A Deep Generative Model for the Design of Synthesizable Ionizable Lipids

Austin Tripp; Jingyi Zhao; Jos\'e Miguel Hern\'andez-Lobato; Morteza Rasoulianboroujeni; Yuxuan Ou

REVIEW 4 major objections 5 minor 15 references

A Deep Generative Model for the Design of Synthesizable Ionizable Lipids

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 This paper shows that a DAG-based generative model trained on a lipid-specific synthesis dataset and paired with the Chemformer reaction predictor produces ionizable lipids with synthesis routes at an 83.4% rate, and that iterative…

desk verdict A useful, honestly-written adaptation of Synthesis-DAGs to ionizable lipids, but the 'synthesizable' claim hinges on an unvalidated reaction predictor. read the letter →

arxiv 2412.00928 v1 pith:WY6QZUSK submitted 2024-12-01 cs.LG cs.AI

Yuxuan Ou , Jingyi Zhao , Austin Tripp , Morteza Rasoulianboroujeni , José Miguel Hernández-Lobato This is my paper · ORCID

classification cs.LGcs.AI

keywords ionizablelipidslipidnanoparticlesgenerativemodelsynthesisDAGsreactionpredictiontransfectionefficiencymRNAdeliverysynthesizability

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 is trying to establish that generative design of ionizable lipids—the delivery component that lets mRNA vaccines and gene therapies enter cells—can be both diverse and synthesizable if each candidate is generated together with its own synthesis route. Small-molecule generators do not transfer directly to lipids, because lipid heads and tails are large, amphiphilic, and under-represented in generic reaction data. The authors assemble a lipid-specific dataset of over 70,000 synthesis paths from commercially available building blocks, adapt a DAG-based molecular generator to this domain, and swap in a stronger reaction predictor. The adapted generator classifies 92.6% of its sampled products as lipids and 83.4% as ionizable lipids, with near-perfect validity, uniqueness, and novelty. Iterative fine-tuning against a learned transfection-efficiency predictor then shifts sampling toward lipids with higher predicted mRNA delivery in HeLa cells, while the generated products remain tied to explicit synthesis paths.

What carries the argument

The load-bearing object is the synthesis-DAG generator, a model that serializes the construction of a molecule's synthesis route into three action types—node addition, building-block identity, and connectivity choice—and uses a shared RNN to model the action sequence, with a reaction predictor applied at each product node. The paper's adaptation adds two ingredients: a lipid-specific training dataset of 70,536 synthesis paths assembled from ZINC20-derived heads and tails, and Chemformer, a pretrained transformer reaction predictor that avoids the copy-paste errors the older Molecular Transformer makes on large lipid reactants. These are supported by two filters: a Chemprop-based lipid classifier and an ionizability check that estimates pKa with MolGpka and applies the Henderson-Hasselbalch equation to require a positive net charge at acidic pH and near-neutral charge at physiological pH.

What would settle it

Take a sample of the generated synthesis DAGs, run the proposed reactions in the lab, and compare the products by mass spectrometry or NMR; if a substantial fraction of the 'successful' paths fail to yield the predicted ionizable lipid, the 83.4% rate would be a computational artifact rather than a synthesizability result.

Watch

Extended reading notes

Core claim

The central claim is that the right representation plus the right domain data is what makes ionizable-lipid generation work. By serializing synthesis routes as directed acyclic graphs and training an RNN on a dataset of 70,536 lipid-specific synthesis paths built from 2.7 million candidate heads and 15,302 tails, the adapted generator (DAG+Chem) reaches an ionizable-lipid rate of 83.4%, compared with 78.3% for a linear-list representation, 50.7% for the same DAG generator paired with the older Molecular Transformer reaction predictor, and 69.2% for random head-tail pairing with Chemformer. The generated molecules are valid, unique, and novel at rates near 1.0, with a Fréchet ChemNet Distance of 3.797, indicating the new lipids stay inside the chemical and biological distribution of the training set. When restricted to two-tail lipids and fine-tuned on AGILE's HeLa transfection-efficiency predictions, the model raises the predicted efficiency of its top 1,000 samples over several iterations, though the trend is not strictly monotonic.

Load-bearing premise

The pipeline's synthesizability claim rests on trusting Chemformer's reaction predictions and the lipid and ionizability classifiers without any wet-lab check; the paper itself states that the validity of the proposed synthesis pathways has not been evaluated.

Editorial extensions

If this is right

Ionizable-lipid design can be treated as an open-ended generative problem rather than a search over fixed head/tail libraries, because every sampled lipid comes with an explicit synthesis route.
The reaction predictor is the main bottleneck: replacing Molecular Transformer with Chemformer more than doubles the ionizable-lipid rate (50.7% to 83.4%) for the same DAG generator.
The DAG representation of synthesis routes carries useful inductive bias over a linear list, yielding both a higher ionizable-lipid rate and a lower Fréchet ChemNet Distance (3.797 vs 4.119).
Iterative fine-tuning on a learned transfection-efficiency predictor can shift the generator toward lipids with higher predicted mRNA delivery in HeLa cells while the synthesis-path constraint is retained, although later iterations can regress.
The generators produce molecules that are novel and unique while staying close to the training distribution, so the approach generalizes rather than memorizing the dataset.

Reading between the lines

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

The 83.4% and 92.6% rates are software-level validity rates: they measure whether the reaction predictor and classifiers accept the product, not whether a chemist can actually make it, and the paper explicitly says the synthesis pathways were not experimentally evaluated.
Because the building block pool comes from ZINC20 and the generator produces only one-to-three-tail lipids, the diversity of the output is bounded by those choices; broadening the head/tail pool or allowing branching chemistry should extend the accessible lipid space.
The optimization loop is only as trustworthy as AGILE's HeLa predictions; if those predictions transfer poorly to other cell types or in vivo conditions, the top-ranked lipids may not be the best candidates.
The architecture is reaction-agnostic in principle, so the same DAG-plus-reaction-predictor recipe could be reused for other modular molecular classes, such as polymers or peptoids, provided a domain-specific synthesis dataset is assembled.

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.

Desk Editor's Note

read the letter

Read it this morning. The thing you should know first: it's a solid engineering contribution, not a breakthrough. The authors adapt Synthesis-DAGs to ionizable lipids by building a new dataset of 70,536 synthesis paths from ZINC building blocks and swapping the reaction predictor from Molecular Transformer to Chemformer. They show that this combination beats baselines on lipid rate and ionizable lipid rate (83.4% vs 50.7% for DAG+MT and 69.2% for random+Chem), with good validity, uniqueness, novelty, and FCD scores. The iterative fine-tuning with AGILE shows a sensible trend in predicted transfection efficiency. The paper is clearly written and the baselines are fair. Credit where due: the dataset construction is careful, the property predictors are validated on an independent set (98.32% accuracy for the lipid classifier, perfect ionizability classification on over 2,500 published ionizable lipids), and the authors explicitly acknowledge in the conclusion that synthesis-path validity has not been evaluated. That honesty is refreshing.

Now the soft spots. The stress-test note is right, and it lands hard. Chemformer is used both to construct the training dataset (Section 4.2) and to predict products during sampling (Section 5.2). The paper shows only two anecdotal examples that Chemformer works on lipid reactions; there is no held-out accuracy or validity rate for Chemformer on the domain it defines. If Chemformer makes systematic errors on large lipid molecules—the exact regime where the paper shows Molecular Transformer fails—those errors are baked into both the training targets and the sampled outputs. So the 83.4% ionizable-lipid rate measures agreement with a pipeline whose chemical ground truth is unverified. The paper concedes this but frames it as future work; the stress-test note is correct that this is a correctness risk, not merely a missing wet-lab confirmation. An inaccurate reaction predictor would invalidate the DAGs on purely computational grounds. Second, there are no error bars anywhere, which the paper acknowledges as a computational-cost decision. For the main table that's a minor issue—the gaps are large—but it matters for the fine-tuning comparison. Third, the AGILE-based optimization uses a surrogate transfection predictor; the paper is appropriately cautious there, but readers should treat the efficiency gains as untested.

Is this paper worth a serious referee? Yes. The adaptation is nontrivial, the dataset is potentially reusable, the comparison stories are informative, and the limitations are addressed head-on. I'd send it to review, with the expectation that the authors add a held-out evaluation of Chemformer on lipid reactions and frame the synthesizability claim accordingly. It's not ready to guide wet-lab synthesis yet, but it's a reasonable foundation.

My summary for you: bring it to a reading group focused on generative chemistry or drug delivery, and if you work in that space, cite it as the current best effort for ionizable lipid generation with synthesis paths, but not as proof that those paths work.

Referee Report

4 major / 5 minor

Summary. The paper adapts the Synthesis-DAGs framework to generate ionizable lipids together with explicit synthesis paths. It constructs a lipid-specific synthesis dataset (70,536 paths) from ZINC20-derived head and tail building blocks, using Chemformer as the reaction predictor and two Chemprop-based classifiers (lipid and ionizable lipid) to define and filter valid products. The proposed DAG+Chem generator is compared with a linear-list variant, the original Synthesis-DAGs with Molecular Transformer, and random generation; it reports an 83.4% ionizable lipid rate, high validity/uniqueness/novelty, and a lower FCD than the other methods. The paper also iteratively fine-tunes the generator with the AGILE predictor to increase predicted mRNA transfection efficiency in HeLa cells. The central claim is that the method generates structurally diverse, synthesizable ionizable lipids with synthesis paths.

Significance. The potential contribution is real: the paper provides a large lipid-specific synthesis-path dataset, adapts a principled generative framework to a new domain, and reports a broad set of quality metrics. The public code, reproducibility-oriented checklist, and comparison against multiple baselines are strengths. However, the main claims are conditional on external validation of the reaction predictor and property predictors. In particular, the 'synthesizable' claim is currently supported only by an unverified computational pipeline, as the paper itself acknowledges. If the missing validation is supplied, this would be a useful advance for computational lipid design.

major comments (4)

[§5.2 and Conclusion] The paper never measures Chemformer's accuracy on lipid or lipid-like reactions. Chemformer is used twice in the same role: §4.2 calls Chemformer to generate the product labels that define the training synthesis paths, and §5.2 uses Chemformer again to predict product nodes at sampling time. Any systematic error on large lipid molecules (the regime in which the paper shows Molecular Transformer fails, Figure 2) is therefore present in both the training targets and the sampled outputs, and the reported 83.4% ionizable lipid rate in Table 2 measures agreement with a pipeline whose chemical ground truth is unverified. The conclusion explicitly states 'the validity of the proposed synthesis pathways has not been evaluated in this work.' To support the 'synthesizable' claim, the authors should report Chemformer's top-1/top-k product-prediction accuracy, product validity rate, and exact-match rate on a held-out set of lipid or lipid-like reactions, or otherwise provide independent chemical evidence for the predicted products.
[§4.2, §6.3, and §3] The lipid classifier and ionizable lipid classifier are used both to filter the training data and to compute the headline lipid rate and ionizable lipid rate at evaluation. A DAG generator trained only on molecules that pass these filters can achieve high rates by reproducing the filters' inductive bias, so these numbers are partially circular as evidence of chemical validity. The validation on 2,500 published ionizable lipids is helpful, but it covers positive examples only; no negative controls or false-positive rates on the generated distribution are reported. Please evaluate classifier agreement on the generated set separately, include non-ionizable negatives in the validation, and complement the learned classifiers with explicit pKa/charge checks or expert chemical review of sampled structures.
[Tables 2 and 3, §6.2] All generation-quality metrics are single-run point estimates with no error bars, repeated seeds, or statistical significance tests. The differences between methods (e.g., DAG+Chem 83.4% vs. List+Chem 78.3% ionizable-lipid rate) could be within sampling noise. Also, Random+Chem uses 309,075 samples while the other rows use roughly 14,000, so the rate comparisons are not matched. Please report standard errors or confidence intervals over multiple training and sampling runs, or at least bootstrap intervals over generated samples, and match the number of evaluated samples across methods.
[§6.4 and Figure 4] The optimization experiment reports the distribution of AGILE-predicted transfection efficiencies, not measured transfection efficiency. Since the top 1,000 DAGs are selected by AGILE and then used for fine-tuning, the upward shift in predicted scores is expected and may reflect overfitting to the predictor rather than genuinely improved lipids. The paper should state more explicitly that this is an in silico demonstration, and should report the predicted scores on an independent validation set, the diversity and novelty of the optimized lipids, and the fraction that still satisfy the ionizable-lipid filters. The non-monotonic behavior between iterations 3 and 4 also deserves discussion rather than a one-sentence remark.

minor comments (5)

[§4.1] The selection thresholds (MW ≤ 500, LogP < 0, Tanimoto > 0.5, 1–3 functional groups) are presented without justification or sensitivity analysis; a short ablation or literature justification would help the reader assess how robust the building-block pool is.
[§6.1] For the Random+Chem baseline, the text says it 'generates raw training data without filtering,' but Table 2 reports its lipid rate and ionizable lipid rate; clarify that these rates are computed after sampling with the same classifiers.
[§6.2] The paper does not report the RNN architecture details (number of layers, hidden size, action embedding size) or Chemformer inference settings; please include them or point to the public code.
[NeurIPS Checklist] The checklist states that all source code is publicly available, but no URL or repository identifier appears in the paper; please add it.
[Figure 4] The text refers to a 'decline is observed between iteration 4 and iteration 3'; this should read 'from iteration 3 to iteration 4' for clarity.

Circularity Check

1 steps flagged · score 4.0 of 10

Chemformer defines both the training reactions and the sampled synthesis paths, so the reported "synthesizable" lipid success rate validates the generator against its own label-generating oracle.

self definitional [Sections 4.2, 5.2, and 7 (dataset construction, reaction predictor choice, conclusion)]
"The chemical reactions that combine a lipid tail with a lipid head or an intermediate product are simulated using Chemformer [Irwin et al., 2022], a reaction prediction model. ... To address the limitations of Molecular Transformer, we adopted Chemformer ... the validity of the proposed synthesis pathways has not been evaluated in this work."

The training synthesis paths are defined by Chemformer's product predictions (Section 4.2), and during sampling Chemformer again computes every product node (Sections 5.2 and 6.2). The reported 83.4% success rate in generating "synthesizable" ionizable lipids is therefore not a check against chemical ground truth; it measures the generator's ability to propose reactant combinations that the same Chemformer model accepts. Any systematic Chemformer error on large lipid reactions is identically present in the training labels and in the sampled outputs, so it cannot be detected or corrected by this pipeline. The conclusion concedes that pathway validity was not evaluated, confirming that the synthesizability claim is self-referential rather than established by an independent oracle.

full rationale

The paper is not a formal derivation, so the circularity question is whether its central empirical claims are validated by an oracle that also produced the training signal. The strongest concern is the reaction predictor: Section 4.2 constructs the ionizable lipid synthesis dataset by simulating reactions with Chemformer, and Section 5.2 selects Chemformer as the sampling-time reaction predictor. Thus the generated synthesis paths and the training paths share the same unverified biochemical authority; the paper explicitly states that the validity of proposed synthesis pathways has not been evaluated. This is a genuine self-referential validation of the "synthesizable" headline claim, but it is not a complete reduction of the method's contribution: the DAG generator itself is a separately trained policy, the lipid classifiers are validated on independent data, and the comparisons against Random+Chem and DAG+MT remain meaningful conditional on Chemformer. There is no load-bearing self-citation chain or imported uniqueness theorem. The score reflects partial circularity in the synthesizability evaluation rather than full equivalence of derivation and input.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The paper relies on several domain assumptions about the reliability of computational predictions (reaction prediction, pKa prediction, transfection efficiency prediction) and on heuristic thresholds for dataset construction. It introduces no fundamentally new entities or forces. The free parameters are all design choices in the data pipeline and optimization loop, none of which are justified by theory or sensitivity analysis.

free parameters (6)

LogP threshold for head selection = 0
Used to filter ionizable heads from ZINC20. Chosen to favor hydrophilic heads. The threshold is a design decision, not derived from first principles. The paper does not provide sensitivity analysis.
Molecular weight threshold for heads = 500 g/mol
Used to filter heads. The paper notes most lipids have MW <= 1000, but sets a stricter limit of 500. The choice is arbitrary and not justified by data.
Tanimoto similarity threshold for tails = 0.5
Used to find similar lipid tails in ZINC. The threshold is chosen to balance accessibility and diversity. No sensitivity analysis is provided.
Number of functional groups required in heads = 1-3
Used to ensure heads can react with tails. The range is chosen based on the observed functional groups in tails. This is a heuristic and not derived from a formal analysis.
Fine-tuning iteration count (sampling top 1000 out of 5000) = 2 rounds
The optimization loop samples 5000 DAGs and selects top 1000 based on AGILE predictions for fine-tuning over two rounds. The precise iterations and selection count are design choices that affect the optimization curve but are not justified by theory.
Lipid tail length constraint for optimization = 10 carbons or longer
Applied in the optimization experiment to improve lipid stability. This is a domain heuristic, not a derived constraint.

assumptions (4)

domain assumption The reaction predictor (Chemformer) accurately predicts the products of reactions between lipid heads and tails.
The whole dataset construction and generation pipeline depends on Chemformer's predictions being correct. The paper shows two qualitative examples but does not validate on a held-out set of lipid reactions. The conclusion explicitly states that the validity of the proposed synthesis pathways has not been evaluated.
domain assumption The lipid classifier and ionizable lipid classifier reliably distinguish lipids and ionizable lipids from other molecules.
The classifiers are used to filter generated products and to construct the training dataset. The paper reports high accuracy/ROC-AUC on held-out data, but the classification is based on computational features (e.g., pKa prediction via MolGpka) that may not perfectly reflect actual ionization behavior in LNP formulation.
domain assumption The AGILE model's predicted mRNA transfection efficiency is a reliable proxy for real transfection efficiency in HeLa cells.
The optimization loop uses AGILE scores to guide fine-tuning. The paper does not verify that the predicted increases translate to actual wet-lab transfection improvements. The limitation is acknowledged in the conclusion.
domain assumption The graph-edit distance and Tanimoto similarity searches in ZINC identify synthetically accessible lipid tails.
The assumption that finding similar molecules in ZINC (which contains commercially available compounds) ensures synthetic accessibility is reasonable but not proven. The paper does not assess how similar is 'similar enough' for practical synthesis.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A Deep Generative Model for the Design of Synthesizable Ionizable Lipids." pith.science (2026). https://pith.science/paper/WY6QZUSK

@misc{pith2026241200928,
  author       = {Pith},
  title        = {Pith review of: A Deep Generative Model for the Design of Synthesizable Ionizable Lipids},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WY6QZUSK}},
  note         = {Machine review of arXiv:2412.00928}
}

read the original abstract

Lipid nanoparticles (LNPs) are vital in modern biomedicine, enabling the effective delivery of mRNA for vaccines and therapies by protecting it from rapid degradation. Among the components of LNPs, ionizable lipids play a key role in RNA protection and facilitate its delivery into the cytoplasm. However, designing ionizable lipids is complex. Deep generative models can accelerate this process and explore a larger candidate space compared to traditional methods. Due to the structural differences between lipids and small molecules, existing generative models used for small molecule generation are unsuitable for lipid generation. To address this, we developed a deep generative model specifically tailored for the discovery of ionizable lipids. Our model generates novel ionizable lipid structures and provides synthesis paths using synthetically accessible building blocks, addressing synthesizability. This advancement holds promise for streamlining the development of lipid-based delivery systems, potentially accelerating the deployment of new therapeutic agents, including mRNA vaccines and gene therapies.

Figures

Figures reproduced from arXiv: 2412.00928 by the authors.

**Figure 1.** An Example of Ionizable Lipid Structure. the application of interest such as activation of stimulator of interferon genes (STING), which is useful for delivering mRNA vaccines. However, despite these advances, the method has limitations. The diversity of ionizable lipids generated remains constrained due to the restricted range of lipid heads and tails available. Therefore, designing and testing a broader and more d… view at source ↗

**Figure 2.** Examples of Predictions from the Original and Improved Reaction Predictors. The original reaction predictor, Molecular Transformer, fails to accurately predict reactions between lipid heads and tails, with errors indicated by blue boxes. In contrast, Chemformer successfully predicts the correct reactions. 5 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Examples of Generated Ionizable Lipids and Their Synthesis Paths. The synthesis paths show the building blocks and intermediate products. Our model can generate ionizable lipids with one to three tails. 7 [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Ionizable Lipids mRNA Transfection Efficiency in HeLa Cells. 6.4 Optimization Towards Ionizable Lipids With High Transfection Efficiency We have developed an ionizable lipid version of Synthesis-DAGs. Our next goal is to optimize these lipids to enhance mRNA transfecti…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 13 canonical work pages

[1]

Guidelines: • The answer NA means that the abstract and introduction do not include the claims made in the paper

Claims Question: Do the main claims made in the abstract and introduction accurately reflect the paper’s contributions and scope? Answer: [Yes] Justification: The paper’s contribution and scope are clearly addressed in the main claim made in the abstract and introduction. Guidelines: • The answer NA means that the abstract and introduction do not include ...

work page
[2]

Limitations

Limitations Question: Does the paper discuss the limitations of the work performed by the authors? Answer: [Yes] Justification: The limitations are addressed in conclusion section. Guidelines: • The answer NA means that the paper has no limitation while the answer No means that the paper has limitations, but those are not discussed in the paper. • The aut...

work page
[3]

Guidelines: • The answer NA means that the paper does not include theoretical results

Theory Assumptions and Proofs Question: For each theoretical result, does the paper provide the full set of assumptions and a complete (and correct) proof? Answer: [NA] 14 Justification: The paper does not include theoretical results. Guidelines: • The answer NA means that the paper does not include theoretical results. • All the theorems, formulas, and p...

work page
[4]

Guidelines: • The answer NA means that the paper does not include experiments

Experimental Result Reproducibility Question: Does the paper fully disclose all the information needed to reproduce the main ex- perimental results of the paper to the extent that it affects the main claims and/or conclusions of the paper (regardless of whether the code and data are provided or not)? Answer: [Yes] Justification: All the details of dataset...

work page
[5]

Guidelines: • The answer NA means that paper does not include experiments requiring code

Open access to data and code Question: Does the paper provide open access to the data and code, with sufficient instruc- tions to faithfully reproduce the main experimental results, as described in supplemental material? 15 Answer: [Yes] Justification: All the source code are publicly available. Guidelines: • The answer NA means that paper does not includ...

work page
[6]

Guidelines: • The answer NA means that the paper does not include experiments

Experimental Setting/Details Question: Does the paper specify all the training and test details (e.g., data splits, hyper- parameters, how they were chosen, type of optimizer, etc.) necessary to understand the results? Answer: [Yes] Justification: All details are included. Guidelines: • The answer NA means that the paper does not include experiments. • Th...

work page
[7]

Guidelines: • The answer NA means that the paper does not include experiments

Experiment Statistical Significance Question: Does the paper report error bars suitably and correctly defined or other appropriate information about the statistical significance of the experiments? Answer: [No] Justification: There is no error bars due to computational cost. Guidelines: • The answer NA means that the paper does not include experiments. • ...

work page
[8]

Guidelines: • The answer NA means that the paper does not include experiments

Experiments Compute Resources Question: For each experiment, does the paper provide sufficient information on the com- puter resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [Yes] Justification: The compute resource is address in experiment implementations section. Guidelines: • The answer NA mean...

work page

Show all 15 references

[9]

Guidelines: • The answer NA means that the authors have not reviewed the NeurIPS Code of Ethics

Code Of Ethics Question: Does the research conducted in the paper conform, in every respect, with the NeurIPS Code of Ethics https://neurips.cc/public/EthicsGuidelines? Answer: [Yes] Justification: We confirm that we followed, in every respect, with the NeurIPS Code of Ethics....
[10]

Guidelines: • The answer NA means that there is no societal impact of the work performed

Broader Impacts Question: Does the paper discuss both potential positive societal impacts and negative societal impacts of the work performed? Answer: [Yes] Justification: Addressed in abstract, introduction and conclusion. Guidelines: • The answer NA means that there is no so...
[11]

Guidelines: • The answer NA means that the paper poses no such risks

Safeguards Question: Does the paper describe safeguards that have been put in place for responsible release of data or models that have a high risk for misuse (e.g., pretrained language models, image generators, or scraped datasets)? Answer: [NA] Justification: The paper poses...
[12]

Guidelines: • The answer NA means that the paper does not use existing assets

Licenses for existing assets Question: Are the creators or original owners of assets (e.g., code, data, models), used in the paper, properly credited and are the license and terms of use explicitly mentioned and properly respected? Answer: [Yes] Justification: All source codes...
[13]

Guidelines: • The answer NA means that the paper does not release new assets

New Assets Question: Are new assets introduced in the paper well documented and is the documentation provided alongside the assets? Answer: [Yes] Justification: All source codes of the paper are publicly available and well documented. Guidelines: • The answer NA means that the...
[14]

Guidelines: • The answer NA means that the paper does not involve crowdsourcing nor research with human subjects

Crowdsourcing and Research with Human Subjects Question: For crowdsourcing experiments and research with human subjects, does the paper include the full text of instructions given to participants and screenshots, if applicable, as well as details about compensation (if any)? A...
[15]

Guidelines: • The answer NA means that the paper does not involve crowdsourcing nor research with human subjects

Institutional Review Board (IRB) Approvals or Equivalent for Research with Human Subjects Question: Does the paper describe potential risks incurred by study participants, whether such risks were disclosed to the subjects, and whether Institutional Review Board (IRB) approvals...

Pith tools

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

[1] [1]

Guidelines: • The answer NA means that the abstract and introduction do not include the claims made in the paper

Claims Question: Do the main claims made in the abstract and introduction accurately reflect the paper’s contributions and scope? Answer: [Yes] Justification: The paper’s contribution and scope are clearly addressed in the main claim made in the abstract and introduction. Guidelines: • The answer NA means that the abstract and introduction do not include ...

work page

[2] [2]

Limitations

Limitations Question: Does the paper discuss the limitations of the work performed by the authors? Answer: [Yes] Justification: The limitations are addressed in conclusion section. Guidelines: • The answer NA means that the paper has no limitation while the answer No means that the paper has limitations, but those are not discussed in the paper. • The aut...

work page

[3] [3]

Guidelines: • The answer NA means that the paper does not include theoretical results

Theory Assumptions and Proofs Question: For each theoretical result, does the paper provide the full set of assumptions and a complete (and correct) proof? Answer: [NA] 14 Justification: The paper does not include theoretical results. Guidelines: • The answer NA means that the paper does not include theoretical results. • All the theorems, formulas, and p...

work page

[4] [4]

Guidelines: • The answer NA means that the paper does not include experiments

Experimental Result Reproducibility Question: Does the paper fully disclose all the information needed to reproduce the main ex- perimental results of the paper to the extent that it affects the main claims and/or conclusions of the paper (regardless of whether the code and data are provided or not)? Answer: [Yes] Justification: All the details of dataset...

work page

[5] [5]

Guidelines: • The answer NA means that paper does not include experiments requiring code

Open access to data and code Question: Does the paper provide open access to the data and code, with sufficient instruc- tions to faithfully reproduce the main experimental results, as described in supplemental material? 15 Answer: [Yes] Justification: All the source code are publicly available. Guidelines: • The answer NA means that paper does not includ...

work page

[6] [6]

Guidelines: • The answer NA means that the paper does not include experiments

Experimental Setting/Details Question: Does the paper specify all the training and test details (e.g., data splits, hyper- parameters, how they were chosen, type of optimizer, etc.) necessary to understand the results? Answer: [Yes] Justification: All details are included. Guidelines: • The answer NA means that the paper does not include experiments. • Th...

work page

[7] [7]

Guidelines: • The answer NA means that the paper does not include experiments

Experiment Statistical Significance Question: Does the paper report error bars suitably and correctly defined or other appropriate information about the statistical significance of the experiments? Answer: [No] Justification: There is no error bars due to computational cost. Guidelines: • The answer NA means that the paper does not include experiments. • ...

work page

[8] [8]

Guidelines: • The answer NA means that the paper does not include experiments

Experiments Compute Resources Question: For each experiment, does the paper provide sufficient information on the com- puter resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [Yes] Justification: The compute resource is address in experiment implementations section. Guidelines: • The answer NA mean...

work page