2017.08.022 2017

  2. [2]

    A review of bottom-up and top-down emission estimates of hydrofluorocarbons (HFCs) in different parts of the world,

    H. Flerlage, G. J. M. Velders, and J. De Boer, “A review of bottom-up and top-down emission estimates of hydrofluorocarbons (HFCs) in different parts of the world,” Chemosphere, vol. 283, p. 131208, Nov. 2021, doi: 10.1016/j.chemosphere.2021.131208

    work page arXiv 2021

  3. [3]

    Electricity savings and greenhouse gas emission reductions from global phase-down of hydrofluorocarbons,

    P. Purohit et al., “Electricity savings and greenhouse gas emission reductions from global phase-down of hydrofluorocarbons,” Mar. 11, 2020. doi: 10.5194/acp-2020-193

    work page doi:10.5194/acp-2020-193 2020

  4. [4]

    Future atmospheric abundances and climate forcings from scenarios of global and regional hydrofluorocarbon (HFC) emissions,

    G. J. M. Velders, D. W. Fahey, J. S. Daniel, S. O. Andersen, and M. McFarland, “Future atmospheric abundances and climate forcings from scenarios of global and regional hydrofluorocarbon (HFC) emissions,” Atmos. Environ., vol. 123, pp. 200–209, Dec. 2015, doi: 10.1016/j.atmosenv.2015.10.071

    work page doi:10.1016/j.atmosenv.2015.10.071 2015

  5. [5]

    Machine learning for sustainable development: leveraging technology for a greener future,

    M. Kagzi, S. Khanra, and S. K. Paul, “Machine learning for sustainable development: leveraging technology for a greener future,” J. Syst. Inf. Technol., vol. 25, no. 4, pp. 440–479, Nov. 2023, doi: 10.1108/JSIT-11-2022-0266

    work page doi:10.1108/jsit-11-2022-0266 2023

  6. [6]

    Searching for Sustainable Refrigerants by Bridging Molecular Modeling with Machine Learning,

    I. I. I. Alkhatib, C. G. Albà, A. S. Darwish, F. Llovell, and L. F. Vega, “Searching for Sustainable Refrigerants by Bridging Molecular Modeling with Machine Learning,” Ind. Eng. Chem. Res., vol. 61, no. 21, pp. 7414–7429, Jun. 2022, doi: 10.1021/acs.iecr.2c00719

    work page doi:10.1021/acs.iecr.2c00719 2022

  7. [7]

    Prediction of global warming potentials of refrigerants and related compounds from their molecular structure – An artificial neural network with group contribution method,

    S. Devotta, A. Chelani, and A. Vonsild, “Prediction of global warming potentials of refrigerants and related compounds from their molecular structure – An artificial neural network with group contribution method,” Int. J. Refrig., vol. 131, pp. 756–765, Nov. 2021, doi: 10.1016/j.ijrefrig.2021.08.011

    work page doi:10.1016/j.ijrefrig.2021.08.011 2021

  8. [8]

    Group contribution-based property estimation methods: advances and perspectives,

    R. Gani, “Group contribution-based property estimation methods: advances and perspectives,” Curr. Opin. Chem. Eng., vol. 23, pp. 184–196, Mar. 2019, doi: 10.1016/j.coche.2019.04.007

    work page doi:10.1016/j.coche.2019.04.007 2019

Show all 21 references
  1. [9]

    Comparing predictive ability of QSAR/QSPR models using 2D and 3D molecular representations,

    A. Sato, T. Miyao, S. Jasial, and K. Funatsu, “Comparing predictive ability of QSAR/QSPR models using 2D and 3D molecular representations,” J. Comput. Aided Mol. Des., vol. 35, no. 2, pp. 179–193, Feb. 2021, doi: 10.1007/s10822-020-00361-7

    2021 doi

  2. [10]

    Improved Machine Learning Models by Data Processing for Predicting Life-Cycle Environmental Impacts of Chemicals,

    Y. Sun, X. Wang, N. Ren, Y. Liu, and S. You, “Improved Machine Learning Models by Data Processing for Predicting Life-Cycle Environmental Impacts of Chemicals,” Environ. Sci. Technol., vol. 57, no. 8, pp. 3434–3444, Feb. 2023, doi: 10.1021/acs.est.2c04945

    2023 doi

  3. [11]

    Effect of Molecular Descriptor Feature Selection in Support Vector Machine Classification of Pharmacokinetic and Toxicological Properties of Chemical Agents,

    Y. Xue, Z. R. Li, C. W. Yap, L. Z. Sun, X. Chen, and Y. Z. Chen, “Effect of Molecular Descriptor Feature Selection in Support Vector Machine Classification of Pharmacokinetic and Toxicological Properties of Chemical Agents,” J. Chem. Inf. Comput. Sci., vol. 44, no. 5, pp. 1630...

    2004 doi

  4. [12]

    Leveraging Machine Learning To Predict the Atmospheric Lifetime and the Global Warming Potential of SF 6 Replacement Gases,

    G. Zhao, H. Kim, C. Yang, and Y. G. Chung, “Leveraging Machine Learning To Predict the Atmospheric Lifetime and the Global Warming Potential of SF 6 Replacement Gases,” J. Phys. Chem. A, vol. 128, no. 12, pp. 2399–2408, Mar. 2024, doi: 10.1021/acs.jpca.3c07339

    2024 doi

  5. [13]

    Cambridge: Cambridge University Press, 2023

    Intergovernmental Panel on Climate Change (IPCC), Climate Change 2021 – The Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press, 2023. doi: 10.1017/9781009157896. 9

    2021 doi

  6. [14]

    Scikit-learn: Machine Learning in Python,

    F. P. Duchesnay Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, Jake Vanderplas, Alexandre Passos, David Cournapeau, Matthieu Brucher, Matthieu Perrot, Édouard, “Scikit-learn...

    2011

  7. [15]

    Using principal component analysis for neural network high-dimensional potential energy surface,

    B. Casier, S. Carniato, T. Miteva, N. Capron, and N. Sisourat, “Using principal component analysis for neural network high-dimensional potential energy surface,” J. Chem. Phys., vol. 152, no. 23, p. 234103, Jun. 2020, doi: 10.1063/5.0009264

    2020 doi

  8. [16]

    Ensemble Learning Models for Food Safety Risk Prediction,

    L.-Y. Wu and S.-S. Weng, “Ensemble Learning Models for Food Safety Risk Prediction,” Sustainability, vol. 13, no. 21, p. 12291, Nov. 2021, doi: 10.3390/su132112291

    2021 doi

  9. [17]

    Forecasting Corn Yield With Machine Learning Ensembles,

    M. Shahhosseini, G. Hu, and S. V. Archontoulis, “Forecasting Corn Yield With Machine Learning Ensembles,” Front. Plant Sci., vol. 11, p. 1120, Jul. 2020, doi: 10.3389/fpls.2020.01120

    2020

  10. [18]

    A new approach for the vanishing gradient problem on sigmoid activation,

    M. Roodschild, J. Gotay Sardiñas, and A. Will, “A new approach for the vanishing gradient problem on sigmoid activation,” Prog. Artif. Intell., vol. 9, no. 4, pp. 351–360, Dec. 2020, doi: 10.1007/s13748-020- 00218-y

    2020 doi

  11. [19]

    Infrared band intensities and global warming potentials of CF4 , C2 F6 , C3 F8 , C4 F10 , C5 F12 , and C6 F14,

    C. M. Roehl, D. Boglu, C. Brühl, and G. K. Moortgat, “Infrared band intensities and global warming potentials of CF4 , C2 F6 , C3 F8 , C4 F10 , C5 F12 , and C6 F14,” Geophys. Res. Lett., vol. 22, no. 7, pp. 815–818, Apr. 1995, doi: 10.1029/95GL00488

    1995 doi

  12. [20]

    The stability of nitrogen-centered radicals,

    J. Hioe, D. Šakić, V. Vrček, and H. Zipse, “The stability of nitrogen-centered radicals,” Org. Biomol. Chem., vol. 13, no. 1, pp. 157–169, 2015, doi: 10.1039/C4OB01656D

    2015 doi

  13. [21]

    Chemistry of Volatile Organic Compounds in the Atmosphere,

    R. Koppmann, “Chemistry of Volatile Organic Compounds in the Atmosphere,” in Hydrocarbons, Oils and Lipids: Diversity, Origin, Chemistry and Fate, H. Wilkes, Ed., Cham: Springer International Publishing, 2020, pp. 811–822. doi: 10.1007/978-3-319-90569-3_24. Acknowledgments The...

    2020 doi

Pith tools

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