Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agents

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read LLM agents with iterative critique and a materials knowledge graph generate hypotheses scoring 80% closeness and 89% quality, aligned with human expert ratings.

desk verdict MATDESIGN is a genuinely useful new benchmark, but the headline results rest on an unverified no-leakage premise and an LLM judge that the paper's own appendix undermines. read the letter →

arxiv 2501.13299 v2 pith:3VVKP7TX submitted 2025-01-23 cs.CL

Shrinidhi Kumbhar , Venkatesh Mishra , Kevin Coutinho , Divij Handa , Ashif Iquebal , Chitta Baral This is my paper · ORCID
classification cs.CL
keywords materialsdiscoveryhypothesisgenerationlargelanguagemodelsmulti-agentsystemsknowledgegraphbenchmarkdatasetevaluationmetricsgoal-drivendesign
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that large language models, organized as a multi-agent system with iterative critique and grounded knowledge, can generate genuinely novel and viable materials hypotheses from a natural-language goal and a list of constraints. To test this, the authors built MATDESIGN, a benchmark of 50 real-world design problems taken from January 2024 journal papers, paired with the materials and methods that actually solved them. Their ACCELMAT framework generates 20 hypotheses per problem, has three LLMs critique them, consolidates the feedback, and refines the hypotheses for up to five rounds; in the best configuration, the generator also receives context from the MatKG materials knowledge graph. On their proposed evaluation, which scores closeness to ground truth and six quality criteria with an LLM evaluator, this best configuration reaches 80% Closeness and 89% Quality, and the authors report that human materials-science experts' ratings 'paralleled' the automated scores. If this is right, an LLM-based pipeline could propose candidate hypotheses and prioritize them for experimental validation, reducing the literature-review and ideation burden in materials discovery.

What carries the argument

The load-bearing machinery is the ACCELMAT agent architecture: a Hypotheses Generation Agent (GPT-4o) that outputs 20 hypotheses from a goal and constraint list; three Critic Agents (GPT-4o, Claude-3.5-Sonnet, Gemini-1.5-Flash) that judge alignment and return feedback; a Summarizer Agent (GPT-4o) that merges feedback; and an Evaluation Agent (OpenAI-o1-preview) that applies the Closeness and Quality rubrics. In the best configuration, goal keywords are used to query MatKG, the largest publicly available materials science knowledge graph, and the retrieved materials and properties are fed to the generator as context. The iterative loop runs up to five cycles, stopping when all 20 hypotheses are unanimously agreed or when all critics' suggestions converge. The dataset MATDESIGN, curated with materials experts from 50 papers published in January 2024, supplies the goals, constraints, and ground-truth materials/methods.

What would settle it

Ask each LLM used in the study—without giving it the goal-solving context—to name the material and method that achieved each MATDESIGN goal, then check whether the models recite the exact ground-truth materials (e.g., 'silyl ester in coaxial electrospun core-shell nanofibers' for the self-healing coating example). If the models name those specific compounds or methods with high accuracy, the no-leakage assumption fails. A second check is to search preprint servers and journal websites to determine whether any of the 50 source papers' full texts were publicly available before the models' knowledge cutoffs.

Watch

Extended reading notes

Core claim

The central claim is that adding structured domain knowledge and iterative multi-LLM feedback to a hypothesis-generation LLM substantially improves both the faithfulness of the generated hypotheses to the intended solution and their overall scientific quality. Using the MATDESIGN benchmark, the paper reports that the full ACCELMAT configuration (critic feedback plus MatKG grounding) achieves the highest performance, with a Closeness score of 80% and a Quality score of 89%, compared with 70% and 79.67% for the no-feedback baseline. The Closeness metric decomposes into concept overlap, property overlap, and keyword matching against ground truth; the Quality metric averages ratings for alignment, plausibility, innovation, testability, feasibility/scalability, and impact potential, all produced by OpenAI-o1-preview as the evaluation agent. The authors further report that four PhD materials science students evaluated 42 suggestions with the same metrics and that 'the human evaluations paralleled the automated results,' which they take as validating the scalable evaluation metric as a substitute for some human assessment. The paper is careful to frame the generated suggestions as a starting point for researchers to refine, not as finished laboratory-ready procedures.

Load-bearing premise

The central premise is that the ground-truth solutions in MATDESIGN, drawn from papers published in January 2024, are entirely absent from the training data of GPT-4o, Claude-3.5-Sonnet, Gemini-1.5-Flash, and OpenAI-o1-preview, so the models cannot be retrieving the answer from memory; if any of these papers appeared as an earlier preprint or online-first within a model's training window, the reported Closeness and Quality scores would overstate genuine hypothesis generation.

Editorial extensions

If this is right

  • The best configuration (knowledge graph + critic feedback) reaches 80% Closeness and 89% Quality, while the no-feedback baseline scores 70% Closeness and 79.67% Quality, a gain the authors attribute to the iterative loop and grounded context.
  • Average critic agreement rises from 11 of 20 hypotheses without feedback to 19 of 20 with feedback and knowledge graph, indicating that the pipeline converges on suggestions that satisfy all three critics.
  • Because the MATDESIGN entries postdate the knowledge cutoffs of all LLMs used, the benchmark is designed to measure whether an LLM can synthesize novel hypotheses rather than recall known solutions.
  • The proposed Closeness and Quality metrics, implemented with OpenAI-o1-preview, approximate the judgment of human materials scientists, suggesting that scalable LLM-based evaluation could be used in place of some human assessment.

Reading between the lines

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

  • The paper's 'no data leakage' premise is never empirically verified: January 2024 journal articles often appear as preprints or online-first months earlier, so some MATDESIGN ground truths may sit inside the models' training data; a direct probe asking each LLM to name the solution for a given goal would settle whether Closeness scores are inflated by memorization.
  • The additional gains from MatKG grounding (Closeness +6.67%, Quality +3.33% over feedback alone) could stem from keyword and entity retrieval rather than deeper scientific reasoning, since the paper does not ablate the knowledge-graph component on its own.
  • Because the evaluation agent is itself an LLM using the same model family as the generator, the claim that its scores parallel human experts rests on a single small human study of 42 suggestions; replicating the human comparison on a larger sample would determine whether the metric can safely substitute for human panels.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper introduces MATDESIGN, a benchmark of 50 materials-science design tasks extracted from January 2024 journal papers, and ACCELMAT, a multi-LLM agent framework that generates hypotheses from a goal and constraints, iteratively refines them with feedback from three critic LLMs and a summarizer, optionally grounds them in the MatKG knowledge graph, and evaluates them with an OpenAI-o1-preview-based rubric covering Closeness to ground truth and six Quality criteria. Three configurations (no feedback, feedback, feedback plus knowledge graph) are compared, with the full configuration reported as best (Closeness 80%, Quality 89%). A human evaluation of 42 suggestions by four materials-science PhD students is reported as paralleling the automated results.

Significance. If the claims hold, the paper provides a useful testbed and framework for LLM-driven materials hypothesis generation: the dataset targets realistic, constraint-rich design goals; the architecture separates generation, critique, summarization, and evaluation; and the proposed rubric is more interpretable than text-similarity scores. The manuscript also includes a BERTScore comparison, which is a healthy sanity check even though it conflicts with the main result. The central findings are, however, conditional on two untested assumptions: that the MATDESIGN ground truths are absent from all employed LLMs' training data, and that the o1-preview-based evaluation is calibrated against human judgment. Because both assumptions are load-bearing for the headline comparisons, the contribution is not yet established at the level claimed.

major comments (4)
  1. [Section 3 and footnote 2] The no-data-leakage assumption is unverified and is contradicted by the model set actually used. Section 3 states that January 2024 publications ensure the ground truth is beyond the training cutoff of 'the LLMs we use', but footnote 2 provides cutoffs only for Llama-3.1-70B (December 2023) and GPT-4o (October 2023). The pipeline in Section 4 also uses Claude-3.5-Sonnet, Gemini-1.5-Flash, and OpenAI-o1-preview, whose public documentation indicates training data extending into 2024. Moreover, January-2024 journal articles may have been available as preprints or online-first months earlier. The paper never runs a memorization probe, such as prompting each model with the goal/constraints alone to see whether it reproduces the exact materials and methods. Because Closeness measures overlap with ground truth, high Closeness scores (80%) could reflect retrieval of memorized content rather than novel hypothesis generation. This assumption directly supports the paper's novelty claim and must be verified or substantially qualified.
  2. [Section 7.1] The human evaluation does not quantitatively validate the automated metric. The paper reports that 42 suggestions were rated by four materials-science PhD students, but gives only a narrative summary ('the human evaluations paralleled the automated results') with no inter-rater agreement statistic (e.g., Cohen's kappa), no per-criterion scores, no effect sizes, and no confidence intervals. With 42 suggestions spread across three configurations and four raters, the narrative conclusion is not a calibration of the o1-preview evaluator. The authors should report rater agreement, per-configuration human scores, and a correlation or agreement measure between human and automated scores (for the same 42 items). Without this, the central claim that the proposed evaluation 'aligns well with expert judgments' (Section 8) is not supported.
  3. [Appendix G] The BERTScore results contradict the paper's main conclusion and are dismissed too quickly. Using MaterialsBERT, Appendix G reports Closeness-like scores of 52.66% (no feedback), 60.59% (feedback), and 50.30% (feedback plus knowledge graph). Thus, the configuration that the paper's own LLM evaluator ranks best is worst under BERTScore. The manuscript says these scores 'do not align' and are hard to interpret, but this does not explain why the knowledge-graph configuration should be lower on a lexical-similarity measure. The discrepancy suggests that the o1-preview evaluator may be systematically favoring outputs that mention knowledge-graph-related terms (e.g., 'graphene', 'zinc', 'epoxy'), rather than outputs genuinely closer to the ground truth. The authors should either provide a detailed analysis of the discrepancy or temper the claim that the full configuration is best; at minimum, both metrics should be reported and discussed in the main text.
  4. [Figure 2 and Section 7] All reported aggregate scores (Closeness 70%, 73.33%, 80%; Quality 79.67%, 85.67%, 89%) are point estimates without error bars, standard deviations, or significance tests. Since each configuration produces 20 hypotheses per goal across 50 goals, per-instance variance is expected to be substantial; the reported 3.33–6.67 percentage-point differences between configurations may not be statistically distinguishable. The authors should report confidence intervals or per-goal score distributions, and ideally a paired significance test, before claiming that feedback and knowledge-graph grounding 'enhance performance' (Section 7).
minor comments (6)
  1. [Abstract / Section 3] The abstract states the dataset is 'from recent journal publications' and the introduction says it 'lies beyond the knowledge cutoff of all LLMs employed'; the latter is only supported for two of the five models used, as noted above, and should be corrected or explicitly scoped.
  2. [Section 7] The Closeness and Quality scores are reported as percentages, but the underlying scales in Tables 4 and 6 are 1–5; the normalization (e.g., (mean score)/5 × 100) should be stated explicitly in Section 5.
  3. [Appendix G] The sentence 'These scores do not align with the evaluations conducted by LLM-based methods or human experts' is imprecise; the human evaluation did not produce comparable BERTScore-style numbers. Consider saying that the rankings differ from the proposed LLM-based metric.
  4. [Ethics Statement] There is a grammatical error: 'We use AI assistants, specifically Grammarly and ChatGPT, were utilized' should be 'We used AI assistants, specifically Grammarly and ChatGPT, to correct grammatical errors...'.
  5. [Throughout] The framework name is inconsistently typeset as 'ACCEL MAT' and 'ACCELMAT'; please standardize.
  6. [Table 1] The column headers in Table 1 ('LLM Agent', 'Diverse Mat', 'Diverse Prop', 'Tool Free') are abbreviated without explanation; expand them in the caption or in the table itself.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: ACCELMAT reports an empirical LLM-agent evaluation, not a derivation; the unverified no-leakage premise and same-rubric human check are validity caveats, not circular reductions.

full rationale

The paper contains no formal derivation chain whose output is equivalent to its input by construction. The central result is an experimental comparison of three agent configurations, with Closeness and Quality scores computed by an LLM judge against ground truth extracted from recent papers. No parameter is fitted to the ground truth and then reported as a prediction. The knowledge-graph configuration injects topic-relevant context retrieved by querying MatKG with keywords from the goal and constraints, but this is not equivalent to supplying the ground-truth materials or methods, and the generation step remains open-ended; the observed improvement is an empirical outcome rather than a mathematical identity. The human-expert evaluation does use the same rubric as the automated system, which weakens its status as an independent validation, but human raters could still disagree with the automated scores, so agreement is not guaranteed by definition. Finally, the Section 3 assertion that January 2024 publications lie beyond the knowledge cutoff of all employed LLMs is not verified for Claude-3.5-Sonnet, Gemini-1.5-Flash, or OpenAI-o1-preview, and the paper only documents cutoffs for Llama-3.1 and GPT-4o. This is a real correctness and data-leakage risk for the novelty claim, but it is an unsupported empirical premise, not a circularity. Accordingly, no circular step meeting the required evidence standard is present.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The work is not a mathematical derivation. Its results rest on the validity of the LLM-based evaluation, the no-leakage assumption, expert curation quality, and the reliability of MatKG. The listed axioms are domain assumptions that the paper does not independently verify.

free parameters (3)
  • Number of hypotheses generated per goal = 20
    Chosen to limit API cost and runtime; authors state more would potentially be better (Section 7).
  • Maximum feedback cycles = 5
    Hard-coded in Algorithm 1 as max_cycles=5 to cap cost.
  • Quality score weights = Unweighted average (1/6 each)
    Section 5.2 says unweighted average is used for simplicity.
assumptions (5)
  • domain assumption Papers published in January 2024 are absent from the training data of GPT-4o, Claude-3.5-Sonnet, Gemini-1.5-Flash, and OpenAI-o1-preview.
    Section 3 frames this as an essential guarantee for evaluating novel hypotheses; no leakage test is performed.
  • domain assumption Unanimous approval by the three LLM critics identifies hypotheses that meet the goal and constraints.
    All configurations use critic consensus to finalize hypotheses. The paper's Limitations admits unanimous LLM agreement does not guarantee scientific accuracy.
  • domain assumption OpenAI-o1-preview's rubric-based scores are a reliable proxy for materials science expert judgment.
    The evaluation agent is the primary outcome measure; human validation covers only 42 suggestions with no quantitative agreement.
  • domain assumption MatKG supplies relevant and trusted materials knowledge for hypothesis grounding.
    Section 6.3 treats MatKG as 'trusted knowledge graphs' and uses it as supplementary context.
  • domain assumption Manual extraction by materials science experts accurately reflects the materials and methods in the source papers.
    Ground truth correctness depends on this; no inter-annotator agreement is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agents." pith.science (2026). https://pith.science/paper/3VVKP7TX

@misc{pith2026250113299,
  author       = {Pith},
  title        = {Pith review of: Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3VVKP7TX}},
  note         = {Machine review of arXiv:2501.13299}
}
read the original abstract

Materials discovery and design are essential for advancing technology across various industries by enabling the development of application-specific materials. Recent research has leveraged Large Language Models (LLMs) to accelerate this process. We explore the potential of LLMs to generate viable hypotheses that, once validated, can expedite materials discovery. Collaborating with materials science experts, we curated a novel dataset from recent journal publications, featuring real-world goals, constraints, and methods for designing real-world applications. Using this dataset, we test LLM-based agents that generate hypotheses for achieving given goals under specific constraints. To assess the relevance and quality of these hypotheses, we propose a novel scalable evaluation metric that emulates the process a materials scientist would use to evaluate a hypothesis critically. Our curated dataset, proposed method, and evaluation framework aim to advance future research in accelerating materials discovery and design with LLMs.

Figures

Figures reproduced from arXiv: 2501.13299 by the authors.

Figure 1
Figure 1. Overview of our iterative hypothesis gener [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The left plot illustrates the Closeness metric scores across three evaluation criteria for the three configu [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. The left plot compares the performance of open-source and closed-source models on the Closeness metric, [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 3 citations worldwide. Full citation record

  1. Visualizing Graph-to-Answer Mechanism Recovery in Materials-Science Hypothesis Generation

    cs.CL 2026-08 conditional novelty 6.0 of 10

    Under graph corruption, mechanism recovery in Graph-PRefLexOR-8B concentrates in late synthesis and answer-start layers 30 and 36, not in transition layers 7 to 10.

Reference graph

Works this paper leans on

23 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [3]

    arXiv preprint arXiv:2412.16720

    Openai o1 system card. arXiv preprint arXiv:2412.16720. Anubhav Jain, Shyue Ping Ong, Geoffroy Hautier, Wei Chen, William Davidson Richards, Stephen T. Dacek, Shreyas Cholia, Dan Gunter, David Skinner, Ger- brand Ceder, and Kristin Aslaug Persson. 2013. Com- mentary: The materials project: A materials genome approach to accelerating materials innovation. ...

    arXiv 2013

  2. [5]

    Expert_1, Expert_2, Expert_3, Expert_4, Expert_5

    The healing mechanism should not rely on complex multi-component reactions but rather a single-component system C.2 Prompt for experts list finder We use the below prompt for dynamically setting the behaviour of Hypotheses Generator relevant to the provided Goal. The output of this prompt is set as the system prompt for the Hypotheses Generator as shown i...

    work page

  3. [6]

    arXiv preprint arXiv:2310.08511

    Honeybee: Progressive instruction finetuning of large language models for materials science. arXiv preprint arXiv:2310.08511. Natalia CM Spera, Cristina Salazar-Castro, Paula C Ál- varez de Eulate, Yury V Kolenko, and Juliana PS Sousa. 2024. Self–healing core–shell nanofibers for corrosion protective coatings for offshore structures. Progress in Organic C...

    arXiv 2024

  4. [14]

    The coating must maintain its structural integrity and protective capabilities even after mechani- cal damage

    work page

  5. [16]

    Give a comprehensive and detailed description of list of materials, along with their names and properties which achieve help the goal and satisfy all constraints

    The healing mechanism should not rely on complex multi-component reactions but rather a single-component system. Generate 20 novel and innovative suggestions that strictly satisfy the provided goal and constraints. While building on existing scientific knowledge, also incorporate creative and innovative thinking to propose unique solutions that push the b...

    work page

  6. [17]

    The material should incorporate a self-healing mechanism triggered by a simple environmental factor (e.g., water)

    work page

  7. [18]

    The self-healing material should allow multiple healing events

    work page

  8. [19]

    The coating must maintain its structural integrity and protective capabilities even after mechanical damage

    work page

Show all 23 references
  1. [20]

    The material should be compatible with scalable application techniques

  2. [21]

    Give a comprehensive and detailed description of list of materials, along with their names and properties which achieve help the goal and satisfy all constraints

    The healing mechanism should not rely on complex multi-component reactions but rather a single-component system. Relevant information from trusted Knowledge Graphs: Suggested Materials: Aluminum: Aluminum is known for its corrosion resistance and is commonly used in offshore e...

  3. [22]

    Scale: 1 - No Overlap: The generated hypothesis contains entirely different concepts with no connection to the core ideas or methods in the ground truth hypothesis

    Concept Overlap Scale: Definition: This scale evaluates how well the core ideas, methods, and scientific concepts in the generated hypothesis align with those in the ground truth hypothesis. Scale: 1 - No Overlap: The generated hypothesis contains entirely different concepts w...

  4. [23]

    Scale: 1 - Not Similar: The generated hypothesis has no overlapping or similar properties to the ground truth hypothesis

    Property Overlap: Definition: This scale assesses how closely the material properties described in the generated hypothesis align with those in the ground truth, including quantitative values and qualitative descriptions. Scale: 1 - Not Similar: The generated hypothesis has no...

  5. [24]

    Scale: 1 - No Match: None of the keywords or entities in the generated hypothesis match those in the ground truth

    Keyword Matching: Definition: This scale evaluates how well specific entities (e.g., material names, chemical compounds, synthesis methods) and keywords in the generated hypothesis match those in the ground truth. Scale: 1 - No Match: None of the keywords or entities in the ge...

  6. [25]

    Scale: 1 - Misaligned: The hypothesis does not relate to the re- search objectives in any meaningful way and ignores or violates specified constraints

    Alignment with Research Objectives and Constraints: Definition: Assesses how directly and effectively the generated hypothesis addresses the objectives specified in the goal statement while adhering to all given constraints and incorporating provided key points or clues. Scale...

  7. [26]

    Scale: 1 - Not Plausible: Contradicts fundamental scientific laws or principles

    Scientific Plausibility: Definition: Assesses whether the hypothesis is grounded in established scientific principles and theories within material science. Scale: 1 - Not Plausible: Contradicts fundamental scientific laws or principles. 2 - Slightly Plausible: Contains signifi...

  8. [27]

    Scale: 1 - Not Innovative: Restates existing knowledge without introducing new ideas

    Innovation and Novelty: Definition: Measures the degree to which the hypothesis introduces original ideas, perspectives, or methods not previously documented. Scale: 1 - Not Innovative: Restates existing knowledge without introducing new ideas. 2 - Slightly Innovative: Provide...

  9. [28]

    Scale: 1 - Not Testable: The hypothesis cannot be tested with current or foreseeable experimental techniques

    Testability: Definition: Evaluates how easily and effectively the hy- pothesis can be tested experimentally, considering the avail- ability of required techniques, equipment, and resources in materials science. Scale: 1 - Not Testable: The hypothesis cannot be tested with curr...

  10. [29]

    Scale: 1 - Not Feasible: Cannot be implemented with current technology, materials, or resources, or faces insurmountable obstacles in scale-up

    Feasibility and Scalability: Definition: Evaluates the practicality of implementing the hypothesis at various scales, from laboratory to industrial applications, focusing on existing infrastructure, technology, cost, and effort required to scale the process. Scale: 1 - Not Fea...

  11. [30]

    silane-based healing agent,

    Impact Potential: Definition: Assesses the hypothesis’s potential to significantly advance the field or address critical challenges in material science. Scale: 1 - No Impact: Unlikely to contribute meaningfully to the field. 2 - Low Impact: May offer minor contributions but la...

    2019

  12. [2012]

    Journal of Materials Science, 47:7317–7340

    From the computer to the laboratory: materials discovery and design using first-principles calcula- tions. Journal of Materials Science, 47:7317–7340. Ryan Jacobs, Maciej Polak, Lane Schultz, Hamed Mah- davi, Vasant Honavar, and Dane Morgan. 2024. Re- gression with large langu...

    2024

  13. [2022]

    Matter, 5(1):314–335

    An invertible crystallographic representation for general inverse design of inorganic crystals with targeted properties. Matter, 5(1):314–335. Pranav Shetty, Arunkumar Chitteth Rajan, Chris Kuen- neth, Sonakshi Gupta, Lakshmi Prerana Panchumarti, Lauren Holm, Chao Zhang, and R...

  14. [2023]

    npj Computational Ma- terials, 9(1):52

    A general-purpose material property data ex- traction pipeline from large polymer corpora using natural language processing. npj Computational Ma- terials, 9(1):52. Karan Singhal, Shekoofeh Azizi, Tao Tu, S Sara Mah- davi, Jason Wei, Hyung Won Chung, Nathan Scales, Ajay Tanwan...

    2022 arXiv

  15. [2024]

    arXiv preprint arXiv:2410.21317

    Matexpert: Decomposing materials discov- ery by mimicking human experts. arXiv preprint arXiv:2410.21317. Victor Fung, Shuyi Jia, Jiaxin Zhang, Sirui Bi, Junqi Yin, and Panchapakesan Ganesh. 2022. Atomic struc- ture generation from reconstructing structural finger- prints. Mac...

    2022 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.