Pith. sign in

REVIEW 5 major objections 5 minor 4 cited by

CAD-Coder: An Open-Source Vision-Language Model for Computer-Aided Design Code Generation

T0 review · 5 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper claims that fine-tuning a LLaVA-style vision-language model on 163k image-CadQuery pairs produces a model that writes valid CadQuery code from images with 100% syntax success and the best 3D solid similarity.

desk verdict Useful open dataset and an honest fine-tuning recipe, but the headline numbers rest on 100 test images with no variance estimate and the abstract overclaims generalization. read the letter →

arxiv 2505.14646 v1 pith:QS6GSQYL submitted 2025-05-20 cs.CV cs.AI

Anna C. Doris , Md Ferdous Alam , Amin Heyrani Nobari , Faez Ahmed This is my paper
classification cs.CVcs.AI
keywords CADcodegenerationvision-languagemodelQueryimage-conditionedfine-tuningGenCAD-Codedataset3Dsolidsimilarityeditable
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 a general-purpose vision-language model, fine-tuned on a large dataset of CAD images paired with CadQuery scripts, can replace the manual step of turning a picture into an editable 3D CAD model. It introduces CAD-Coder, built on a LLaVA-1.5-style architecture, and GenCAD-Code, a dataset of 163,671 image-CadQuery pairs converted from the GenCAD dataset. On a 100-image subset of the test set, the model generates syntactically valid CadQuery code in 100% of cases and reaches the highest intersection-over-union similarity to ground-truth solids, ahead of closed and open VLMs including much larger ones. The paper also reports partial generalization to photographs of real 3D-printed objects and, for a lower-learning-rate variant, to a fillet operation not present in the fine-tuning data. A sympathetic reader would take the central claim to be that domain-specific fine-tuning of a foundation VLM is a viable route to accurate, editable, image-conditioned CAD generation.

What carries the argument

The load-bearing machinery is the pairing of a LLaVA-1.5-style vision-language architecture with a two-stage training recipe: first align a CLIP-ViT-L-336px vision encoder to a Vicuna-13B language model through a two-layer MLP on 595k image-caption pairs, then fine-tune the MLP and LLM end-to-end on 147k image-CadQuery pairs with the vision encoder frozen. The data side is GenCAD-Code, obtained by converting the command-vector CAD programs of GenCAD/DeepCAD into CadQuery scripts, which gives a complete, executable, editable code representation rather than a restricted DSL. Evaluation uses a normalized IOUbest metric in which solids are centered and scale-normalized by the radius of gyration, then aligned by principal axes with the best of four valid rotations, justified by a Procrustes-style optimal alignment argument.

What would settle it

Re-run the evaluation on all 7,355 test-set images, or on a larger stratified sample, and check whether VSR remains 100% and IOUbest stays near 0.675; additionally, verify that each sampled test CAD program and its five rendered views are absent from the training split. If the 100-example subset contains only simple solids or shares programs with training, the reported margins would shrink.

Watch

Extended reading notes

Core claim

CAD-Coder's central claim is that an open-source VLM fine-tuned end-to-end on the GenCAD-Code dataset can generate executable, editable CadQuery Python code directly from an input image, and that this method outperforms state-of-the-art image-conditioned code-generating baselines. In the paper's evaluation on 100 randomly sampled test examples, CAD-Coder attains a 100% valid syntax rate (VSR) and an IOUbest score of 0.675, compared with 0.524 for GPT-4.5 and 0.352 for Qwen2.5-VL-72B, the next-best baselines on each metric. The paper further claims that the model shows signs of generalization beyond its fine-tuning distribution: it produces reasonable CAD code from photographs of real objects, and one variant trained with a halved learning rate can apply a fillet operation that never appeared in the fine-tuning data. The authors position this as evidence that fine-tuned foundation models, rather than bespoke CAD-only networks or DSL-based systems, are a promising path for automating editable CAD modeling.

Load-bearing premise

The quantitative claims rest on the assumption that the 100 randomly sampled test examples fairly represent the full 7,355-example test set and that the train/test split separates CAD programs, not just rendered images, so no test solid appeared in training through a different view.

Editorial extensions

If this is right

  • If CAD-Coder's results hold, engineers could hand a photo or rendering to a model and get a starting CadQuery script that compiles, cutting the manual CAD modeling step down to editing generated code.
  • A 100% VSR means generated scripts can be run automatically, enabling pipelines that batch-generate CAD candidates from image collections.
  • The open release of GenCAD-Code (163k image-code pairs) gives other researchers a large English-code dataset for further fine-tuning beyond this specific model.
  • The generalization experiments suggest foundation-model pretraining can partially compensate for a narrow fine-tuning distribution, and the fillet result on the low-LR variant shows hyperparameter choice controls how much pretrained CAD knowledge survives.
  • The IOUbest metric offers a more principled replacement for chamfer-distance comparisons in editable-CAD evaluation.

Reading between the lines

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

  • The headline numbers rest on 100 test examples, so a natural extension is to run the same evaluation on the full 7,355-example test set; if the 100-sample subset is biased toward simpler solids, VSR and IOU would likely drop.
  • Because the paper does not specify whether the train/test split is at the CAD-program level, the same solid might appear in training through one of its five rendered views; checking program-level exclusivity would settle whether part of the gain is memorization.
  • The real-photo generalization was tested on only five objects, so a larger real-image benchmark paired with ground-truth CAD code would be needed to quantify the domain gap.
  • The fillet generalization is prompt-dependent, so more robust methods for preserving pretrained knowledge during fine-tuning could expand the set of usable CAD operations.
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

5 major / 5 minor

Summary. The paper presents CAD-Coder, a LLaVA-1.5-style vision-language model fine-tuned on GenCAD-Code, a new dataset of 163,671 rendered-CAD-image and CadQuery-code pairs converted from the GenCAD dataset. The model is evaluated on a 100-example subset of the GenCAD-Code test set against six open- and closed-source VLMs, and the authors report a 100% valid syntax rate (VSR) and an IOUbest of 0.675, outperforming GPT-4.5 and Qwen2.5-VL-72B. Additional experiments examine generalization to photographs of real 3D-printed objects and to the fillet operation, which was not present in the fine-tuning data. The paper also releases the model weights, evaluation prompts, and the dataset.

Significance. If the reported results are robust, the work is valuable: it provides a relatively large open dataset of image-CadQuery pairs, an open fine-tuned VLM for editable CAD code generation, and a formal alignment metric for solid-shape comparison. The comparison with strong external VLMs addresses a timely question, and the release of weights and exact prompts supports reproducibility. The Appendix's derivation of the optimal rigid-body alignment is a genuine technical contribution. However, the headline quantitative claims rest on a single 100-example evaluation subset, on an unvalidated conversion script that defines ground truth, and on dataset construction choices whose leakage risk is not quantified; the significance is therefore conditional on these issues being resolved.

major comments (5)
  1. [§3.3, Table 2] The central state-of-the-art claim rests on a single 100-example subset of the 7,355-example test set. The paper reports no sampling seed, no sample identifiers, and no confidence intervals. Given the right-skewed complexity distribution in Figure 2, the 100% VSR and the 0.151 IOUbest margin over GPT-4.5 may be sampling artifacts. Please report bootstrap or exact binomial confidence intervals (for VSR with 100/100 successes, the 95% Clopper-Pearson interval is approximately [96.4%, 100%]), release the exact sample list, and, if feasible, report results on the full test set.
  2. [§3.1, §3.3] The paper does not state whether the train/test/validation split is at the CAD-program level or at the image level, and it does not describe any deduplication of the underlying ABC/DeepCAD/GenCAD data, even though Section 2.3 notes that ABC includes duplicates. Because GenCAD supplies five rendered images per CAD program, an image-level split could place different views of the same solid in both training and test, directly inflating VSR and IOUbest. Please specify the split granularity, implement and describe solid-level deduplication, or provide a quantitative analysis of train/test similarity to rule out leakage.
  3. [§3.1] The GenCAD-Code conversion script defines both the training targets and the evaluation ground truth, but the paper reports no validation that the converted CadQuery scripts reproduce the original GenCAD/ABC solids. A systematic conversion error (for example, in arc or line parameter ordering) would be invisible to the reported metrics because the same converter generates both the prediction targets and the reference solids. Please validate the conversion on a sample by comparing rendered images or STL meshes of converted solids against the original GenCAD outputs and report the failure rate.
  4. [§3.3.2, Table 2] IOUbest is computed only on scripts that compile successfully, while VSR differs across models (100% for CAD-Coder versus 82-94% for the baselines). This makes the IOU comparison conditional on successful execution and can favor a model with a low VSR if its few successful outputs happen to be accurate. Please also report an unconditional metric, such as treating failed scripts as IOU=0, and discuss how the model ranking changes under that metric.
  5. [§4.3.2, Abstract] The abstract's claim that CAD-Coder succeeds at 'executing CAD operations unseen during fine-tuning' is contradicted by the body of the paper. Section 4.3.2 states that the main CAD-Coder model cannot add fillets, and that only CAD-Coder-Qwen2.5-14B-LowLR succeeds, and only when given a prescriptive prompt; more abstract prompts do not elicit correct behavior. Please soften the abstract, the introduction, and contribution 3 so that they match the actual results and specify the prompt conditions under which the capability appears.
minor comments (5)
  1. [§4.2] The statement that Qwen2.5-VL-72B is the next-best open-source model is incorrect: Ovis2-34B has IOUbest 0.408 and InternVL2_5-78B-MPO has 0.379, both higher than Qwen2.5-VL's 0.352. The relative improvement figure is also arithmetically incorrect: 0.675 versus 0.352 is about a 92% relative improvement, not 60%.
  2. [§4.3.1] The real-image generalization evidence is qualitative and limited to five 3D-printed objects, with no quantitative metric or comparison to baselines. Please either add quantitative scores for these five examples or explicitly describe the result as anecdotal.
  3. [§3.3.1] The baseline selection criterion ('top-three models on leaderboards as of March 17, 2025') is not reproducible because leaderboard rankings change over time. Please report exact model version identifiers or checkpoints for each baseline.
  4. [§4.3.2] The fillet experiment relies on a two-query setup with a prescriptive prompt, but the exact prompt text is only summarized in Figure 5. Please include the verbatim prompts in the appendix or repository.
  5. [§4.1] Training details do not mention the learning-rate schedule, warmup, or random seed for Stage 2 fine-tuning. This information is needed to reproduce the LowLR variant and the reported results.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the core evaluation is an external benchmark against other VLMs, and the same-group GenCAD citation is transparent data provenance rather than a load-bearing derivation.

full rationale

CAD-Coder's central claim is empirical: a LLaVA-style VLM is fine-tuned on a converted CadQuery dataset and measured against external closed- and open-source VLMs using VSR and IOUbest. The evaluation has no fitted constant that is later renamed a prediction, and no quantity in Table 2 is defined in terms of the model's own outputs. VSR is computed by executing the generated CadQuery scripts, and IOUbest is computed by comparing the resulting solids with ground-truth solids under a Procrustes-style alignment derived in Appendix A. The one same-group citation, GenCAD [9], is explicit data provenance: GenCAD-Code is a conversion of GenCAD's CAD programs into CadQuery code, and the paper also states that GenCAD derives from the DeepCAD/ABC datasets. That citation is not used to justify the claimed superiority over GPT-4.5, Qwen2.5-VL-72B, and the other baselines, which are external models evaluated on the same 100-example subset. Concerns about the small test subset, missing confidence intervals, and possible duplication or split-granularity leakage are validity risks, not circular reasoning. The fillet-generalization claim is explicitly acknowledged by the authors to be prompt-dependent and non-robust, so it is not presented as a forced or hidden success. No specific equation or step can be quoted as reducing to its own input, so the appropriate finding is no significant circularity.

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

The paper's quantitative claims rest on the fidelity of the conversion script, on the IOUbest normalization discarding absolute scale, and on transfer from CLIP pretraining; none of these are independently verified beyond the paper's own experiments.

assumptions (4)
  • domain assumption The direct GenCAD-to-CadQuery conversion script is a semantically exact translation of each CAD program.
    Section 3.1 states the conversion is done by a script but does not report validation that every converted CadQuery script renders the same solid as the original GenCAD program; both training labels and ground-truth test labels inherit any conversion errors.
  • domain assumption A single grayscale isometric rendered image carries enough information to determine the target solid up to scale, so scale normalization is appropriate for evaluation.
    Section 3.3.2 and Appendix A justify the IOUbest metric by normalizing translation and scale and aligning principal axes; if absolute dimensions are considered part of the CAD generation task, this metric understates a real limitation.
  • domain assumption Pre-trained CLIP features learned on natural images transfer to rendered CAD images and real photographs in the fine-tuned model.
    The real-image generalization experiments in Section 4.3.1 rely on this transfer, but only five real-image examples are shown and no quantitative accuracy is reported for them.
  • standard math Standard rigid-body alignment results, including change of variables, inertia tensors, and orthogonal Procrustes, are correct and applicable to solid shape comparison.
    Appendix A, Lemmas 1 and 2 use textbook integration and SVD arguments that appear mathematically sound.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CAD-Coder: An Open-Source Vision-Language Model for Computer-Aided Design Code Generation." pith.science (2026). https://pith.science/paper/QS6GSQYL

@misc{pith2026250514646,
  author       = {Pith},
  title        = {Pith review of: CAD-Coder: An Open-Source Vision-Language Model for Computer-Aided Design Code Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QS6GSQYL}},
  note         = {Machine review of arXiv:2505.14646}
}
read the original abstract

Efficient creation of accurate and editable 3D CAD models is critical in engineering design, significantly impacting cost and time-to-market in product innovation. Current manual workflows remain highly time-consuming and demand extensive user expertise. While recent developments in AI-driven CAD generation show promise, existing models are limited by incomplete representations of CAD operations, inability to generalize to real-world images, and low output accuracy. This paper introduces CAD-Coder, an open-source Vision-Language Model (VLM) explicitly fine-tuned to generate editable CAD code (CadQuery Python) directly from visual input. Leveraging a novel dataset that we created--GenCAD-Code, consisting of over 163k CAD-model image and code pairs--CAD-Coder outperforms state-of-the-art VLM baselines such as GPT-4.5 and Qwen2.5-VL-72B, achieving a 100% valid syntax rate and the highest accuracy in 3D solid similarity. Notably, our VLM demonstrates some signs of generalizability, successfully generating CAD code from real-world images and executing CAD operations unseen during fine-tuning. The performance and adaptability of CAD-Coder highlights the potential of VLMs fine-tuned on code to streamline CAD workflows for engineers and designers. CAD-Coder is publicly available at: https://github.com/anniedoris/CAD-Coder.

Figures

Figures reproduced from arXiv: 2505.14646 by the authors.

Figure 1
Figure 1. FIGURE 1: OVERVIEW OF CAD-CODER. THE VLM ACCEPTS AN IMAGE AS INPUT AND OUTPUTS CADQUERY CODE, WHICH CAN BE RUN [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2: DISTRIBUTION OF TOKEN COUNTS FOR THE CAD [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. FIGURE 3: TWO EXAMPLES COMPARING CAD-CODER’S GENERATED SOLIDS WITH BASELINE GENERATED SOLIDS. THE IOU [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: FIGURE 4: WE TEST CAD-CODER’S GENERALIZABILITY TO REAL-IMAGE-CONDITIONED CAD GENERATION, A TASK NOT INCLUDED IN [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: FIGURE 5: EXAMPLES OF CAD-CODER VARIANTS ATTEMPTING TO ADD FILLETS TO CAD SOLIDS. THE FIGURE COMPARES THE PERFOR [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. CADENA: Stepwise CAD Reverse Engineering

    cs.CV 2026-08 conditional novelty 7.0 of 10

    A vision-language model that grows a CAD program step by step, rendering target and current build together after every operation, outperforms single-pass baselines on mechanical part reconstruction.

  2. Nova3D: Code-Native Generation of Programmable 3D Assets

    cs.GR 2026-07 conditional novelty 6.0 of 10

    Nova3D generates 3D assets as executable Blender source, yielding named parts, assembly hierarchies, measurable constraints, and native joints that mesh-native generators do not expose.

  3. STEP-LLM: Generating CAD STEP Models from Natural Language with Large Language Models

    cs.AI 2026-01 conditional novelty 6.0 of 10

    STEP-LLM fine-tunes LLMs to emit complete STEP CAD files from natural-language captions, reporting lower median Chamfer distance than Text2CAD.

  4. Toward Knowledge-Guided AI for Inverse Design in Manufacturing: A Perspective on Domain, Physics, and Human-AI Synergy

    cs.AI 2025-05 unverdicted novelty 3.0 of 10

    A perspective arguing that inverse design in manufacturing improves when expert-guided problem definition, physics-informed ML, and LLM interfaces are combined.

Reference graph

Works this paper leans on

37 extracted references · 24 canonical work pages · cited by 4 Pith papers

  1. [1]

    Generating CAD Code with Vision-Language Models for 3D Designs

    Alrashedy,Kamel,Tambwekar,Pradyumna,Zaidi,Zulfiqar, Langwasser, Megan, Xu, Wei and Gombolay, Matthew. 11 “Generating CAD Code with Vision-Language Models for 3D Designs.”arXiv preprint arXiv:2410.05340(2024)

    arXiv 2024

  2. [2]

    3d shape gener- ation and completion through point-voxel diffusion

    Zhou, Linqi, Du, Yilun and Wu, Jiajun. “3d shape gener- ation and completion through point-voxel diffusion.”Pro- ceedingsoftheIEEE/CVFinternationalconferenceoncom- puter vision: pp. 5826–5835. 2021

    work page 2021

  3. [3]

    Diffusion probabilistic mod- els for 3d point cloud generation

    Luo, Shitong and Hu, Wei. “Diffusion probabilistic mod- els for 3d point cloud generation.” Proceedings of the IEEE/CVF conference on computer vision and pattern recognition: pp. 2837–2845. 2021

    work page 2021

  4. [4]

    Shap-e: Generat- ing conditional 3d implicit functions

    Jun, Heewoo and Nichol, Alex. “Shap-e: Generat- ing conditional 3d implicit functions.” arXiv preprint arXiv:2305.02463(2023)

    arXiv 2023

  5. [5]

    Towards implicit text-guided 3d shape generation

    Liu, Zhengzhe, Wang, Yi, Qi, Xiaojuan and Fu, Chi-Wing. “Towards implicit text-guided 3d shape generation.”Pro- ceedings ofthe IEEE/CVF Conference onComputer Vision and Pattern Recognition: pp. 17896–17906. 2022

    work page 2022

  6. [6]

    Text2mesh: Text-driven neural styl- ization for meshes

    Michel, Oscar, Bar-On, Roi, Liu, Richard, Benaim, Sagie and Hanocka, Rana. “Text2mesh: Text-driven neural styl- ization for meshes.” Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition: pp. 13492–13502. 2022

    work page 2022

  7. [7]

    Skexgen: Autoregressive generation of cad con- struction sequences with disentangled codebooks

    Xu,Xiang,Willis,KarlDD,Lambourne,JosephG,Cheng, Chin-Yi, Jayaraman, Pradeep Kumar and Furukawa, Ya- sutaka. “Skexgen: Autoregressive generation of cad con- struction sequences with disentangled codebooks.”arXiv preprint arXiv:2207.04632(2022)

    arXiv 2022

  8. [8]

    Deep- cad: A deep generative network for computer-aided design models

    Wu, Rundi, Xiao, Chang and Zheng, Changxi. “Deep- cad: A deep generative network for computer-aided design models.” ProceedingsoftheIEEE/CVFInternationalCon- ference on Computer Vision: pp. 6772–6782. 2021

    work page 2021

Show all 37 references
  1. [9]

    Gencad: Image-conditioned computer-aided design generation with transformer-based contrastive representation and diffusion priors

    Alam, Md Ferdous and Ahmed, Faez. “Gencad: Image-conditioned computer-aided design generation with transformer-based contrastive representation and diffusion priors.” arXiv preprint arXiv:2409.16294(2024)

    2024 arXiv

  2. [10]

    Visual Instruction Tuning

    Liu, Haotian, Li, Chunyuan, Wu, Qingyang and Lee, Yong Jae. “Visual Instruction Tuning.” (2023)

    2023

  3. [11]

    LLM4CAD:Multi-ModalLargeLanguageModelsfor3D Computer-Aided Design Generation

    Li, Xingang, Sun, Yuewan and Sha, Zhenghui. “LLM4CAD:Multi-ModalLargeLanguageModelsfor3D Computer-Aided Design Generation.” International De- sign Engineering Technical Conferences and Computers and Information in Engineering Conference, Vol. 88407: p. V006T06A015. 2024. Ameri...

    2024

  4. [12]

    BlenderLLM: Training Large Language Models forComputer-AidedDesignwithSelf-improvement

    Du,Yuhao,Chen,Shunian,Zan,Wenbo,Li,Peizhao,Wang, Mingxuan, Song, Dingjie, Li, Bo, Hu, Yan and Wang, Benyou. “BlenderLLM: Training Large Language Models forComputer-AidedDesignwithSelf-improvement.” arXiv preprint arXiv:2412.14203(2024)

    2024 arXiv

  5. [13]

    Cadvlm: Bridging language and vision in the gen- eration of parametric cad sketches

    Wu, Sifan, Khasahmadi, Amir Hosein, Katz, Mor, Jayara- man, Pradeep Kumar, Pu, Yewen, Willis, Karl and Liu, Bang. “Cadvlm: Bridging language and vision in the gen- eration of parametric cad sketches.”European Conference on Computer Vision: pp. 368–384. 2024. Springer

    2024

  6. [14]

    OpenECAD: Anefficientvisuallanguagemodelforeditable3D-CADde- sign

    Yuan,Zhe,Shi,JianqiandHuang,Yanhong. “OpenECAD: Anefficientvisuallanguagemodelforeditable3D-CADde- sign.” Computers & GraphicsVol. 124 (2024): p. 104048

    2024

  7. [15]

    Csgnet: Neural shape parser for constructive solid geometry

    Sharma, Gopal, Goyal, Rishabh, Liu, Difan, Kalogerakis, Evangelos and Maji, Subhransu. “Csgnet: Neural shape parser for constructive solid geometry.”Proceedings of the IEEE Conference on Computer Vision and Pattern Recog- nition: pp. 5515–5523. 2018

    2018

  8. [16]

    Text2cad: Generating sequential cad designs from beginner-to-expert level text prompts

    Khan, Mohammad Sadil, Sinha, Sankalp, Uddin, Talha, Stricker, Didier, Ali, Sk Aziz and Afzal, Muhammad Ze- shan. “Text2cad: Generating sequential cad designs from beginner-to-expert level text prompts.”Advances in Neural Information Processing SystemsVol. 37 (2024): pp. 7552– 7579

    2024

  9. [17]

    Generating CAD Code with Vision-Language Modelsfor3DDesigns

    Alrashedy, Kamel, Tambwekar, Pradyumna, Zaidi, Zul- fiqar Haider, Langwasser, Megan, Xu, Wei and Gombolay, Matthew. “Generating CAD Code with Vision-Language Modelsfor3DDesigns.” TheThirteenthInternationalCon- ference on Learning Representations

  10. [18]

    FromIdeatoCAD:ALanguageModel-DrivenMulti- Agent System for Collaborative Design

    Ocker, Felix, Menzel, Stefan, Sadik, Ahmed and Rios, Thi- ago. “FromIdeatoCAD:ALanguageModel-DrivenMulti- Agent System for Collaborative Design.”arXiv preprint arXiv:2503.04417(2025)

    2025 arXiv

  11. [19]

    CAD-Assistant: Tool-Augmented VLLMs as Generic CAD Task Solvers?

    Mallis, Dimitrios, Karadeniz, Ahmet Serdar, Cavada, Sebastian, Rukhovich, Danila, Foteinopoulou, Niki, Cherenkova, Kseniya, Kacem, Anis and Aouada, Djamila. “CAD-Assistant: Tool-Augmented VLLMs as Generic CAD Task Solvers?” arXiv preprint arXiv:2412.13810 (2024)

    2024 arXiv

  12. [20]

    LargeLan- guage Models for Computer-Aided Design (LLM4CAD) Fine-Tuned: Dataset and Experiments

    Sun,Yuewan,Li,XingangandSha,Zhenghui. “LargeLan- guage Models for Computer-Aided Design (LLM4CAD) Fine-Tuned: Dataset and Experiments.” Journal of Me- chanical Design(2025): pp. 1–19

    2025

  13. [21]

    Attention is all you need

    Vaswani, Ashish, Shazeer, Noam, Parmar, Niki, Uszkoreit, Jakob, Jones, Llion, Gomez, Aidan N, Kaiser, Łukasz and Polosukhin, Illia. “Attention is all you need.”Advances in neural information processing systemsVol. 30 (2017)

    2017

  14. [22]

    Learningtransferablevisualmodelsfromnaturallanguage supervision

    Radford, Alec, Kim, Jong Wook, Hallacy, Chris, Ramesh, Aditya, Goh, Gabriel, Agarwal, Sandhini, Sastry, Girish, Askell, Amanda, Mishkin, Pamela, Clark, Jack et al. “Learningtransferablevisualmodelsfromnaturallanguage supervision.” International conference on machine learn- ing...

    2021

  15. [23]

    Engineering sketch generation for computer-aided design

    Willis, Karl DD, Jayaraman, Pradeep Kumar, Lambourne, Joseph G, Chu, Hang and Pu, Yewen. “Engineering sketch generation for computer-aided design.” Proceedings of the IEEE/CVF conference on computer vision and pattern recognition: pp. 2105–2114. 2021

    2021

  16. [24]

    Abc: Abigcadmodeldatasetforgeometricdeeplearning

    Koch, Sebastian, Matveev, Albert, Jiang, Zhongshi, Williams, Francis, Artemov, Alexey, Burnaev, Evgeny, Alexa, Marc, Zorin, Denis and Panozzo, Daniele. “Abc: Abigcadmodeldatasetforgeometricdeeplearning.” Pro- ceedings of the IEEE/CVF conference on computer vision and pattern r...

    2019

  17. [25]

    Brep- gen: A b-rep generative diffusion model with structured latent geometry

    Xu,Xiang,Lambourne,Joseph,Jayaraman,Pradeep,Wang, Zhengqing, Willis, Karl and Furukawa, Yasutaka. “Brep- gen: A b-rep generative diffusion model with structured latent geometry.” ACM Transactions on Graphics (TOG) Vol. 43 No. 4 (2024): pp. 1–14. 12

    2024

  18. [26]

    Solidgen: An autoregressive model for direct b-rep synthesis

    Jayaraman, Pradeep Kumar, Lambourne, Joseph G, De- sai, Nishkrit, Willis, Karl DD, Sanghi, Aditya and Morris, Nigel JW. “Solidgen: An autoregressive model for direct b-rep synthesis.”arXiv preprint arXiv:2203.13944(2022)

    2022 arXiv

  19. [27]

    Visual instruction tuning

    Liu, Haotian, Li, Chunyuan, Wu, Qingyang and Lee, Yong Jae. “Visual instruction tuning.”Advances in neural information processing systemsVol. 36 (2024)

    2024

  20. [28]

    Improved baselines with visual instruction tuning

    Liu,Haotian,Li,Chunyuan,Li,YuhengandLee,YongJae. “Improved baselines with visual instruction tuning.”arXiv preprint arXiv:2310.03744(2023)

    2023 arXiv

  21. [29]

    Mmmu: A massive multi-discipline multimodal understanding and reasoning benchmark for expert agi

    Yue, Xiang, Ni, Yuansheng, Zhang, Kai, Zheng, Tianyu, Liu, Ruoqi, Zhang, Ge, Stevens, Samuel, Jiang, Dongfu, Ren, Weiming, Sun, Yuxuan et al. “Mmmu: A massive multi-discipline multimodal understanding and reasoning benchmark for expert agi.”Proceedings of the IEEE/CVF Conferen...

    2024

  22. [30]

    Vlmevalkit: An open- source toolkit for evaluating large multi-modality models

    Duan, Haodong, Yang, Junming, Qiao, Yuxuan, Fang, Xinyu,Chen,Lin,Liu,Yuan,Dong,Xiaoyi,Zang,Yuhang, Zhang, Pan, Wang, Jiaqi et al. “Vlmevalkit: An open- source toolkit for evaluating large multi-modality models.” Proceedings of the 32nd ACM International Conference on Multimedi...

    2024

  23. [31]

    Deeplearningfortechnicaldocumentclassification

    Jiang,Shuo,Hu,Jie,Magee,ChristopherLandLuo,Jianxi. “Deeplearningfortechnicaldocumentclassification.” IEEE Transactions on Engineering Management(2022)

    2022

  24. [32]

    An empirical study of catastrophic forgetting in large language models during continual fine- tuning

    Luo, Yun, Yang, Zhen, Meng, Fandong, Li, Yafu, Zhou, Jie and Zhang, Yue. “An empirical study of catastrophic forgetting in large language models during continual fine- tuning.” arXiv preprint arXiv:2308.08747(2023)

    2023 arXiv

  25. [33]

    Fine-tuningcancrippleyourfoundationmodel; preserving features may be the solution

    Mukhoti, Jishnu, Gal, Yarin, Torr, Philip HS and Dokania, PuneetK. “Fine-tuningcancrippleyourfoundationmodel; preserving features may be the solution.”arXiv preprint arXiv:2308.13320(2023)

    2023 arXiv

  26. [34]

    Activation-Informed Merging of Large Language Models

    HeyraniNobari,Amin,Alimohammadi,Kaveh,Arjomand- Bigdeli, Ali, Srivastava, Akash, Ahmed, Faez and Azizan, Navid. “Activation-Informed Merging of Large Language Models.” arXiv e-prints(2025): pp. arXiv–2502

    2025

  27. [35]

    Pro- crustes Problems

    Gower, John C and Dijksterhuis, Garmt B. Pro- crustes Problems. Oxford University Press (2004). DOI 10.1093/acprof:oso/9780198510581.001.0001. URLhttps: //doi.org/10.1093/acprof:oso/9780198510581.001.0001. APPENDIX A. OPTIMAL SOLID MODEL ALIGNMENT In this section, we detail the...

    2004

  28. [36]

    Its Jacobian determinant is det(︁𝑠 R)︁ =𝑠3 det(R)=𝑠3, because det(R)= 1 given𝑅∈𝑆𝑂(3)

    Jacobian Determinant and Volume Scaling: Since 𝑓 is affine, its differential𝐷𝑓(x) is constant and equal to𝑠 R for all x∈Ω1. Its Jacobian determinant is det(︁𝑠 R)︁ =𝑠3 det(R)=𝑠3, because det(R)= 1 given𝑅∈𝑆𝑂(3)

  29. [37]

    Change of Variables and Relative Volume Preserva- tion: By the change-of-variable, for any integrable function𝑔 defined onΩ2, ∫ Ω2 𝑔(y)𝑑y= ∫ Ω1 𝑔(︁𝑓(x))︁ |︁|︁det(︁𝐷𝑓(x))︁|︁|︁ 𝑑x=𝑠3 ∫ Ω1 𝑔(︁𝑓(x))︁𝑑x. In particular, choosing𝑔(y)≡ 1 yields Vol(Ω2)=𝑠3 Vol(Ω1)⇒ 𝑠3= Vol(Ω2) Vol(Ω1),...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.