{"id":"843799d5-af41-4ba1-9958-8d0e27256d4e","arxiv_id":"2606.12994","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"DeepJEB++ expands a small seed set of jet engine brackets into 15,360 labeled 3D designs via 2D latent diffusion augmentation, VLM filtering, generative 3D lifting, and automated finite-element labeling.","lead":"The paper presents DeepJEB++, a pipeline that augments fewer than 400 seed jet engine bracket designs into 15,360 simulation-labeled 3D meshes by operating in 2D latent space then lifting to 3D. A smart generalist might read it to understand practical ways to scale training data for AI in mechanical engineering without massive manual simulation effort.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"VLM filter + 3D lift may admit geometries whose interface recognition produces invalid BCs or labels","rationale":"The reader's weakest assumption directly identifies the same pipeline step whose correctness is least evidenced by the abstract numbers. No stronger internal inconsistency appears in the given description; the concern is therefore the same one already flagged.","tokens_in":1794,"tokens_out":295,"duration_ms":14707,"concrete_test":"Sample 100 generated meshes; run the Stage-3 interface recognizer, then have two independent FEM experts manually re-assign loads/bolts and re-simulate; report the fraction where expert labels differ from the pipeline by >10 % in max stress or displacement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The 40× expansion claim (Stage 3) requires that the VLM filter (Stage 1) and domain-adapted generator (Stage 2) together ensure every retained mesh has correctly identifiable load/bolt interfaces whose assigned mass/stress/displacement labels are faithful. The abstract states that label fidelity is assessed against SimJEB ground truth, yet supplies no quantitative failure rates, confusion matrices for interface detection, or ablation showing how often the automated pipeline assigns incorrect boundary conditions on out-of-distribution shapes. If interface recognition error exceeds a few percent, the resulting 15 360-entry dataset contains a non-negligible fraction of physically invalid samples, undermining downstream use.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper introduces DeepJEB++, a foundation-model-driven framework that augments fewer than 400 seed 3D jet engine bracket designs into 15,360 simulation-labeled meshes (40x expansion) via 2D latent-space augmentation. Stage 1 fine-tunes a pretrained 2D latent diffusion model on multi-view renders, synthesizes novel views by interpolation, and applies a VLM quality filter for manufacturability. Stage 2 lifts validated images to 3D meshes using a domain-adapted generative foundation model. Stage 3 automates recognition of load/bolt interfaces and assignment of finite-element labels (mass, stress, displacement). Quality is assessed along manufacturability, label fidelity to SimJEB ground truth, and distributional consistency; the dataset is to be released publicly, with all stages using a single GPU.","tokens_in":1928,"tokens_out":447,"duration_ms":18661,"significance":"If the quality assessments hold with supporting quantitative evidence, the work would deliver a valuable public resource for data-driven engineering design, directly addressing the scarcity of large-scale 3D geometry-physics paired datasets. The single-GPU pipeline and reliance on external pretrained models demonstrate practical resource efficiency. Public release of the dataset would support reproducible research in engineering AI.","major_comments":[{"comment":"Abstract: The claim that quality was assessed along manufacturability, label fidelity, and distributional consistency supplies no quantitative metrics, error bars, failure rates, confusion matrices for interface detection, or ablation results on out-of-distribution shapes. This is load-bearing for the central 40x expansion claim, because even modest rates of invalid boundary-condition assignments would render a non-negligible fraction of the 15,360 samples physically unusable for downstream tasks.","section":"Abstract"},{"comment":"Abstract (Stage 3): The automated interface-recognition and labeling pipeline is described as operating without manual intervention and preserving label fidelity against SimJEB, yet no validation statistics or comparison to manual labeling are reported to bound the error rate on generated geometries.","section":"Abstract"}],"minor_comments":[],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for highlighting the need for explicit quantitative validation to support the dataset's claimed quality and usability. We address each major comment below and commit to revisions that strengthen the evidence without altering the core claims.","responses":[{"response":"We agree that the abstract's quality-assessment claim requires supporting quantitative evidence to substantiate the 40x expansion. The manuscript evaluates manufacturability via VLM filtering rates, label fidelity via comparison to SimJEB ground truth, and distributional consistency via statistical tests, but these are not quantified with the specific metrics, error bars, failure rates, confusion matrices, or OOD ablations noted. We will revise the abstract to report key numbers (e.g., pass rates, correlation coefficients, MMD scores) and expand the results section with error bars, interface-detection confusion matrices, failure rates, and OOD ablations. This directly addresses the concern about physically unusable samples.","revision_made":"yes","referee_comment":"[Abstract] Abstract: The claim that quality was assessed along manufacturability, label fidelity, and distributional consistency supplies no quantitative metrics, error bars, failure rates, confusion matrices for interface detection, or ablation results on out-of-distribution shapes. This is load-bearing for the central 40x expansion claim, because even modest rates of invalid boundary-condition assignments would render a non-negligible fraction of the 15,360 samples physically unusable for downstream tasks."},{"response":"We concur that bounding the error rate on generated geometries is essential for the automated Stage 3 pipeline. The current text reports fidelity preservation against SimJEB for seed designs and states the pipeline runs without manual intervention, but does not include validation statistics or manual-labeling comparisons for the augmented set. We will add these in revision: agreement rates and error bounds on a manually labeled subset of generated samples, plus any relevant confusion matrices, updating both the abstract and Stage 3 description.","revision_made":"yes","referee_comment":"[Abstract] Abstract (Stage 3): The automated interface-recognition and labeling pipeline is described as operating without manual intervention and preserving label fidelity against SimJEB, yet no validation statistics or comparison to manual labeling are reported to bound the error rate on generated geometries."}],"tokens_in":1527,"tokens_out":480,"duration_ms":18737,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The punchline is that DeepJEB++ describes a workable way to grow a small seed set of jet engine brackets into a much larger simulation-labeled collection by doing most of the creative work in 2D latent space before lifting to 3D.\n\nWhat is new is the specific sequence: fine-tune a 2D diffusion model on multi-view renders, interpolate in latent space, run a VLM filter for manufacturability, lift the survivors with a domain-adapted foundation model, then run automated interface recognition to attach load, bolt, mass, stress, and displacement labels. That combination is not in the earlier literature the abstract cites, and the decision to avoid direct 3D augmentation makes sense given the stated difficulties.\n\nThe paper does one thing clearly right: it releases the resulting 15,360-entry dataset publicly. Anyone working on ML for mechanical design now has a larger starting point than the original SimJEB set.\n\nThe soft spot is exactly where the stress-test note points. The abstract says quality was checked on manufacturability, label fidelity against SimJEB, and distributional consistency, yet it gives no failure rates, confusion matrices, or ablation numbers. If the automated interface step mislabels boundary conditions on more than a small fraction of the new shapes, the dataset carries systematic errors that would affect any model trained on it. That evidence is simply not shown.\n\nThis is for researchers who need more 3D geometry-plus-physics data for engineering tasks and are willing to do their own downstream validation. It is concrete enough and the data release is useful enough that it deserves a serious referee rather than a desk reject. Reviewers will want the missing quantitative checks, but the pipeline itself is worth that discussion.","headline":"The paper gives a concrete 2D-to-3D pipeline that turns a few hundred brackets into 15k labeled meshes, but the abstract supplies no numbers on how often the VLM filter or interface detector actually fails.","tokens_in":2429,"tokens_out":441,"would_cite":false,"duration_ms":17810,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"A pipeline augments fewer than 400 seed jet-engine brackets into 15,360 simulation-labeled 3D meshes by operating in 2D latent space before lifting to 3D.","keywords":["3D dataset augmentation","jet engine brackets","latent diffusion models","foundation models","finite element labeling","engineering design data","simulation labels","data scarcity"],"falsifier":"A random sample of 200 generated meshes is run through an independent finite-element solver; if more than 10 percent produce stress or displacement values that deviate by more than 15 percent from the SimJEB reference under identical boundary conditions, the claim of label fidelity fails.","tokens_in":2705,"feed_emoji":"🔧","tokens_out":806,"duration_ms":13920,"temperature":0.7,"pith_summary":"The paper shows how to overcome the scarcity of large 3D engineering datasets that include both geometry and physics labels. It does so by first expanding designs inside a data-rich 2D latent space with a fine-tuned diffusion model, then using a vision-language filter to keep only manufacturable outputs, lifting the survivors to 3D meshes with a domain-adapted generative model, and finally running an automated pipeline that detects load and bolt interfaces to attach finite-element labels. A sympathetic reader would care because data-driven design methods cannot scale without thousands of geometry-plus-performance pairs, and manual creation of such pairs is prohibitively expensive. The work claims this staged process achieves a 40-fold increase while staying within single-GPU budgets per stage and preserving the geometric and label properties needed for downstream use.","feed_headline":"2D latent augmentation yields 15,360 labeled 3D brackets from under 400 seeds","feed_subtitle":"Foundation-model stages expand the set fortyfold while automating load detection and finite-element labeling on a single GPU per stage.","key_machinery":"2D latent-space augmentation with a fine-tuned diffusion model followed by vision-language filtering and 3D lifting with a domain-adapted generative model.","core_discovery":"Starting from fewer than 400 seed designs, the three-stage process produces 15,360 simulation-labeled 3D brackets. Stage 1 fine-tunes a pretrained 2D latent diffusion model on multi-view renders and synthesizes new views by latent interpolation, then applies a vision-language-model filter to retain only manufacturable results. Stage 2 lifts the validated 2D images to 3D meshes using a domain-adapted generative foundation model. Stage 3 automatically identifies load and bolt interfaces on each mesh and computes finite-element labels for mass, stress, and displacement. Quality is assessed on manufacturability, fidelity to SimJEB ground truth, and distributional consistency.","pith_inferences":["The same 2D-to-3D lifting pattern could be tested on other load-bearing components such as brackets in different industries.","If the filter and lifting steps generalize, the approach reduces the data bottleneck that currently limits physics-informed machine learning in mechanical design.","Public release of the 15,360-mesh set creates a concrete benchmark that later papers can use to measure further augmentation gains."],"forward_implications":["A 40x increase in labeled 3D engineering data becomes feasible from a few hundred seeds using only one GPU per stage.","Automated interface detection removes the need for manual boundary-condition assignment before simulation.","The resulting dataset supports reproducible training of AI models for structural design tasks.","Quality checks along manufacturability, label fidelity, and distributional axes provide a repeatable evaluation template for other augmentation pipelines."],"fun_headline_variants":["Under 400 seeds expand to 15,360 labeled 3D brackets via 2D latent space","DeepJEB++ creates 15,360 simulation-labeled 3D brackets from small seed set","Three foundation model stages yield 15,360 labeled jet brackets","Latent augmentation in 2D lifts to 15,360 simulation-labeled 3D brackets"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The vision-language filter and the 3D generative model together keep manufacturability and simulation-label accuracy high enough that no systematic geometric or physics errors invalidate later engineering use.","fun_headline_variants_meta":{"raw":{"variants":["Under 400 seeds expand to 15,360 labeled 3D brackets via 2D latent space","DeepJEB++ creates 15,360 simulation-labeled 3D brackets from small seed set","Three foundation model stages yield 15,360 labeled jet brackets","Latent augmentation in 2D lifts to 15,360 simulation-labeled 3D brackets"]},"model":"grok-4.3","cost_usd":0.009537,"raw_usage":{"total_tokens":4328,"prompt_tokens":811,"num_sources_used":0,"completion_tokens":92,"cost_in_usd_ticks":95374500,"prompt_tokens_details":{"text_tokens":811,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":3425,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":811,"tokens_out":92,"duration_ms":18116,"temperature":1.0,"reasoning_tokens":3425,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-27T07:45:39.592468+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A random sample of 200 generated meshes is run through an independent finite-element solver; if more than 10 percent produce stress or displacement values that deviate by more than 15 percent from the SimJEB reference under identical boundary conditions, the claim of label fidelity fails.","supporting_citations":[],"review_version":1}