datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
design-patents-not-in-impact
US Design Patents Not Included in IMPACT (2008-2026)
Original drawing images (TIFF) and grant full-text XML for 165,917 US design patents that are
absent from the AI4Patents/IMPACT dataset.
IMPACT covers 2007-2022 and contains 434,498 rows. This dataset supplies the design patents that
IMPACT does not have: 161,093 patents granted in 2023-2026, which are outside IMPACT's period,
plus 4,824 patents from years IMPACT does cover but did not include. There is no patent
overlap with… See the full description on the dataset page: https://huggingface.co/datasets/SoichiOnozuka/design-patents-not-in-impact.SAMTOR_Novel_Target_Designs_GA-II
SAMTOR — SAM-competitive de novo designs (Technetium GA-II)
196 small molecules across two generations, generated de novo by the Technetium TC-43.ai engine (GA-II) and conditioned on the S-adenosylmethionine (SAM) pocket of human SAMTOR.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the site. With no approved drug, clinical candidate or… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/SAMTOR_Novel_Target_Designs_GA-II.FKBP12_F36V_Binder_Designs_1BL4
FKBP12 F36V Binder Designs (PDB 1BL4)
165 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the engineered FKBP12 F36V bump-hole cavity of 1BL4 (1.9 Å).
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the site. Each is supplied as a complete protein–ligand complex.
Molecules were generated by the… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/FKBP12_F36V_Binder_Designs_1BL4.FSHR_Agonist_Designs_8I2G
FSHR Allosteric Agonist Designs — 8I2G (GA-II)
292 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the transmembrane allosteric pocket of the follicle-stimulating hormone receptor (FSHR), using PDB 8I2G as the receptor.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the site. All 292 are profiled for… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/FSHR_Agonist_Designs_8I2G.PDE5_Inhibitor_Designs_2H42
PDE5 Inhibitor Designs — 2H42 / GA-II
Why this target matters. PDE5 inhibitors are among the most widely prescribed drugs in the world, and the target remains actively pursued in pulmonary arterial hypertension and cardiac indications far beyond its original use.
2,090 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the catalytic pocket of human cGMP-specific phosphodiesterase 5 (PDE5A), taken from the sildenafil co-crystal 2H42 (2.3… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/PDE5_Inhibitor_Designs_2H42.SMARCA2_VHL_MolecularGlue_Designs_6HAY
SMARCA2–VHL Molecular Glue Designs (PDB 6HAY)
884 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the SMARCA2 bromodomain–VHL interface of the ternary complex 6HAY (2.24 Å), so that one molecule spans both partners.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the site. Each is supplied as a… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/SMARCA2_VHL_MolecularGlue_Designs_6HAY.Lipoprotein-a-Binder-Designs
Lipoprotein(a) Binder Designs — apo(a) KIV-8 / 8TCE
Why this target matters. Lp(a) is the most common genetic cardiovascular risk factor in the world, affecting roughly a fifth of the population, and the only major one with no approved therapy.
100 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the lysine-binding site of kringle IV type 8 (KIV-8) of apolipoprotein(a), using the 1.07 Å co-crystal 8TCE — Lp(a) KIV-8 in complex with… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/Lipoprotein-a-Binder-Designs.GPR75_Intracellular_Antagonist_Designs_GA-II
GPR75 Intracellular Antagonist Designs — Technetium GA-II
156 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the intracellular allosteric site of GPR75 — a class A orphan GPCR and first-in-class anti-obesity target with human genetic validation.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/GPR75_Intracellular_Antagonist_Designs_GA-II.VHL_NonProline_Binder_Designs_9GIO
VHL Non-Proline Binder Designs (9GIO)
279 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the VHL substrate-recognition pocket of the VHL–ElonginC–ElonginB (VCB) complex (PDB 9GIO), with the field's usual hydroxyproline anchor withheld during generation.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/VHL_NonProline_Binder_Designs_9GIO.GPR75_Inhibitor_Designs_GA-II
GPR75 Inhibitor Designs — Technetium GA-II
139 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the extracellular vestibule of GPR75, a first-in-class anti-obesity target.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the site. Each is docked into a rigid receptor afterwards and supplied as a full… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/GPR75_Inhibitor_Designs_GA-II.RPIA_Inhibitor_Designs_GA-II
RPIA Inhibitor Designs — Technetium GA-II
21 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the active site of RPIA (ribose-5-phosphate isomerase A), a first-in-class longevity target with no prior inhibitor to screen for.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the site. Each is docked into a… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/RPIA_Inhibitor_Designs_GA-II.LHCGR_Agonist_Designs_7FIH
LHCGR Allosteric Agonist Designs — 7FIH (GA-II)
993 small molecules generated de novo by the Technetium TC-43.ai engine (GA-II), conditioned on the transmembrane allosteric pocket of the luteinizing hormone / choriogonadotropin receptor (LHCGR), using PDB 7FIH as the receptor.
Each molecule was constructed against this pocket rather than selected from a compound library — docking (AutoDock Vina) came afterwards, to place and score the generated molecules in the site. All 993 are… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/LHCGR_Agonist_Designs_7FIH.THRB_Agonist_Designs_1R6G
THRβ Agonist Designs — 1R6G (GA-II)
Why this target matters. Thyroid hormone drives lipid clearance and hepatic fat oxidation,
but thyroid hormone itself is unusable as a drug because THRα engagement causes tachycardia,
arrhythmia and bone loss. Resmetirom's 2024 approval for MASH proved that a β-selective
thyromimetic can separate the metabolic benefit from the cardiac liability — and made
β-over-α selectivity the whole game for this target.
188 small molecules generated de… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/THRB_Agonist_Designs_1R6G.Nav1.7_VSD4_NonAnionic_Designs_8I5G
Nav1.7 VSD4 Non-Anionic Designs — 8I5G (GA-II)
Why this target matters. Human genetics validates Nav1.7 as thoroughly as any pain target
in existence — loss-of-function carriers are congenitally insensitive to pain, gain-of-function
mutations cause erythromelalgia — yet no Nav1.7 blocker has reached approval, and the clinical
failures of the aryl sulfonamide class are widely attributed to the properties that come with
its anionic warhead.
459 small molecules generated de novo… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/Nav1.7_VSD4_NonAnionic_Designs_8I5G.africa-synth-concrete-mix-design-nigeria-nigeria
Africa Synth Concrete Mix Design Nigeria Nigeria | Africa (Electric Sheep Africa metadata inventory)
Size category: 10K<n<100K - Formats: csv - Sector: infrastructure_transport - Engineered by Electric Sheep Africa
TL;DR
This dataset is part of the Electric Sheep Africa catalog on Hugging Face. It is indexed for African data discovery with standardized metadata, loading guidance, provenance notes, and analyst-oriented context.
What This Dataset… See the full description on the dataset page: https://huggingface.co/datasets/electricsheepafrica/africa-synth-concrete-mix-design-nigeria-nigeria.truss-design-studyThis dataset containe a variety of structural truss designs from a human subjects research experiment.
ant-design-commitsDesign2Code-VLMEvalKitcontrollable-amp-design-dataset
Controllable AMP Design — Curated E. coli MIC Dataset
10,044 curated antimicrobial peptide (AMP) sequences paired with a continuous minimum
inhibitory concentration (MIC) activity score against E. coli, cleaned and deduplicated from
DBAASP v3. Used to train the CVAE generator and Judge predictor in
Sloudis/controllable-amp-design
(GitHub repo).
This is the cleaned/derived dataset only. The raw DBAASP bulk exports it was built from are
not redistributed here — their… See the full description on the dataset page: https://huggingface.co/datasets/Sloudis/controllable-amp-design-dataset.fleets-of-wwii-design-commentary
Fleets of World War II — warship classes & design commentary
The structured database inside Fleets of World War II: Design History and Analysis
for Every Ship of Every Navy (Richard Worth, Nimble Books, ISBN 9781608881604). Two layers:
config
rows
what
ship_classes
1009
one record per warship class — nation, ship type, class name, ships in class, key specs, and Worth's design commentary (why it was built that way, its strengths/flaws)
section_overview
216
the Nation… See the full description on the dataset page: https://huggingface.co/datasets/wfzimmerman/fleets-of-wwii-design-commentary.clinical-therapeutic-protocol-protocol-design-tradeoff-navigation-v0.1What this dataset tests
Whether a model can design a multi-step treatment protocolthat manages trade-offs across systems over time.
Required outputs
protocol_timeline
interaction_risk_map
monitoring_plan
adherence_plan
Typical failures
single-step recommendations without time staging
ignoring drug interactions and contraindications
no monitoring cadence
no adherence plan
Suggested prompt wrapper
System
You design a long-horizon treatment protocol under multi-system… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-therapeutic-protocol-protocol-design-tradeoff-navigation-v0.1.ORMDL3_Ceramide_Site_Designs_7YIU
ORMDL3–SPT Ceramide-Sensing Site — GA-II Designed Ligand–Receptor Complexes
Why this target matters. The 17q21 locus is the most reproducible genetic association in childhood-onset asthma and acts through ORMDL3 expression — and it drives airway hyperreactivity independently of allergy and inflammation. Every approved asthma biologic targets inflammation, so the patients this mechanism describes are precisely the ones those drugs fail.
477 small molecules generated de novo by… See the full description on the dataset page: https://huggingface.co/datasets/Tc-43/ORMDL3_Ceramide_Site_Designs_7YIU.iot-design-studyfan-design-dataset
Fan Design Dataset
A synthetic dataset of axial fan aerodynamic performance for training surrogate models.
Dataset Description
This dataset contains 10,000 samples (8000 train / 2000 test) of axial fan designs with their predicted aerodynamic performance. Each sample maps 14 geometric and operating parameters to 8 performance metrics.
The data was generated using established turbomachinery correlations (Euler equation, Lieblein loss model, Denton tip clearance… See the full description on the dataset page: https://huggingface.co/datasets/harshaperla/fan-design-dataset.Yonsei_DesignITLAB_Int.DesignStyleLLM_Brutalism_eng_ver.3Gemma2-Design-interviewstudy-design-classification-PubMed