Pith. sign in

REVIEW 4 major objections 5 minor 55 references

DiffractGPT: Atomic Structure Determination from X-ray Diffraction Patterns using Generative Pre-trained Transformer

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read DiffractGPT, a generative pre-trained transformer, predicts atomic crystal structures directly from powder X-ray diffraction patterns, and giving it the chemical composition sharply improves accuracy.

desk verdict A sensible transformer-for-XRD idea that currently lacks the numbers and experimental validation needed to back its 'robust tool' claim. read the letter →

arxiv 2508.08349 v1 pith:3VCQNVXV submitted 2025-08-11 cond-mat.mtrl-sci

Kamal Choudhary This is my paper
classification cond-mat.mtrl-sci
keywords crystalstructuredeterminationX-raydiffractiongenerativepre-trainedtransformerinversedesignpowdermaterialsinformaticsdeeplearning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that a generative pre-trained transformer, DiffractGPT, can predict atomic crystal structures directly from powder X-ray diffraction patterns, turning pattern analysis into a fast sequence-generation task. The model is trained on simulated XRD patterns computed from thousands of DFT-relaxed structures. It is evaluated in three modes: no chemical information, an element list, and an explicit chemical formula. The central finding is that adding chemical information, especially the full formula, substantially improves prediction accuracy. A sympathetic reader would care because crystal structure determination from powder diffraction is normally slow and expertise-heavy, and this points toward a fully automated alternative.

What carries the argument

The central mechanism is a generative pre-trained transformer (GPT-style model) applied to structure determination as sequence generation. Diffraction patterns are encoded as input tokens and crystal structures as output tokens, so the model learns a conditional distribution over structures given a pattern. Optional conditioning tokens carry the element list or chemical formula, and the paper shows these composition tokens are what most strongly disambiguate the inverse mapping. The key work it does is replacing a hand-crafted search over candidate structures with a learned generative model that can propose structures directly.

What would settle it

Take a held-out set of experimental powder XRD patterns whose crystal structures are already known, run DiffractGPT on them, and compare predicted structures with the known ground truth. If the success rate is no better than a simple baseline or random guessing, the claim that the model learns a transferable pattern-to-structure mapping is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that a generative pre-trained transformer can learn the inverse mapping from a powder XRD pattern to an atomic structure well enough to serve as a practical structure-determination tool. The model takes a simulated diffraction pattern as input and generates candidate structures as output, with optional conditioning on chemical composition. Across the three evaluated scenarios, the paper finds that predictions are weakest without composition, improve when an element list is provided, and are strongest when the explicit chemical formula is given. The paper further reports that the training process is straightforward and fast, which it presents as making the method

Load-bearing premise

The load-bearing premise is that simulated XRD patterns computed from DFT-relaxed structures are representative enough of real experimental powder diffraction data that a model trained only on simulations will make useful predictions on measured patterns.

Editorial extensions

If this is right

  • A researcher with only a powder XRD pattern could obtain candidate atomic structures without specialized crystallographic expertise or expensive search.
  • Supplying an element list or chemical formula makes the model's predictions markedly more reliable, enabling a practical composition-aware XRD analysis workflow.
  • Because training is reported as fast, the same approach could be re-run on new or larger structure databases to keep pace with materials discovery.
  • Fast automated structure prediction from diffraction could slot into high-throughput pipelines, flagging candidate structures for further refinement.

Reading between the lines

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

  • The clearest extension not demonstrated in the paper is transfer to experimental XRD data; if simulated and measured patterns differ in peak shape, background, or preferred orientation, fine-tuning on a modest set of experimentally measured patterns may be necessary.
  • Since the model is generative, it can plausibly output multiple candidate structures for one pattern, which would make it naturally suited to polymorphism and pseudo-symmetric cases where a single best structure is ambiguous.
  • The strong effect of chemical formula suggests a natural follow-up: predict composition from the pattern first, then use it to condition structure generation, making the pipeline usable when the chemistry is completely unknown.
  • The generated structures could be used as starting guesses for Rietveld or other refinement methods, bridging a fully automated proposal step with quantitative fitting.
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 / 5 minor

Summary. The paper proposes DiffractGPT, a generative pre-trained transformer that predicts atomic crystal structures directly from powder X-ray diffraction (XRD) patterns. The model is trained on simulated XRD patterns generated from JARVIS-DFT structures and is evaluated under three input conditions: no chemical information, an element list, and an explicit chemical formula. The abstract claims that adding chemical information significantly improves accuracy and that the method provides fast, accurate inverse design and a robust tool for materials discovery. No quantitative accuracy metrics, baselines, error bars, or experimental validation are reported in the supplied text.

Significance. The problem is important and timely: an automated, generative inverse mapping from powder XRD to crystal structure would be of broad use in materials characterization. The choice of a GPT-style architecture on a standardized DFT-derived dataset is a plausible and interesting idea, and the three-scenario ablation is a clean way to isolate the value of chemical priors. If the claimed performance is substantiated, this would be a meaningful step toward automated structure determination. However, as presented, the abstract provides no quantitative evidence, and the simulated-only evaluation leaves the central practical claim unsubstantiated. The contribution cannot be judged until concrete metrics and transfer-to-experiment evidence are supplied.

major comments (4)
  1. [Abstract] The central claim of 'fast and accurate inverse design' is not supported by any quantitative result. The abstract gives no accuracy metric (e.g., fraction of correctly predicted space groups, lattice-parameter error, composition recovery), no error bars, and no comparison baseline. A reader cannot tell whether the model outperforms chance or existing structure-solution methods. The manuscript must report test-set metrics for all three scenarios, together with baselines, before this claim can be evaluated.
  2. [JARVIS-DFT simulated XRD data] All results described are on simulated XRD patterns derived from JARVIS-DFT structures. Real experimental powder XRD includes background, preferred orientation, peak-shape asymmetries, wavelength-dependent effects, and sample-dependent intensities that are absent or simplified in simulation. Without evaluation on experimental patterns (e.g., from ICSD/COD or measured samples), or at least a domain-adaptation analysis with simulated noise, the claim that DiffractGPT 'bridges ... experimental communities' is unsupported. A concrete test would be to apply the trained model to a set of experimental powder patterns and quantify the accuracy drop relative to simulated data.
  3. [Full text (as supplied)] The body of the manuscript is not readable: it consists largely of replacement characters and scrambled text, with no recoverable sections, equations, tables, figures, or data-generation details. An unrelated arXiv header (2508.08354, hep-ph) is also mixed in. As a result, the training procedure, architecture details, dataset split, and evaluation protocol cannot be inspected. This is load-bearing: the paper's technical claims are not verifiable from the submitted text.
  4. [Scenario comparison] The abstract states that 'incorporating chemical information significantly enhances prediction accuracy,' but no comparative numbers are given for the three scenarios. Without reporting performance for the no-information, element-list, and explicit-formula cases (including the magnitude of the improvement and its statistical significance), the claim remains qualitative. A single table comparing all three settings is necessary.
minor comments (5)
  1. [Abstract] Specify the exact dataset size and JARVIS-DFT version instead of 'thousands of atomic structures'.
  2. [Abstract] Define 'prediction accuracy' operationally (e.g., top-1 space group success rate, mean absolute error in lattice parameters) in the abstract so claims are falsifiable.
  3. [Full text] Numerous glyphs appear as replacement characters, making the methods section unreadable. Please ensure the PDF/text encoding is correct and that the intended arXiv identifier matches the manuscript.
  4. [Abstract] The phrase 'bridging gaps between computational, data science, and experimental communities' is overstated given the absence of any experimental data; consider softening or substantiating it.
  5. [General] The manuscript should include comparisons with established structure-solution methods (e.g., Rietveld refinement, simulated annealing) and prior deep-learning approaches for inverse XRD analysis, with references, to support the claim of a significant advancement.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found; the simulated-data pipeline is a domain-transfer concern, not a logical circularity.

full rationale

The provided full text is largely unreadable due to encoding corruption, leaving the abstract as the only usable evidence. The abstract reports training a generative model on simulated XRD patterns from the JARVIS-DFT dataset and evaluating it under three input conditions (no chemical information, element list, explicit formula). This is an empirical machine-learning pipeline, not a formal derivation chain. No equation is presented, no parameter is fitted to a target quantity and then renamed as a prediction, and no load-bearing self-citation or imported uniqueness theorem is invoked. The possible weakness is that the model is trained and evaluated on simulated patterns, so its transfer to experimental XRD data is unproven; however, that is a generalization/validation risk, not circularity. Nothing in the available text reduces a claimed prediction to its own input by construction. Accordingly, no circular steps are identified and the circularity score is 0.

Assumptions & free parameters 1 free parameters · 2 assumptions · 0 invented entities

The central claim rests on the quality and representativeness of the simulated training data and on the untested assumption that a model trained on that data transfers to real measurements. No new physical entities are introduced.

free parameters (1)
  • Model hyperparameters = not reported
    The transformer's architecture size, number of layers, training epochs, and loss weights are not given. These are chosen by the author and directly affect the reported accuracy.
assumptions (2)
  • domain assumption Simulated XRD patterns from JARVIS-DFT are a valid proxy for experimental XRD patterns
    The model is trained and evaluated on simulated patterns. The abstract gives no evidence that this transfer holds, and this is the load-bearing premise for any practical use.
  • domain assumption The JARVIS-DFT dataset provides sufficient structural diversity for generalization
    The abstract says 'thousands of atomic structures' but does not state composition or structure diversity. The model's coverage of chemical space is unknown.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DiffractGPT: Atomic Structure Determination from X-ray Diffraction Patterns using Generative Pre-trained Transformer." pith.science (2026). https://pith.science/paper/3VCQNVXV

@misc{pith2026250808349,
  author       = {Pith},
  title        = {Pith review of: DiffractGPT: Atomic Structure Determination from X-ray Diffraction Patterns using Generative Pre-trained Transformer},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3VCQNVXV}},
  note         = {Machine review of arXiv:2508.08349}
}
read the original abstract

Crystal structure determination from powder diffraction patterns is a complex challenge in materials science, often requiring extensive expertise and computational resources. This study introduces DiffractGPT, a generative pre-trained transformer model designed to predict atomic structures directly from X-ray diffraction (XRD) patterns. By capturing the intricate relationships between diffraction patterns and crystal structures, DiffractGPT enables fast and accurate inverse design. Trained on thousands of atomic structures and their simulated XRD patterns from the JARVIS-DFT dataset, we evaluate the model across three scenarios: (1) without chemical information, (2) with a list of elements, and (3) with an explicit chemical formula. The results demonstrate that incorporating chemical information significantly enhances prediction accuracy. Additionally, the training process is straightforward and fast, bridging gaps between computational, data science, and experimental communities. This work represents a significant advancement in automating crystal structure determination, offering a robust tool for data-driven materials discovery and design.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

55 extracted references · 48 canonical work pages

  1. [1]

    Elements of modern X-ray physics; John Wiley & Sons, West Sussex, UK, 2011

    Als-Nielsen, J.; McMorrow, D. Elements of modern X-ray physics; John Wiley & Sons, West Sussex, UK, 2011

    work page 2011

  2. [2]

    Fundamentals of crystallography; Oxford university press, Oxford, UK, 2002

    Giacovazzo, C. Fundamentals of crystallography; Oxford university press, Oxford, UK, 2002

    work page 2002

  3. [3]

    X-ray metrology for advanced microelectronics

    Wyon, C. X-ray metrology for advanced microelectronics. The European Physical Journal-Applied Physics 2010, 49, 20101

    work page 2010

  4. [4]

    F.; Schaak, R

    Holder, C. F.; Schaak, R. E. Tutorial on powder X-ray diffraction for characterizing nanoscale materials. ACS Nano 2019, 13, 7359--7365

    work page 2019

  5. [5]

    Brown, J. G. X-rays and Their Applications; Springer New York, 2012

    work page 2012

  6. [6]

    FullProf

    Rodriguez-Carvajal, J.; Roisnel, T. FullProf. 98 and WinPLOTR: new windows 95/NT applications for diffraction. Commission for Powder Diffraction, International Union of Crystallography, Newsletter 1998, 20, May--August

    work page 1998

  7. [7]

    C.; Von Dreele, R

    Larson, A. C.; Von Dreele, R. B. General Structure Analysis System (GSAS). Los Alamos National Laboratory Report lAUR 1985, 86

    work page 1985

  8. [8]

    GenX 3: the latest generation of an established tool

    Glavic, A.; Bj \"o rck, M. GenX 3: the latest generation of an established tool. Journal of applied crystallography 2022, 55, 1063--1071

    work page 2022

Show all 55 references
  1. [9]

    Coelho, A. A. TOPAS and TOPAS-Academic: an optimization program integrating computer algebra and crystallographic objects written in C++. Journal of Applied Crystallography 2018, 51, 210--218

    2018

  2. [10]

    Rietveld texture analysis from TOF neutron diffraction data

    Wenk, H.-R.; Lutterotti, L.; Vogel, S. Rietveld texture analysis from TOF neutron diffraction data. Powder Diffraction 2010, 25, 283--296

    2010

  3. [11]

    W.; Choudhary, A.; Agrawal, A.; Billinge, S

    Choudhary, K.; DeCost, B.; Chen, C.; Jain, A.; Tavazza, F.; Cohn, R.; Park, C. W.; Choudhary, A.; Agrawal, A.; Billinge, S. J.; others Recent advances and applications of deep learning methods in materials science. npj Computational Materials 2022, 8, 59

    2022

  4. [12]

    K.; Choudhary, K.; Mehta, A.; Smith, R.; Kusne, G.; Tavazza, F.; Vlcek, L.; Ziatdinov, M.; Kalinin, S

    Vasudevan, R. K.; Choudhary, K.; Mehta, A.; Smith, R.; Kusne, G.; Tavazza, F.; Vlcek, L.; Ziatdinov, M.; Kalinin, S. V.; Hattrick-Simpers, J. Materials science in the artificial intelligence age: high-throughput library generation, machine learning, and a pathway from correlat...

    2019

  5. [13]

    R.; Botti, S.; Marques, M

    Schmidt, J.; Marques, M. R.; Botti, S.; Marques, M. A. Recent advances and applications of machine learning in solid-state materials science. npj computational materials 2019, 5, 83

    2019

  6. [14]

    G ov --- nist.gov

    C H I P S . G ov --- nist.gov. https://www.nist.gov/chips, [Accessed 10-10-2024]

    2024

  7. [15]

    X-ray diffraction data analysis by machine learning methods—a review

    Surdu, V.-A.; Gy o rgy, R. X-ray diffraction data analysis by machine learning methods—a review. Applied Sciences 2023, 13, 9992

    2023

  8. [16]

    B.; Chung, J.; Jung, J.; Sohn, K.; Singh, S

    Park, W. B.; Chung, J.; Jung, J.; Sohn, K.; Singh, S. P.; Pyo, M.; Shin, N.; Sohn, K.-S. Classification of crystal structure using a convolutional neural network. IUCrJ 2017, 4, 486--494

    2017

  9. [17]

    Machine learning-assisted close-set X-ray diffraction phase identification of transition metals

    Zhdanov, M.; Zhdanov, A. Machine learning-assisted close-set X-ray diffraction phase identification of transition metals. arXiv preprint arXiv:2305.15410 2023, [Accessed 10-10-2024]

    2023 arXiv

  10. [18]

    D.; Lee, J.-W.; Park, W

    Lee, B. D.; Lee, J.-W.; Park, W. B.; Park, J.; Cho, M.-Y.; Pal Singh, S.; Pyo, M.; Sohn, K.-S. Powder X-ray diffraction pattern is all you need for machine-learning-based symmetry identification and property prediction. Advanced Intelligent Systems 2022, 4, 2200042

    2022

  11. [19]

    M.; Naujoks, D.; Olds, D.; Ludwig, A

    Banko, L.; Maffettone, P. M.; Naujoks, D.; Olds, D.; Ludwig, A. Deep learning for visualization and novelty detection in large X-ray diffraction datasets. Npj Computational Materials 2021, 7, 104

    2021

  12. [20]

    Artifact identification in X-ray diffraction data using machine learning methods

    Yanxon, H.; Weng, J.; Parraga, H.; Xu, W.; Ruett, U.; Schwarz, N. Artifact identification in X-ray diffraction data using machine learning methods. Journal of Synchrotron Radiation 2023, 30, 137--146

    2023

  13. [21]

    N.; Stanovov, V

    Zaloga, A. N.; Stanovov, V. V.; Bezrukova, O. E.; Dubinin, P. S.; Yakimov, I. S. Crystal symmetry classification from powder X-ray diffraction patterns using a convolutional neural network. Materials Today Communications 2020, 25, 101662

    2020

  14. [22]

    Proceedings of the National Academy of Sciences 2022, 119, e2109665119

    Venderley, J.; Mallayya, K.; Matty, M.; Krogstad, M.; Ruff, J.; Pleiss, G.; Kishore, V.; Mandrus, D.; Phelan, D.; Poudel, L.; others Harnessing interpretable and unsupervised machine learning to address big data from modern X-ray diffraction. Proceedings of the National Academ...

    2022

  15. [23]

    B.; Lee, J

    Lee, J.-W.; Park, W. B.; Lee, J. H.; Singh, S. P.; Sohn, K.-S. A deep-learning technique for phase identification in multiphase inorganic compounds using synthetic XRD powder patterns. Nature communications 2020, 11, 86

    2020

  16. [24]

    M.; Banko, L.; Cui, P.; Lysogorskiy, Y.; Little, M

    Maffettone, P. M.; Banko, L.; Cui, P.; Lysogorskiy, Y.; Little, M. A.; Olds, D.; Ludwig, A.; Cooper, A. I. Crystallography companion agent for high-throughput materials discovery. Nature Computational Science 2021, 1, 290--297

    2021

  17. [25]

    Oviedo, F.; Ren, Z.; Sun, S.; Settens, C.; Liu, Z.; Hartono, N. T. P.; Ramasamy, S.; DeCost, B. L.; Tian, S. I.; Romano, G.; others Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks. npj Computational Mat...

    2019

  18. [26]

    Crystal Structure Assignment for Unknown Compounds from X-ray Diffraction Patterns with Deep Learning

    Chen, L.; Wang, B.; Zhang, W.; Zheng, S.; Chen, Z.; Zhang, M.; Dong, C.; Pan, F.; Li, S. Crystal Structure Assignment for Unknown Compounds from X-ray Diffraction Patterns with Deep Learning. Journal of the American Chemical Society 2024, 146, 8098

    2024

  19. [27]

    Machine learning-accelerated discovery of novel 2D ferromagnetic materials with strong magnetization

    Xin, C.; Yin, Y.; Song, B.; Fan, Z.; Song, Y.; Pan, F. Machine learning-accelerated discovery of novel 2D ferromagnetic materials with strong magnetization. Chip 2023, 2, 100071

    2023

  20. [28]

    Advancements in High-Throughput Screening and Machine Learning Design for 2D Ferromagnetism: A Comprehensive Review

    Xin, C.; Song, B.; Jin, G.; Song, Y.; Pan, F. Advancements in High-Throughput Screening and Machine Learning Design for 2D Ferromagnetism: A Comprehensive Review. Advanced Theory and Simulations 2023, 6, 2300475

    2023

  21. [29]

    AtomGPT: Atomistic Generative Pretrained Transformer for Forward and Inverse Materials Design

    Choudhary, K. AtomGPT: Atomistic Generative Pretrained Transformer for Forward and Inverse Materials Design. The Journal of Physical Chemistry Letters 2024, 15, 6909--6917

    2024

  22. [30]

    N.; Kaiser, .; Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; Polosukhin, I. Attention is all you need. Advances in neural information processing systems 2017, 30

    2017

  23. [31]

    O'Reilly Media, Inc

    Tunstall, L.; Von Werra, L.; Wolf, T. Natural language processing with transformers; " O'Reilly Media, Inc.", 2022

    2022

  24. [32]

    Rothman, D. Transformers for Natural Language Processing: Build innovative deep neural network architectures for NLP with Python, PyTorch, TensorFlow, BERT, RoBERTa, and more; Packt Publishing Ltd, 2021

    2021

  25. [33]

    A brief overview of ChatGPT: The history, status quo and potential future development

    Wu, T.; He, S.; Liu, J.; Sun, S.; Liu, K.; Han, Q.-L.; Tang, Y. A brief overview of ChatGPT: The history, status quo and potential future development. IEEE/CAA Journal of Automatica Sinica 2023, 10, 1122--1136

    2023

  26. [34]

    F.; Picciani, P.; Salles, B.; Follmer, C.; Oliveira Jr, O

    Pimentel, A.; Wagener, A.; da Silveira, E. F.; Picciani, P.; Salles, B.; Follmer, C.; Oliveira Jr, O. N. Challenging ChatGPT with Chemistry-Related Subjects. ChemarXiv preprint 2023, https://doi.org/10.26434/chemrxiv-2023-xl6w3, Accessed 10-10-2024

    2023 doi

  27. [35]

    M.; Schwaller, P.; Ortega-Guerrero, A.; Smit, B

    Jablonka, K. M.; Schwaller, P.; Ortega-Guerrero, A.; Smit, B. Leveraging large language models for predictive chemistry. Nature Machine Intelligence 2024, 1--9

    2024

  28. [36]

    P.; Morgan, D

    Polak, M. P.; Morgan, D. Extracting accurate materials data from research papers with conversational language models and prompt engineering. Nature Communications 2024, 15, 1569

    2024

  29. [37]

    F.; DeCost, B.; Biacchi, A

    Wines, D.; Gurunathan, R.; Garrity, K. F.; DeCost, B.; Biacchi, A. J.; Tavazza, F.; Choudhary, K. Recent progress in the JARVIS infrastructure for next-generation data-driven materials design. Applied Physics Reviews 2023, 10

    2023

  30. [38]

    F.; Reid, A

    Choudhary, K.; Garrity, K. F.; Reid, A. C.; DeCost, B.; Biacchi, A. J.; Hight Walker, A. R.; Trautt, Z.; Hattrick-Simpers, J.; Kusne, A. G.; Centrone, A.; others The joint automated repository for various integrated simulations (JARVIS) for data-driven materials design. npj co...

    2020

  31. [39]

    Q.; Sablayrolles, A.; Mensch, A.; Bamford, C.; Chaplot, D

    Jiang, A. Q.; Sablayrolles, A.; Mensch, A.; Bamford, C.; Chaplot, D. S.; Casas, D. d. l.; Bressand, F.; Lengyel, G.; Lample, G.; Saulnier, L.; others Mistral 7B. arXiv preprint arXiv:2310.06825 2023, [Accessed 10-10-2024]

    2023 arXiv

  32. [40]

    Nature 2021, 596, 583--589

    Jumper, J.; Evans, R.; Pritzel, A.; Green, T.; Figurnov, M.; Ronneberger, O.; Tunyasuvunakool, K.; Bates, R.; Z \' dek, A.; Potapenko, A.; others Highly accurate protein structure prediction with AlphaFold. Nature 2021, 596, 583--589

    2021

  33. [41]

    C.; Chowdhury, S.; Van Nguyen, N.; Trautt, Z.; Newrock, M

    Choudhary, K.; Zhang, Q.; Reid, A. C.; Chowdhury, S.; Van Nguyen, N.; Trautt, Z.; Newrock, M. W.; Congo, F. Y.; Tavazza, F. Computational screening of high-performance optoelectronic materials using OptB88vdW and TB-mBJ formalisms. Scientific data 2018, 5, 1--12

    2018

  34. [42]

    J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; Chen, W

    Hu, E. J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; Chen, W. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 2021, [Accessed 10-10-2024]

    2021 arXiv

  35. [43]

    https://github.com/unslothai/unsloth, Accessed: 2024-10-10

    Unsloth AI Unsloth - GitHub Repository. https://github.com/unslothai/unsloth, Accessed: 2024-10-10

    2024

  36. [44]

    arXiv preprint arXiv:2307.09288 2023, [Accessed 10-10-2024]

    Touvron, H.; Martin, L.; Stone, K.; Albert, P.; Almahairi, A.; Babaei, Y.; Bashlykov, N.; Batra, S.; Bhargava, P.; Bhosale, S.; others Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288 2023, [Accessed 10-10-2024]

    2023 arXiv

  37. [45]

    arXiv preprint arXiv:2302.13971 2023, [Accessed 10-10-2024]

    Touvron, H.; Lavril, T.; Izacard, G.; Martinet, X.; Lachaux, M.-A.; Lacroix, T.; Rozi \`e re, B.; Goyal, N.; Hambro, E.; Azhar, F.; others Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971 2023, [Accessed 10-10-2024]

    2023 arXiv

  38. [46]

    Taori, R.; Gulrajani, I.; Zhang, T.; Dubois, Y.; Li, X.; Guestrin, C.; Liang, P.; Hashimoto, T. B. Stanford alpaca: an instruction-following llama model (2023). URL https://github. com/tatsu-lab/stanford\_alpaca 2023, [Accessed 10-10-2024]

    2023

  39. [47]

    Roformer: Enhanced transformer with rotary position embedding

    Su, J.; Ahmed, M.; Lu, Y.; Pan, S.; Bo, W.; Liu, Y. Roformer: Enhanced transformer with rotary position embedding. Neurocomputing 2024, 568, 127063

    2024

  40. [48]

    P.; Richards, W

    Ong, S. P.; Richards, W. D.; Jain, A.; Hautier, G.; Kocher, M.; Cholia, S.; Gunter, D.; Chevrier, V. L.; Persson, K. A.; Ceder, G. Python Materials Genomics (pymatgen): A robust, open-source python library for materials analysis. Computational Materials Science 2013, 68, 314--319

    2013

  41. [49]

    Crystal diffusion variational autoencoder for periodic material generation

    Xie, T.; Fu, X.; Ganea, O.-E.; Barzilay, R.; Jaakkola, T. Crystal diffusion variational autoencoder for periodic material generation. arXiv preprint arXiv:2110.06197 2021, [Accessed 10-10-2024]

    2021 arXiv

  42. [50]

    Mlatticeabc: generic lattice constant prediction of crystal materials using machine learning

    Li, Y.; Yang, W.; Dong, R.; Hu, J. Mlatticeabc: generic lattice constant prediction of crystal materials using machine learning. ACS omega 2021, 6, 11585--11594

    2021

  43. [51]

    Unified graph neural network force-field for the periodic table: solid state applications

    Choudhary, K.; DeCost, B.; Major, L.; Butler, K.; Thiyagalingam, J.; Tavazza, F. Unified graph neural network force-field for the periodic table: solid state applications. Digital Discovery 2023, 2, 346--355

    2023

  44. [52]

    npj Computational Materials 2021, 7, 115

    Lim, B.; Bellec, E.; Dupraz, M.; Leake, S.; Resta, A.; Coati, A.; Sprung, M.; Almog, E.; Rabkin, E.; Schulli, T.; others A convolutional neural network for defect classification in Bragg coherent X-ray diffraction. npj Computational Materials 2021, 7, 115

    2021

  45. [53]

    Convolutional neural network analysis of x-ray diffraction data: strain profile retrieval in ion beam modified materials

    Boulle, A.; Debelle, A. Convolutional neural network analysis of x-ray diffraction data: strain profile retrieval in ion beam modified materials. Machine Learning: Science and Technology 2023, 4, 015002

    2023

  46. [54]

    J.; Cabana, J.; Cherukara, M

    Judge, W.; Chan, H.; Sankaranarayanan, S.; Harder, R. J.; Cabana, J.; Cherukara, M. J. Defect identification in simulated Bragg coherent diffraction imaging by automated AI. MRS Bulletin 2023, 48, 124--133

    2023

  47. [55]

    and ``The chemical formula is ... The XRD is ... Generate atomic structure description with lattice lengths, angles, coordinates, and atom types,

    Motamedi, M.; Shidpour, R.; Ezoji, M. LSTM-based framework for predicting point defect percentage in semiconductor materials using simulated XRD patterns. Scientific Reports 2024, 14, 24353 mcitethebibliography final.tex0000664000000000000000000014572115046366154011415 0ustar ...

    2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.