pith. sign in

arxiv: 2605.28077 · v1 · pith:3U6ZZNVKnew · submitted 2026-05-27 · 💻 cs.AI

MACReD: A Multi-Agent Collaborative Reasoning Framework for Reaction Diagram Parsing

Chuang Tang , Chenhao Lin , Yin Xu , Hao Wang , Jinrui Zhou , Xin Li , Mingjun Xiao , Enhong Chen This is my paper

Pith reviewed 2026-06-29 12:30 UTC · model grok-4.3

classification 💻 cs.AI
keywords multi-agent frameworkreaction diagram parsingchemical reaction extractionvision-language modelsmultigraph fusionRxnScribe benchmarkhierarchical reasoning
0
0 comments X

The pith

A multi-agent framework parses complex chemical reaction diagrams by coordinating specialized agents and fusing their outputs into consistent reactions.

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

The paper presents MACReD as a hierarchical system that assigns separate agents to detect molecules, arrows, and text in reaction diagrams, then combines their outputs through multigraph fusion to enforce overall spatial and chemical consistency. Existing single-model approaches often lose coherence on intertwined layouts and fail to integrate multiple types of visual information at once. The authors test the approach on the RxnScribe benchmark and report higher accuracy than the prior baseline method. If the coordination works as described, literature diagrams that previously produced fragmented or invalid extractions become usable for automated reaction databases.

Core claim

MACReD uses a planning and perception layer with fine-grained detection agents for molecular structures, arrows, and text, followed by a reasoning layer that applies multigraph fusion to merge heterogeneous cues and produce chemically valid global reaction reconstructions, reaching F1 scores of 75.2 percent under hard match and 84.6 percent under soft match on the RxnScribe benchmark while the baseline reaches 69.1 percent and 80.0 percent.

What carries the argument

Hierarchical multi-agent coordination inside a vision-language model, where perception agents feed a reasoning layer that performs multigraph fusion to enforce chemically consistent global reasoning across the diagram.

If this is right

  • The method improves extraction accuracy on multi-step and tree-structured reactions that appear in literature.
  • It maintains performance across varied diagram layouts that include intertwined visual elements.
  • The multigraph fusion step allows integration of recognition outputs with higher-level chemical constraints.
  • Benchmark gains indicate that agent specialization reduces errors that arise when a single model attempts all subtasks simultaneously.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same agent-division pattern could be tested on other types of scientific diagrams that mix graphics and text, such as flow charts in engineering papers.
  • If the multigraph fusion proves stable, it might serve as a template for combining outputs from multiple vision tools without retraining the underlying models.
  • The approach leaves open whether the same hierarchy would still hold when diagrams are drawn in non-standard styles not represented in the current benchmark.

Load-bearing premise

Dividing diagram parsing among specialized agents and fusing their results with a multigraph will preserve spatial coherence and chemical validity on diagrams that single models cannot handle.

What would settle it

A collection of diagrams containing overlapping elements or multi-step branches where the agents produce locally correct detections but the fused output yields chemically invalid reactions or broken spatial relations.

Figures

Figures reproduced from arXiv: 2605.28077 by Chenhao Lin, Chuang Tang, Enhong Chen, Hao Wang, Jinrui Zhou, Mingjun Xiao, Xin Li, Yin Xu.

Figure 1
Figure 1. Figure 1: Examples of reaction diagrams in chemistry litera [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of MACReD, illustrating agent-level collaboration across planning, perception, and reasoning layers. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Example of the reaction diagram parsing task. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Example of routing mechanism. controller that jointly analyzes the user query and the visual con￾tent of the input reaction diagram, and dynamically determines an adaptive sequence of downstream agent invocations. Rather than relying on explicit search-based planning or hand￾crafted utility maximization, the Planning Agent formulates deci￾sion making as a context-conditioned agent routing problem. Its role… view at source ↗
Figure 5
Figure 5. Figure 5: Hard Match and Soft Match evaluation protocols. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: MACReD’s reaction diagram parsing performance [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Planner Agent prompt for context-aware agent selection. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Molecule-Recognition Agent prompt for recognition molecular entities. [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Reaction-Combiner Agent prompt for refining reaction graphs into chemically valid reactions. [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Structured JSON output from Reaction Diagram Parsing, showing identified reactants, products, conditions, and [PITH_FULL_IMAGE:figures/full_fig_p014_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Visualization of Reaction Diagram Parsing JSON output, illustrating the spatial layout of molecules, conditions, and [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Qualitative comparison between MACReD (top) and RxnScribe (bottom). MACReD produces more complete reaction [PITH_FULL_IMAGE:figures/full_fig_p015_12.png] view at source ↗
read the original abstract

Parsing chemical reaction diagrams from scientific literature is challenging due to heterogeneous layouts, intertwined visual elements, and the difficulty of integrating recognition and reasoning. Existing vision-language models advance multimodal understanding but still fail on complex diagrams, struggling to maintain spatial coherence and to integrate multidimensional information during reasoning. To address these issues, we propose MACReD, a hierarchical multi-agent framework that coordinates specialized agents for molecular perception, arrow understanding, text extraction, and reaction reconstruction within a unified VLM-guided architecture. The planning and perception layers use flexible, fine-grained detection to handle visual complexity, while the reasoning layer uses a multigraph fusion mechanism to integrate heterogeneous cues and enforce chemically consistent global reasoning. Experiments on the RxnScribe benchmark show that MACReD achieves state-of-the-art performance, with F1 scores of 75.2% and 84.6% under hard and soft match criteria, outperforming the RxnScribe baseline, which obtains 69.1% and 80.0%, respectively. These results demonstrate the robustness of MACReD across diverse diagram layouts, including multi-step and tree-structured reactions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The paper proposes MACReD, a hierarchical multi-agent framework for parsing chemical reaction diagrams. It coordinates specialized agents for molecular perception, arrow understanding, text extraction, and reaction reconstruction inside a VLM-guided architecture, using flexible detection in planning/perception layers and a multigraph fusion mechanism in the reasoning layer to enforce spatial coherence and chemical consistency. On the RxnScribe benchmark the method reports F1 scores of 75.2% (hard match) and 84.6% (soft match), outperforming the RxnScribe baseline (69.1%/80.0%).

Significance. If the reported gains are shown to be robust and attributable to the multi-agent coordination and multigraph fusion rather than backbone strength alone, the work would demonstrate a concrete advance in multimodal scientific-document understanding, offering a template for integrating heterogeneous visual and textual cues in complex diagrams.

major comments (1)
  1. [§4] §4 (Experiments) and abstract: the central SOTA claim rests on F1 scores of 75.2/84.6 vs. 69.1/80.0 without any description of data splits, evaluation protocol (exact definition of hard/soft match), whether the baseline was re-run under identical conditions, error bars, random seeds, or ablation studies isolating the contribution of hierarchical agent coordination versus a stronger VLM. These omissions make it impossible to attribute the performance delta to the claimed architectural mechanisms.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the single major comment below.

read point-by-point responses

  1. Referee: [§4] §4 (Experiments) and abstract: the central SOTA claim rests on F1 scores of 75.2/84.6 vs. 69.1/80.0 without any description of data splits, evaluation protocol (exact definition of hard/soft match), whether the baseline was re-run under identical conditions, error bars, random seeds, or ablation studies isolating the contribution of hierarchical agent coordination versus a stronger VLM. These omissions make it impossible to attribute the performance delta to the claimed architectural mechanisms.

    Authors: We agree that the current manuscript does not provide sufficient detail on these experimental aspects, which limits the ability to attribute gains specifically to the multi-agent coordination and multigraph fusion. In the revised manuscript we will expand §4 to include: (i) the data splits used from the RxnScribe benchmark, (ii) the exact definitions of hard and soft match as applied, (iii) confirmation that the baseline was re-run under identical conditions, (iv) error bars and random seeds from multiple runs, and (v) ablation studies that isolate the hierarchical agent coordination and multigraph fusion components versus backbone strength alone. These additions will directly address the attribution concern. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical framework evaluated on external benchmark with no derivation chain

full rationale

The paper describes a multi-agent architecture (MACReD) for diagram parsing and reports F1 scores on the named RxnScribe benchmark against a stated baseline. No equations, first-principles derivations, fitted parameters renamed as predictions, or load-bearing self-citations appear in the abstract or described content. The central claim reduces to an empirical performance delta on an external dataset, which is self-contained and does not reduce to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No free parameters, axioms, or invented entities are identifiable from the abstract; the work relies on standard VLM and multi-agent concepts from prior literature.

pith-pipeline@v0.9.1-grok · 5742 in / 1135 out tokens · 29427 ms · 2026-06-29T12:30:32.612105+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

44 extracted references · 16 canonical work pages · 10 internal anchors

  1. [1]

    Hu Ding, Pengxiang Hua, and Zhen Huang. 2025. Survey on Recent Progress of AI for Chemistry: Methods, Applications, and Opportunities.arXiv preprint arXiv:2502.17456(2025)

    work page arXiv 2025

  2. [2]

    Seoin Back, Alán Aspuru-Guzik, Michele Ceriotti, Ganna Gryn’ova, Bartosz Grzybowski, Geun Ho Gu, Jason Hein, Kedar Hippalgaonkar, Rodrigo Hormáz- abal, Yousung Jung, et al. 2024. Accelerated chemical science with AI.Digital Discovery3, 1 (2024), 23–33

    2024

  3. [3]

    Joe R McDaniel and Jason R Balmuth. 1992. Kekule: OCR-optical chemical (structure) recognition.Journal of chemical information and computer sciences 32, 4 (1992), 373–378

    1992

  4. [4]

    Richard Casey, Stephen Boyer, Paul Healey, Alex Miller, Bernadette Oudot, and Karl Zilles. 1993. Optical recognition of chemical graphics. InProceedings of 2nd International Conference on Document Analysis and Recognition (ICDAR’93). IEEE, 627–631

    1993

  5. [5]

    P Ibison, M Jacquot, F Kam, AG Neville, Richard W Simpson, C Tonnelier, T Venczel, and A Peter Johnson. 1993. Chemical literature data extraction: the CLiDE Project.Journal of Chemical Information and Computer Sciences33, 3 (1993), 338–344

    1993

  6. [6]

    Aniko T Valko and A Peter Johnson. 2009. CLiDE Pro: the latest generation of CLiDE, a tool for optical chemical structure recognition.Journal of chemical information and modeling49, 4 (2009), 780–787

    2009

  7. [7]

    Igor V Filippov and Marc C Nicklaus. 2009. Optical structure recognition software to recover chemical information: OSRA, an open source solution

    2009

  8. [8]

    Paolo Frasconi, Francesco Gabbrielli, Marco Lippi, and Simone Marinai. 2014. Markov logic networks for optical chemical structure recognition.Journal of chemical information and modeling54, 8 (2014), 2380–2390

    2014

  9. [9]

    Joshua Staker, Kyle Marshall, Robert Abel, and Carolyn M McQuaw. 2019. Molecu- lar structure extraction from documents using deep learning.Journal of chemical information and modeling59, 3 (2019), 1017–1029

    2019

  10. [10]

    Kohulan Rajan, Achim Zielesny, and Christoph Steinbeck. 2021. DECIMER 1.0: deep learning for chemical image recognition using transformers.Journal of Cheminformatics13, 1 (2021), 61

    2021

  11. [11]

    Sanghyun Yoo, Ohyun Kwon, and Hoshik Lee. 2022. Image-to-graph transform- ers for chemical structure recognition. InICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 3393–3397

    2022

  12. [12]

    Yujie Qian, Jiang Guo, Zhengkai Tu, Zhening Li, Connor W Coley, and Regina Barzilay. 2023. MolScribe: robust molecular structure recognition with image- to-graph generation.Journal of chemical information and modeling63, 7 (2023), 1925–1934

    2023

  13. [13]

    Yufan Chen, Ching Ting Leung, Yong Huang, Jianwei Sun, Hao Chen, and Hanyu Gao. 2024. MolNexTR: a generalized deep learning model for molecular image recognition.Journal of Cheminformatics16, 1 (2024), 141

    2024

  14. [14]

    Dat Quoc Nguyen, Zenan Zhai, Hiyori Yoshikawa, Biaoyan Fang, Christian Druckenbrodt, Camilo Thorne, Ralph Hoessel, Saber A Akhondi, Trevor Cohn, Timothy Baldwin, et al. 2020. ChEMU: named entity recognition and event extrac- tion of chemical reactions from patents. InEuropean conference on information retrieval. Springer, 572–579

    2020

  15. [15]

    Jiang Guo, A Santiago Ibanez-Lopez, Hanyu Gao, Victor Quach, Connor W Coley, Klavs F Jensen, and Regina Barzilay. 2021. Automated chemical reaction extraction from scientific literature.Journal of chemical information and modeling 62, 9 (2021), 2035–2045

    2021

  16. [16]

    Damian M Wilary and Jacqueline M Cole. 2021. ReactionDataExtractor: A tool for automated extraction of information from chemical reaction schemes.Journal of chemical information and modeling61, 10 (2021), 4962–4974

    2021

  17. [17]

    Zihan Qiu, Zekun Wang, Bo Zheng, Zeyu Huang, Kaiyue Wen, Songlin Yang, Rui Men, Le Yu, Fei Huang, Suozhi Huang, et al . 2025. Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Free.arXiv preprint arXiv:2505.06708(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  18. [18]

    Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al. 2023. A survey of large language models.arXiv preprint arXiv:2303.182231, 2 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  19. [19]

    Yupeng Chang, Xu Wang, Jindong Wang, Yuan Wu, Linyi Yang, Kaijie Zhu, Hao Chen, Xiaoyuan Yi, Cunxiang Wang, Yidong Wang, et al . 2024. A survey on evaluation of large language models.ACM transactions on intelligent systems and technology15, 3 (2024), 1–45

    2024

  20. [20]

    Jingyi Zhang, Jiaxing Huang, Sheng Jin, and Shijian Lu. 2024. Vision-language models for vision tasks: A survey.IEEE transactions on pattern analysis and machine intelligence46, 8 (2024), 5625–5644

    2024

  21. [21]

    Akash Ghosh, Arkadeep Acharya, Sriparna Saha, Vinija Jain, and Aman Chadha

  22. [22]

    Exploring the frontier of vision-language models: A survey of current methodologies and future directions.arXiv preprint arXiv:2404.07214(2024)

    work page arXiv 2024

  23. [23]

    Yujie Qian, Jiang Guo, Zhengkai Tu, Connor W Coley, and Regina Barzilay. 2023. RxnScribe: a sequence generation model for reaction diagram parsing.Journal of chemical information and modeling63, 13 (2023), 4030–4041

    2023

  24. [24]

    Yufan Chen, Ching Ting Leung, Jianwei Sun, Yong Huang, Linyan Li, Hao Chen, and Hanyu Gao. 2025. Towards Large-scale Chemical Reaction Image Parsing via a Multimodal Large Language Model.arXiv preprint arXiv:2503.08156(2025)

    work page arXiv 2025

  25. [25]

    Ali Dorri, Salil S Kanhere, and Raja Jurdak. 2018. Multi-agent systems: A survey. Ieee Access6 (2018), 28573–28593

    2018

  26. [26]

    Khanh-Tung Tran, Dung Dao, Minh-Duong Nguyen, Quoc-Viet Pham, Barry O’Sullivan, and Hoang D Nguyen. 2025. Multi-agent collaboration mechanisms: A survey of llms.arXiv preprint arXiv:2501.06322(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  27. [27]

    GPT-4o System Card

    OpenAI Aaron Hurst, Adam Lerer, Adam P. Goucher, Adam Perelman, Aditya Ramesh, and Aidan Clark et al. 2024. GPT-4o System Card.ArXivabs/2410.21276 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  28. [28]

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, and et al. 2025. Gemini: A Family of Highly Capable Multimodal Models. arXiv:2312.11805 [cs.CL] https://arxiv.org/abs/2312.11805

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  29. [29]

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, and Lianghao Deng et al. 2025. Qwen3-VL Technical Report.arXiv preprint arXiv:2511.21631(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  30. [30]

    Junxian Li, Di Zhang, Xunzhi Wang, Zeying Hao, Jingdi Lei, Qian Tan, Cai Zhou, Wei Liu, Yaotian Yang, Xinrui Xiong, et al. 2025. Chemvlm: Exploring the power of multimodal large language models in chemistry area. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 415–423

    2025

  31. [31]

    Lei Bai, Zhongrui Cai, Yuhang Cao, Maosong Cao, Weihan Cao, Chiyu Chen, Haojiong Chen, Kai Chen, Pengcheng Chen, and Ying Chen et al. 2025. Intern-S1: A Scientific Multimodal Foundation Model. arXiv:2508.15763 [cs.LG] https: //arxiv.org/abs/2508.15763

    work page arXiv 2025

  32. [32]

    Yashar Talebirad and Amirhossein Nadiri. 2023. Multi-agent collaboration: Harnessing the power of intelligent llm agents.arXiv preprint arXiv:2306.03314 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  33. [33]

    Yuheng Cheng, Ceyao Zhang, Zhengwen Zhang, Xiangrui Meng, Sirui Hong, Wenhao Li, Zihao Wang, Zekai Wang, Feng Yin, Junhua Zhao, et al. 2024. Ex- ploring large language model based intelligent agents: Definitions, methods, and prospects.arXiv preprint arXiv:2401.03428(2024)

    work page arXiv 2024

  34. [34]

    Mingcheng Li, Xiaolu Hou, Ziyang Liu, Dingkang Yang, Ziyun Qian, Jiawei Chen, Jinjie Wei, Yue Jiang, Qingyao Xu, and Lihua Zhang. 2025. MCCD: Multi-Agent Collaboration-based Compositional Diffusion for Complex Text-to-Image Gener- ation. InProceedings of the Computer Vision and Pattern Recognition Conference. 13263–13272

    2025

  35. [35]

    Binpeng Shi, Yu Luo, Jingya Wang, Yongxin Zhao, Shenglin Zhang, Bowen Hao, Chenyu Zhao, Yongqian Sun, Zhi Zhang, Ronghua Sun, et al. 2025. FlowXpert: Tang et al. Expertizing Troubleshooting Workflow Orchestration with Knowledge Base and Multi-Agent Coevolution. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2. 4839–4850

    2025

  36. [36]

    Zhangtao Cheng, Yuhao Ma, Jian Lang, Kunpeng Zhang, Ting Zhong, Yong Wang, and Fan Zhou. 2025. Generative Thinking, Corrective Action: User- Friendly Composed Image Retrieval via Automatic Multi-Agent Collaboration. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2. 334–344

    2025

  37. [37]

    Damian M Wilary and Jacqueline M Cole. 2023. ReactionDataExtractor 2.0: a deep learning approach for data extraction from chemical reaction schemes. Journal of Chemical Information and Modeling63, 19 (2023), 6053–6067

    2023

  38. [38]

    Cheng Cui, Ting Sun, Manhui Lin, Tingquan Gao, Yubo Zhang, Jiaxuan Liu, Xueqing Wang, Zelun Zhang, Changda Zhou, Hongen Liu, Yue Zhang, Wenyu Lv, Kui Huang, Yichao Zhang, Jing Zhang, Jun Zhang, Yi Liu, Dianhai Yu, and Yanjun Ma. 2025. PaddleOCR 3.0 Technical Report. arXiv:2507.05595 [cs.CV] https://arxiv.org/abs/2507.05595

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  39. [39]

    Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and Philip S Yu. 2020. A comprehensive survey on graph neural networks.IEEE transactions on neural networks and learning systems32, 1 (2020), 4–24

    2020

  40. [40]

    Si Zhang, Hanghang Tong, Jiejun Xu, and Ross Maciejewski. 2019. Graph convo- lutional networks: a comprehensive review.Computational Social Networks6, 1 (2019), 1–23

    2019

  41. [41]

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, Humen Zhong, Yuanzhi Zhu, and Mingkun Yang et al. 2025. Qwen2.5-VL Technical Report. arXiv:2502.13923 [cs.CV] https://arxiv.org/abs/2502.13923

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  42. [42]

    Jinguo Zhu, Weiyun Wang, Zhe Chen, Zhaoyang Liu, Shenglong Ye, Lixin Gu, Hao Tian, and Yuchen Duan et al. 2025. InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models. arXiv:2504.10479 [cs.CV] https://arxiv.org/abs/2504.10479

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  43. [43]

    LLM Xiaomi, Bingquan Xia, Bowen Shen, Dawei Zhu, Di Zhang, Gang Wang, Hailin Zhang, Huaqiu Liu, Jiebao Xiao, Jinhao Dong, et al. 2025. MiMo: Unlocking the Reasoning Potential of Language Model–From Pretraining to Posttraining. arXiv preprint arXiv:2505.07608(2025)

    work page arXiv 2025

  44. [44]

    molecule_expert

    Mark Martori and Daniel Probst. 2022.Machine Learning approach for chemical reactions digitalisation.https://github.com/markmartorilopez/ A Details of the Ablation Study Table 3 presents a detailed ablation study of MACReD across four common layouts: Single Line, Multiple Line, Tree, and Graph. Eval- uation metrics include Precision, Recall, and F1-score ...

    2022