Biaggio74's picture
Upload folder using huggingface_hub
2f49884 verified
|
Raw
History Blame Contribute Delete
7.06 kB
metadata
license: cc-by-nc-4.0
pretty_name: Formeo Portal Frame Dataset v1
task_categories:
  - tabular-regression
  - reinforcement-learning
tags:
  - structural-engineering
  - eurocode
  - portal-frame
  - physics-informed
  - sizing-trajectories
language:
  - en
size_categories:
  - 1K<n<10K
configs:
  - config_name: samples
    default: true
    data_files:
      - split: train
        path: splits/samples_train.parquet
      - split: validation
        path: splits/samples_validation.parquet
      - split: test
        path: splits/samples_test.parquet
  - config_name: trajectories
    data_files:
      - split: all
        path: trajectories.parquet
  - config_name: checks
    data_files:
      - split: all
        path: checks.parquet

Formeo Portal Frame Dataset

50k samples / hour. Physics-verified structural samples from Formeo's Dataset Factory — generator, Eurocode verifier, and sizing trajectories. Need another structure type or engineering domain? Get in touch: info@formeo.ai.

Summary

Physics-verified structural engineering dataset for steel portal frames: design inputs, reference solutions, per-clause Eurocode checks, sizing trajectories, labeled ULS/SLS outcomes, and alignment with a deterministic scoring checker.

The public sample is the proof. The sellable asset is the generator + verifier + reasoning traces for a verifiable engineering domain.

Formeo is an AI-driven structural engineering design platform developed by Kisfali Engineering Srls. This repository publishes a public sample of the corpus under Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). The generation pipeline, verifier API, and full corpus remain proprietary.

  • HF dataset: Formeo/portal-frame-dataset

  • Factory throughput (measured): ~50k samples / hour

  • Structure family: single-bay pitched-roof steel portal frame

  • Public sample: 2,000 samples (cc-by-nc-4.0)

  • Full corpus (licensed): 20,000 samples (regeneration on request)

  • Schema version: 1.1

  • Units: SI (m, kN, kNm, kg, kgCO2e)

  • Feasible rate (ULS and SLS ≤ 1.0) on full run: 99.8%

Dataset contents

Content Purpose
Frame inputs (geometry / loads / grade) Task specification
Final steel sections + mass / eCO₂ Reference solution
Per-member EN 1993 check table Clause-level labels
SLS deflection (actual vs L/200, H/150) Serviceability labels for supervised learning
failure_mode: none / uls / sls / both Strength vs deflection failure attribution
status Design outcome including labeled non-optimal cases
Sizing trajectory Ordered catalog walk (agent / RL training signal)

Acceptance rule (dataset labels and checker are identical): a design is feasible only when both ULS strength utilization and SLS deflection utilization are ≤ 1.0. Governing utilization = max(ULS, SLS).

This package status mix: infeasible_catalog_limit=3, non_converged_oscillation=1, optimal=1179, suboptimal=817

Intended use

Formeo provides a programmatic task generator and a deterministic Eurocode scoring checker for structural design. Downstream systems can train or evaluate models that propose member sections; the checker returns pass/fail and per-clause diagnostics.

Parameter space (this release)

Ranges below match the factory sampler (CONTINUOUS_BOUNDS / CATEGORICALS) used to generate the data.

Parameter Field Range / levels Type
Span span 12–25 m float
Eaves height h_eaves 4–10 m float
Roof pitch pitch 3–15 deg float
Frame spacing frame_spacing 5–8 m float
Basic wind velocity v_b0 22–26 m/s float
Snow on ground s_k 0.4–0.8 kN/m² float
Cladding + services DL g_k_add 0.15–0.25 kN/m² float
Base fixity base_fixity pinned, fixed categorical
Terrain terrain II, III categorical
Steel grade grade S275, S355 categorical
Column family column_family IPE, HEA categorical
Rafter family rafter_family IPE, HEA categorical

Initial section guesses are sampled within the chosen IPE/HEA family; final sections result from the sizing catalog walk (same families). Catalog upper bound in this release: IPE600 / HEA as in the Formeo materials DB.

Design status labels

Status Meaning
optimal ULS and SLS both ≤ 1.0; governing utilization in [0.90, 1.00]
suboptimal Both pass but governing utilization < 0.90 (code-ok, overdesigned)
infeasible_catalog_limit Still over-utilised; catalog could not step up
non_converged_oscillation Over-utilised for other sizing reasons
error Pipeline failure (not included in the published sample)

failure_mode indicates whether the overshoot is strength (uls), deflection (sls), or both.

Check fidelity

  • EN 1993 member checks: Formeo simplified implementation (not a full commercial design suite covering every clause of EN 1993-1-1).
  • SLS: vertical apex δz ≤ L/200, horizontal eaves δx ≤ H/150. Actual displacements and limits are stored per sample.
  • Linear 3D frame FEA; idealized single portal frame; no haunches, connections, fire, or fatigue.
  • Frame-only eCO₂ (default 1.5 kgCO2e/kg steel A1–A3); purlins/cladding excluded.
  • Single CEN-style annex for v1.

Package contents

  • samples.jsonl — full sample records
  • samples.parquet — all public samples (includes a split column: train / val / test)
  • splits/samples_*.parquet — same rows partitioned for the Hub Dataset Viewer
  • trajectories.parquet / checks.parquet — analytics tables
  • splits.json — train / val / test sample ID lists
  • assets/ — UI screenshot + sample calculation report PDF (when included)
  • qa_plots/ — utilization, status, span–mass diagnostic plots (when included)
  • figures/ — curated learnability / env-demo plots (when included)
  • manifest.json — sample size and status counts

Legal & Copyright

  • Copyright: © 2026 Kisfali Engineering Srls. All rights reserved.
  • Dataset public license: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) (cc-by-nc-4.0).
  • Proprietary assets: The generation pipeline, verifier API, full corpus, and commercial usage rights are owned exclusively by Kisfali Engineering Srls.

For commercial licensing, custom structure families, verifier API access, or synthetic data for another engineering need, contact Kisfali Engineering Srls. at info@formeo.ai.

Citation

@dataset{formeo_portal_frame_2026,
  author       = {Kisfali Engineering Srls.},
  title        = {Formeo Dataset Factory: Portal Frame v1},
  year         = 2026,
  publisher    = {Hugging Face},
  url          = {https://huggingface.co/datasets/Formeo/portal-frame-dataset}
}