Evaluating Molecule Synthesizability via Retrosynthetic Planning and Reaction Prediction

Dandan Zhang; Hanjun Dai; Peng Liu; Songtao Liu; Zhengkai Tu

REVIEW 3 major objections 5 minor 12 references

Evaluating Molecule Synthesizability via Retrosynthetic Planning and Reaction Prediction

T0 review · 3 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 The paper proposes the round-trip score—plan a route backward, simulate it forward, and measure how well the target comes back—and shows it outperforms the search success rate for judging whether generated molecules are synthesizable.

desk verdict A sensible round-trip synthesizability metric, but the headline recall number is inflated by a training-set leak that the authors need to fix. read the letter →

arxiv 2411.08306 v2 pith:YXV7BWD7 submitted 2024-11-13 cs.LG q-bio.QM

Songtao Liu , Dandan Zhang , Zhengkai Tu , Hanjun Dai , Peng Liu This is my paper · ORCID

classification cs.LGq-bio.QM

keywords synthesizabilityevaluationround-tripscoreretrosyntheticplanningforwardreactionpredictionstructure-baseddrugdesigngenerativemodelssyntheticaccessibilityTanimotosimilarity

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 proposes a new metric, the round-trip score, for deciding whether a molecule generated by a drug-design model can actually be synthesized. The score is computed in three stages: a retrosynthetic planner finds a route backward from the molecule to purchasable starting materials, a forward reaction model tries to simulate that route forward from those materials, and the Tanimoto similarity between the original molecule and the simulated product becomes the score. The paper argues this score distinguishes feasible from infeasible routes far better than the search success rate, reporting 76.4% precision versus 62.1% on a 1,068-molecule test set, and shows the SA score cannot separate the two groups. It then uses the metric to benchmark seven structure-based drug-design generative models, finding large differences in the share of molecules with at least one reproducible route.

What carries the argument

The load-bearing object is the round-trip score $S(m) = \mathrm{Sim}(m, m')$, where $m'$ is the product obtained by running a retrosynthetic planner backward from $m$ to starting materials and then a forward reaction model forward from those starting materials back to a molecule. Tanimoto similarity here is the Jaccard similarity between molecular fingerprint bit vectors, so a score of 1 means the forward simulation reproduced the target exactly; lower scores mean the simulated route leads somewhere else. The forward model is a Transformer decoder trained on about 916k cleaned USPTO reactions, and the retrosynthetic planner is the template-based Neuralsym with beam search. The argument works because the forward model acts as a wet-lab proxy: if the planner proposes a reaction that is not chemically faithful, the forward simulation will not return the target, and the route is penalized. Reference routes from the same reaction network provide the feasible/infeasible labels used to measure the metric's discrimination.

What would settle it

Retrain the forward model on a cleaned USPTO split that excludes the reactions used to build the test-set reference routes, recompute round-trip scores on the same 1,068 targets, and check whether the 93.9% recall and the ranking of the seven generative models persist; if recall collapses or rankings change, the reported discrimination depends on memorization rather than generalizable forward prediction. A complementary wet-lab check would be to take a sample of generated molecules with round-trip score 1, follow the predicted routes in a real laboratory, and measure how many actually produce the target.

Watch

Extended reading notes

Core claim

The central claim is that synthesizability can be quantified by round-tripping: if the starting materials of a predicted synthetic route can be fed through a learned forward reaction model to reproduce the target molecule, the route is feasible; if not, it is not. Concretely, the paper defines $S(m) = \mathrm{Sim}(m, f_\Phi(g_\Theta(m)))$, where $g$ is a retrosynthetic planner, $f$ is a forward reaction predictor, and $\mathrm{Sim}$ is Tanimoto similarity. On 1,027 molecules for which the planner produced routes, the round-trip score identifies feasible routes with 76.4% precision and 93.9% recall (F1 84.2%), while the search success rate treats all 1,027 as found and has 62.1% precision. The paper also reports that the SA score gives nearly identical averages for feasible and infeasible routes (2.68 vs 2.73), so it cannot discriminate. Applied to 10,000 molecules from seven generative models, the round-trip score ranks Pocket2Mol highest, with 22.05% of its molecules having at least one route among the top five that the forward model can reproduce, versus 2.87% for LiGAN.

Load-bearing premise

The metric is only as reliable as the forward reaction model, which is assumed to predict wet-lab outcomes on newly designed molecules even though it was trained on about 916k literature reactions; if that model memorizes training reactions or fails on out-of-distribution molecules, a high round-trip score will not guarantee a real synthesis.

Editorial extensions

If this is right

The round-trip score can replace SA score and search success rate as the primary synthesizability check for generative drug-design models, since it directly verifies route reproducibility instead of relying on structural heuristics or mere route existence.
A model's synthesizability is measured by the fraction of its generated molecules for which at least one of the top five predicted routes reproduces the molecule, so multi-route evaluation gives a fuller picture than single-route success.
Because the score is computed entirely from reaction data, expanding reaction databases should make the metric stricter and more reliable without changing the evaluation protocol.
Hallucinated retrosynthetic reactions are penalized automatically: a route counts only if the forward model can walk it back to the target, fixing a known weakness of the search success rate.

Reading between the lines

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

The reliability ceiling of the round-trip score is set by forward-model accuracy, so an explicit out-of-distribution benchmark—evaluating the forward model on reactions with unseen functional groups—would quantify how much of the reported recall transfers to truly novel molecules.
The same round-trip procedure could be turned into a training reward for generative models, so molecules are optimized not just for docking scores but for surviving forward simulation; the paper stops at evaluation.
The score could also be used to compare route quality continuously, rather than thresholding at 1, by retaining the Tanimoto value as a graded measure of how close the simulated route gets to the target.
The observed near-zero round-trip rates for most models (top-5 under 22%) suggest that current SBDD training objectives ignore synthesizability; a round-trip filter during sampling could be a cheap way to boost the fraction of synthesizable hits.

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

Two things to know. The idea is a natural extension of Schwaller's round-trip accuracy to multi-step routes, and the benchmark across seven SBDD models is a useful reference point. But the headline validation result in Table 1 is not trustworthy as reported, because the forward model was trained on the same 916k USPTO reactions used to build the test routes. That makes the 93.9% recall largely a memorization score, not a generalization result.

What is actually new: the multi-step round-trip score S(m)=Sim(m, f(g(m))) composed from a retrosynthetic planner and a forward predictor, and the evaluation of seven SBDD generative models on it. The paper is clearly written, the arithmetic is internally consistent, and the comparison with search success rate is a legitimate improvement in principle. The motivation is sound: search success rate is too lenient, and SA score is not route-aware.

The soft spot is the validation. The paper trains the forward Transformer on 'about 916k reactions' — the same cleaned USPTO set used to construct the reaction network and extract reference routes for the test molecules. So a large fraction of the 'feasible' test routes are composed of reactions the forward model has memorized. The manual subset is supposed to break the circularity, but it is small (112 feasible routes) and was labeled by the authors with CAS and expertise, without a blinding protocol. The paper does not report the recall on that manual subset; if you aggregate the given numbers, it is about 79.5%, well below the 97% on reference-matched routes. The stress-test note gets this right. This means the absolute top-k percentages in Table 2 lack a validated calibration, though the ranking of models is probably informative.

Who should read this: anyone working on synthesizability scoring or generative drug design. It is a fair contribution to the subfield, not a breakthrough. As a referee, I'd send it out, but I would insist the authors retrain or reselect the forward model so that test routes are strictly out-of-distribution, and report recall separately for matched and manually labeled routes. If they fix the leak, the central claim becomes credible. Right now, the concept is fine and the execution is not.

Referee Report

3 major / 5 minor

Summary. The paper proposes a round-trip score S(m) = Sim(m, f(g(m))) for evaluating the synthesizability of molecules generated by drug-design models. A retrosynthetic planner g first proposes a synthetic route from purchasable starting materials; a forward reaction prediction model f then simulates the route in the forward direction; the Tanimoto similarity between the original molecule and the reconstructed molecule is the score. The authors validate the score on a test set of 1,068 molecules drawn from a reaction network built from about 916k cleaned USPTO reactions, labeling predicted routes as feasible either by matching starting materials to reference routes or by manual CAS/expert evaluation. They report that the round-trip score has precision 76.4%, recall 93.9%, and F1 84.2%, compared with search success rate precision 62.1% (Table 1). They then benchmark seven structure-based drug design generative models and report top-k round-trip success percentages (Table 2), concluding that the metric is more reliable than the search success rate and that existing models still generate mostly molecules without feasible synthetic routes.

Significance. If the validation were sound, the round-trip score would be a valuable, parameter-free evaluation metric that combines retrosynthetic planning and forward reaction prediction, directly addressing a known weakness of SA-score-based and search-success-based synthesizability measures. The paper is also useful in assembling a large-scale benchmark of 1,068 test molecules and applying the metric to seven SBDD models, producing a concrete ranking. The score itself has no fitted parameters beyond the trained retrosynthesis and forward models, which is a genuine strength. However, the empirical validation is the load-bearing part of the paper, and that validation has a training/test overlap problem that substantially weakens the central claim. The metric idea is defensible and the comparison with search success rate is meaningful, but the current evidence does not yet establish generalization to out-of-distribution generated molecules.

major comments (3)

[Section 4.1, Settings and Evaluation Protocol] The forward reaction model is trained on 'about 916k reactions' from the cleaned USPTO set, and the same set is used to build the reaction network from which the reference routes and the feasibility labels are derived. A predicted route is labeled feasible when its starting materials match a reference route in the test set, so every reaction in that reference route is present in the forward model's training data. The 93.9% recall (510/526) therefore reflects, at least in part, the forward model's ability to recall reactions it has already seen, not its ability to simulate novel chemistry. This is load-bearing for the claim that the round-trip score 'surpasses the search success rate' as a general synthesizability metric. Please retrain the forward model on reactions excluding all reactions that appear in the test reference routes (or otherwise report recall on test routes whose reactions were not in the training set), and report the corresponding Table 1 statistics.
[Section 4.1, Evaluation Protocol] The manually labeled feasible routes provide a partially independent check, but the paper does not report their recall separately. Aggregating the stated numbers gives (36 + 53)/(44 + 68) = 89/112 ≈ 79.5% for the manual subset, which is materially below the 97.0% recall on reference-matched routes. This gap is consistent with the memorization concern raised above and should be reported and discussed. In addition, the manual labels were assigned by the authors using CAS and their own domain expertise without a blinded protocol or inter-annotator agreement, so the label noise is uncontrolled. At minimum, report recall on the manual subset separately and specify the criteria used to decide that a partial CAS match 'is also accurate'.
[Section 4.2, Table 2] The absolute top-k percentages in Table 2 are presented as measures of synthesizability for generated molecules, but their calibration depends entirely on the Table 1 validation, which is weakened by the training/test overlap described above. For out-of-distribution generated molecules there are no reference routes to match, so the forward model is being applied in a regime the current experiment does not validate. At most the ranking across models may be supported, and even that assumes error rates are similar across the seven generative models. Please add a calibrated evaluation on a held-out set of novel reactions or generated molecules, and report uncertainty estimates for the Table 2 percentages.

minor comments (5)

[Section 4.1, Dataset Construction] The term 'data points' is used for molecules; please clarify that the 105,218/1,068/1,068 split is over target molecules, not reactions, and state explicitly how reactions from the test molecules are handled when building the forward model's training set.
[Section 4.1, Evaluation Protocol] The claim that the scale is 'highly convincing' because 1,068 exceeds 189 by a factor of five is not by itself an argument for reliability; please supplement with confidence intervals or a statistical comparison of the precision/recall estimates.
[Figure 3] The notation m_tar&, m_I1&, and m_I2& is used in the figure but not defined in the text; please define the reconstructed intermediates and clarify how they are matched to the original route.
[Section 4.2, Settings] The statement that the authors 'verify the validity and plausibility of these molecules' is not backed by a description of the filters used; please specify the validity and plausibility criteria.
[Section 4.1, Results] The conclusion that the SA score 'lacks the ability to differentiate' because the average SA scores are 2.68 and 2.73 would be stronger with a reported standard deviation or significance test, given that the two groups are large.

Circularity Check

1 steps flagged · score 5.0 of 10

Validation of the round-trip score is partly circular: the forward model is trained on the same USPTO reactions used to label routes feasible, so the 93.9% recall measures memorization as much as generalization; the independent manual subset shows lower recall.

fitted input called prediction [Section 4.1, Settings / Evaluation Protocol / Table 1]
"We use the Transformer (Vaswani et al., 2017) Decoder as our forward reaction prediction model, training it on about 916k reactions. ... If the starting materials of the predicted route match the starting materials of any reference route, the route is deemed feasible. ... 510 of these molecules have round-trip scores of 1."

The forward model is trained on the same ~916k cleaned USPTO reactions used to build the reaction network from which the test reference routes are extracted. A test route is labeled feasible exactly when its starting materials match one of those reference routes, so every reaction in a reference-matched feasible route is already present in the forward model's training data. The round-trip score of 1 on such routes can be achieved by recalling seen reactions rather than by generalizing to new chemistry. The reported recall of 93.9% (510/526) is therefore partly a training-set fit and does not by itself validate the metric on out-of-distribution generated molecules.

full rationale

The round-trip score itself, S(m)=Sim(m,f(g(m))), contains no fitted parameters and is not self-definitional: it is a cycle-consistency quantity that can in principle be evaluated on any molecule, and the Section 4.2 benchmark on SBDD-generated molecules is out-of-distribution for the trained models, so the rankings there are not forced by the metric's definition. The circularity lies in the validation protocol of Section 4.1. Feasible ground truth is defined by matching reference routes extracted from the same cleaned USPTO reaction set on which the forward Transformer is trained, so reference-matched feasible routes are training-set reactions; the 93.9% recall on those routes partly reflects memorization. The paper itself notes that 'the reference routes in the test set do not cover all possible feasible routes' and supplements with CAS/manual evaluation, which provides a genuinely external check; however, the manual labels are assigned by the authors without a blinded protocol, and their aggregated recall (89/112 ≈ 79.5%) is materially below the reference-matched recall. Because the central claim 'the round-trip score surpasses the search success rate' rests mainly on the reference-matched numbers, the empirical support is partially circular. No load-bearing self-citation chain or uniqueness import was found: FusionRetro is cited for the matching-based evaluation methodology, but the round-trip metric and the generated-molecule benchmark have independent content. Overall, a moderate partial circularity score is appropriate.

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

The paper adds no invented physical entities. Its central claim rests on two fitted neural models (retrosynthesis and forward reaction prediction), on feasibility labels derived from the same USPTO reaction network, and on the assumption that exact in-silico reconstruction approximates wet-lab synthesizability. The only hand-set knobs are beam size, route depth, and the score=1 threshold.

free parameters (3)

beam size for retrosynthetic search = 5
Chosen due to computational constraints (Section 4.2); affects all top-k benchmark numbers.
maximum route depth = 15
Set to the highest depth found in the route dataset (Section 4.1); changes which molecules get a route and therefore the score distribution.
round-trip success threshold = 1.0
The paper counts only a Tanimoto similarity of exactly 1 as a successful reproduction, collapsing a continuous score to a binary label.

assumptions (4)

domain assumption Retrosynthetic planner and forward reaction predictor trained on USPTO are valid proxies for wet-lab chemistry.
Sections 3.1 and 4.1 use these models in place of experiments; the central reliability claim depends on model quality.
domain assumption Route feasibility can be labeled by matching starting materials to reference USPTO routes or by authors' manual CAS review.
Section 4.1 defines ground truth this way; not all feasible routes are in the database and manual labels are subjective.
domain assumption A molecule is synthesizable if a predicted route's starting materials can undergo the recorded sequence of reactions to reproduce it exactly.
Equation (1) defines the score; does not account for byproducts, yields, or experimental conditions.
domain assumption ZINC purchasable compounds are the starting-material space.
Footnote 1 in Section 1 defines starting materials via ZINC; this choice bounds all routes.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Evaluating Molecule Synthesizability via Retrosynthetic Planning and Reaction Prediction." pith.science (2026). https://pith.science/paper/YXV7BWD7

@misc{pith2026241108306,
  author       = {Pith},
  title        = {Pith review of: Evaluating Molecule Synthesizability via Retrosynthetic Planning and Reaction Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YXV7BWD7}},
  note         = {Machine review of arXiv:2411.08306}
}

read the original abstract

A significant challenge in wet lab experiments with current drug design generative models is the trade-off between pharmacological properties and synthesizability. Molecules predicted to have highly desirable properties are often difficult to synthesize, while those that are easily synthesizable tend to exhibit less favorable properties. As a result, evaluating the synthesizability of molecules in general drug design scenarios remains a significant challenge in the field of drug discovery. The commonly used synthetic accessibility (SA) score aims to evaluate the ease of synthesizing generated molecules, but it falls short of guaranteeing that synthetic routes can actually be found. Inspired by recent advances in top-down synthetic route generation and forward reaction prediction, we propose a new, data-driven metric to evaluate molecule synthesizability. This novel metric leverages the synergistic duality between retrosynthetic planners and reaction predictors, both of which are trained on extensive reaction datasets. To demonstrate the efficacy of our metric, we conduct a comprehensive evaluation of round-trip scores across a range of representative molecule generative models.

Figures

Figures reproduced from arXiv: 2411.08306 by the authors.

**Figure 1.** For a given molecule, multiple synthetic routes can be identified within the reaction database, illustrating the diverse routes available for its synthesis. 2.5. Retrosynthetic Planning Retrosynthetic planning aims to predict synthetic routes for target molecules. This process works backward from the desired target, predicting potential precursor molecules that could be transformed into the target through chemical r… view at source ↗

**Figure 2.** Comparison of evaluation metrics for retrosynthetic planning. The search success rate deems both routes successful, while the matching-based metric correctly identifies the top route as incorrect and the bottom route as correct, demonstrating its superior reliability. prediction within the highlighted circle is incorrect, while the bottom part illustrates a correctly predicted synthetic route. Despite this, both rou… view at source ↗

**Figure 3.** Illustration of the round-trip score calculation process. It consists of three stages: Retrosynthetic Planning, Forward Reproduction, and Similarity Computation. molecule m′ . Finally, we compute the Tanimoto similarity between m and m′ , which we define as the round-trip score [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: The feasibility of this reaction is confirmed by the reactions documented in the CAS database. routes. Please note that the number of molecules evaluated far exceeds the number of molecules used in the benchmarks for assessing retrosynthetic planning search algorith…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 10 canonical work pages

[3]

K., Torren-Peraire, P., Genheden, S., Verho- even, J., Preuss, M., and Tetko, I

Hassen, A. K., Torren-Peraire, P., Genheden, S., Verho- even, J., Preuss, M., and Tetko, I. Mind the retrosyn- thesis gap: Bridging the divide between single-step and multi-step retrosynthesis prediction. arXiv preprint arXiv:2212.11809,

arXiv
[9]

Re-evaluating chemical synthesis planning algorithms

Tripp, A., Maziarz, K., Lewis, S., Liu, G., and Segler, M. Re-evaluating chemical synthesis planning algorithms. In NeurIPS 2022 AI for Science: Progress and Promises ,

work page 2022
[11]

Leveraging Reaction-aware Substructures for Retrosynthesis Analysis

Zhao, M., Fang, L., Tan, L., Lou, J.-G., and Lepage, Y . Leveraging reaction-aware substructures for ret- rosynthesis and reaction prediction. arXiv preprint arXiv:2204.05919,

work page Pith review arXiv
[12]

The hyper-parameters for the reaction model. Max Length Embedding Size Decoder Layers Attention Heads FFN Hidden Dropout 402 64 6 8 2048 0.1 Epochs Batch Size Warmup LR Factor Scheduling 2000 128 16000 20 lr = lr factor × min(1.0, 0.1 numstep warmup ) max(0.1 num step,warmup) Table 3 reports the hyper-parameter setting of our reaction model. For Neuralsym...

work page 2000
[2012]

Retrosynthetic Planning with Experience-Guided Monte Carlo Tree Search

Hong, S., Zhuo, H. H., Jin, K., and Zhou, Z. Retrosynthetic planning with experience-guided monte carlo tree search. arXiv preprint arXiv:2112.06028,

work page Pith review arXiv
[2018]

Seo, S., Kim, M., Shen, T., Ester, M., Park, J., Ahn, S., and Kim, W. Y . Generative flows on synthetic pathway for drug design. arXiv preprint arXiv:2410.04542,

work page Pith review arXiv
[2019]

Koziarski, M., Rekesh, A., Shevchuk, D., van der Sloot, A., Gai´nski, P., Bengio, Y ., Liu, C.-H., Tyers, M., and Batey, R. A. Rgfn: Synthesizable molecular generation using gflownets. arXiv preprint arXiv:2406.08506,

arXiv
[2020]

Directly Optimizing for Synthesizability in Generative Molecular Design using Retrosynthesis Models

Guan, J., Qian, W. W., Peng, X., Su, Y ., Peng, J., and Ma, J. 3d equivariant diffusion for target-aware molecule gener- ation and affinity prediction. In International Conference on Learning Representations, 2023a. Guan, J., Zhou, X., Yang, Y ., Bao, Y ., Peng, J., Ma, J., Liu, Q., Wang, L., and Gu, Q. DecompDiff: Diffusion models with decomposed priors ...

work page Pith review arXiv

Show all 12 references

[2021]

A., Bekas, C., and Lee, A

Schwaller, P., Laino, T., Gaudin, T., Bolgar, P., Hunter, C. A., Bekas, C., and Lee, A. A. Molecular transformer: a model for uncertainty-calibrated chemical reaction prediction. ACS Central Science, 2019a. Schwaller, P., Petraglia, R., Nair, V ., and Laino, T. Eval- uation me...

2019
[2022]

Syn- flownet: Towards molecule design with guaranteed syn- thesis pathways

Cretu, M., Harris, C., Roy, J., Bengio, E., and Li`o, P. Syn- flownet: Towards molecule design with guaranteed syn- thesis pathways. In ICLR 2024 Workshop on Generative and Experimental Perspectives for Biomolecular Design,

2024
[2023]

Retro-bleu: Quantifying chemical plausibility of retrosynthesis routes through reaction template sequence analysis

Li, J., Fang, L., and Lou, J.-G. Retro-bleu: Quantifying chemical plausibility of retrosynthesis routes through reaction template sequence analysis. arXiv preprint arXiv:2311.06304,

arXiv
[2024]

Evo- lutionary retrosynthetic route planning

Zhang, Y ., Hao, H., He, X., Gao, S., and Zhou, A. Evo- lutionary retrosynthetic route planning. arXiv preprint arXiv:2310.05186,

arXiv

Pith tools

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

[1] [3]

K., Torren-Peraire, P., Genheden, S., Verho- even, J., Preuss, M., and Tetko, I

Hassen, A. K., Torren-Peraire, P., Genheden, S., Verho- even, J., Preuss, M., and Tetko, I. Mind the retrosyn- thesis gap: Bridging the divide between single-step and multi-step retrosynthesis prediction. arXiv preprint arXiv:2212.11809,

arXiv

[2] [9]

Re-evaluating chemical synthesis planning algorithms

Tripp, A., Maziarz, K., Lewis, S., Liu, G., and Segler, M. Re-evaluating chemical synthesis planning algorithms. In NeurIPS 2022 AI for Science: Progress and Promises ,

work page 2022

[3] [11]

Leveraging Reaction-aware Substructures for Retrosynthesis Analysis

Zhao, M., Fang, L., Tan, L., Lou, J.-G., and Lepage, Y . Leveraging reaction-aware substructures for ret- rosynthesis and reaction prediction. arXiv preprint arXiv:2204.05919,

work page Pith review arXiv

[4] [12]

The hyper-parameters for the reaction model. Max Length Embedding Size Decoder Layers Attention Heads FFN Hidden Dropout 402 64 6 8 2048 0.1 Epochs Batch Size Warmup LR Factor Scheduling 2000 128 16000 20 lr = lr factor × min(1.0, 0.1 numstep warmup ) max(0.1 num step,warmup) Table 3 reports the hyper-parameter setting of our reaction model. For Neuralsym...

work page 2000

[5] [2012]

Retrosynthetic Planning with Experience-Guided Monte Carlo Tree Search

Hong, S., Zhuo, H. H., Jin, K., and Zhou, Z. Retrosynthetic planning with experience-guided monte carlo tree search. arXiv preprint arXiv:2112.06028,

work page Pith review arXiv

[6] [2018]

Seo, S., Kim, M., Shen, T., Ester, M., Park, J., Ahn, S., and Kim, W. Y . Generative flows on synthetic pathway for drug design. arXiv preprint arXiv:2410.04542,

work page Pith review arXiv

[7] [2019]

Koziarski, M., Rekesh, A., Shevchuk, D., van der Sloot, A., Gai´nski, P., Bengio, Y ., Liu, C.-H., Tyers, M., and Batey, R. A. Rgfn: Synthesizable molecular generation using gflownets. arXiv preprint arXiv:2406.08506,

arXiv

[8] [2020]

Directly Optimizing for Synthesizability in Generative Molecular Design using Retrosynthesis Models

Guan, J., Qian, W. W., Peng, X., Su, Y ., Peng, J., and Ma, J. 3d equivariant diffusion for target-aware molecule gener- ation and affinity prediction. In International Conference on Learning Representations, 2023a. Guan, J., Zhou, X., Yang, Y ., Bao, Y ., Peng, J., Ma, J., Liu, Q., Wang, L., and Gu, Q. DecompDiff: Diffusion models with decomposed priors ...

work page Pith review arXiv