datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Deepface_Annotated_3K
Deepface Annotated 3K Dataset
Deepface_Annotated_3K is a synthetic facial image dataset containing 3K AI-generated faces from StyleGAN2.Each image is automatically annotated with demographic attributes like:
Age (in years)
Gender (with prediction confidence)
Dominant Race (White, Asian, Latino Hispanic, Indian, etc.)
The dataset is designed for research on fairness, bias detection, demographic classification, and synthetic face representation.
Structure information… See the full description on the dataset page: https://huggingface.co/datasets/Subh775/Deepface_Annotated_3K.D05-annotated
D05 Annotated — the full MMAD chain-of-thought corpus
The complete reasoning-augmented version of AI4Manufacturing/D05
(MMAD, multimodal industrial anomaly-detection MCQ): 34,414 records, every one with a
teacher-written chain-of-thought ending in FINAL ANSWER. This is the merge of two tracks
that were built (and are also released) separately, because they make different claims about the
answer:
Track
Records
Answer (annot)
Also released as
Track 1 — rationalized
31… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/D05-annotated.D05-2a-annotated
D05 Track-2a Annotated — re-solved defect classification with chain-of-thought
Want the full corpus in one download? This track is also released merged with
D05-1-annotated as
AI4Manufacturing/D05-annotated
(34,414 records; filter tracks via metadata.cot.source).
A reasoning-augmented, label-corrected subset of AI4Manufacturing/D05
(MMAD, multimodal industrial anomaly-detection MCQ). This release covers Track 2a:
the Defect-Classification questions on the label-poor MMAD… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/D05-2a-annotated.CWRU-annotated
Roles
Roles: reasoning view of CWRU — annot is the source label (ball / inner_race / normal / outer_race), kept machine-parseable as the gold for verification and reward parsing; the model reads query + image, where the image is an envelope spectrum with the theoretical fault frequencies marked. The reasoning column is filled on all 690 records and is the SFT imitation target for this repo; the query enumerates the closed set of labels the answer must come from, and annot… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/CWRU-annotated.XJTU-annotated
Roles
Roles: reasoning view of XJTU — annot is the source label (inner_race / normal / outer_race), kept machine-parseable as the gold for verification and reward parsing; the model reads query + image, where the image is an envelope spectrum with the theoretical fault frequencies marked. The reasoning column is filled on all 907 records and is the SFT imitation target for this repo; the query enumerates the closed set of labels the answer must come from, and annot remains the… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/XJTU-annotated.PADERBORN-annotated
Roles
Roles: reasoning view of PADERBORN — annot is the source label (healthy / inner_race / outer_race), kept machine-parseable as the gold for verification and reward parsing; the model reads query + image, where the image is an envelope spectrum with the theoretical fault frequencies marked. The reasoning column is filled on all 1,074 records and is the SFT imitation target for this repo; the query enumerates the closed set of labels the answer must come from, and annot… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/PADERBORN-annotated.IMS-annotated
Roles
Roles: reasoning view of IMS — annot is the source label (ball / inner_race / normal / outer_race), kept machine-parseable as the gold for verification and reward parsing; the model reads query + image, where the image is an envelope spectrum with the theoretical fault frequencies marked. The reasoning column is filled on all 542 records and is the SFT imitation target for this repo; the query enumerates the closed set of labels the answer must come from, and annot remains… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/IMS-annotated.MFPT-annotated
Roles
Roles: reasoning view of MFPT — annot is the source label (inner_race / normal / outer_race), kept machine-parseable as the gold for verification and reward parsing; the model reads query + image, where the image is an envelope spectrum with the theoretical fault frequencies marked. The reasoning column is filled on all 78 records and is the SFT imitation target for this repo; the query enumerates the closed set of labels the answer must come from, and annot remains the… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/MFPT-annotated.OTTAWA-VARSPEED-annotated
Roles
Roles: reasoning view of OTTAWA-VARSPEED — annot is the source label (combined / healthy / inner_race), kept machine-parseable as the gold for verification and reward parsing; the model reads query + image, where the image is an ORDER spectrum, whose axis counts events per shaft revolution. The reasoning column is filled on all 35 records and is the SFT imitation target for this repo; the query enumerates the closed set of labels the answer must come from, and annot… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/OTTAWA-VARSPEED-annotated.MAFAULDA-rotor-annotated
Roles
Roles: reasoning view of MAFAULDA-rotor — annot is the source label (imbalance / misalignment), kept machine-parseable as the gold for verification and reward parsing; the model reads query + image, where the image is a vibration spectrum with the shaft-order positions marked. The reasoning column is filled on all 706 records and is the SFT imitation target for this repo; the query enumerates the closed set of labels the answer must come from, and annot remains the string… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/MAFAULDA-rotor-annotated.ITSC-annotated
ITSC-annotated
Park-vector current loci of a three-phase induction motor, as a four-class stator
winding task: normal / phase_A / phase_B / phase_C. 183 records.
reasoning is filled on every record; the twin repo ITSC is identical except that field is empty.
The reading
Both coordinates are divided by the radius of the equal-area circle, so that circle is
the 1.0 ring on every image and the tick numbers carry no current amplitude at all. Two
steps, and both are… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/ITSC-annotated.IEEE-BRB-annotated
IEEE-BRB-annotated
Stator-current spectra of a mains-fed induction motor around the supply line, as a
binary rotor-cage task: healthy / broken_bars. 477 records.
reasoning is filled on every record; the twin repo IEEE-BRB is identical except that field is empty.
The reading
One quantity, and it is what an eye judges on a spectrum: at the pair of marks either
side of the supply line, how far does the line stand above its own two shoulders.
Measured over the 477… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/IEEE-BRB-annotated.SNU-PLANETARY-annotated
SNU-PLANETARY-annotated
Order spectra about the gear mesh line of a single-stage planetary gearbox
(ring fixed; sun 31 teeth, 3 planets of 31, ring 95), as a three-class fault
location task: normal / sun_gear / planet_gear. 917 records. reasoning is filled on every record; the twin repo SNU-PLANETARY is identical except that field is empty.
The reading
The abscissa is the order offset from the mesh line; the ordinate is every line's
height divided by the mesh… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/SNU-PLANETARY-annotated.IEEE-BRB-mcq-annotated
IEEE-BRB-mcq-annotated
Stator-current spectra of a mains-fed induction motor around the supply line, as a
multiple-choice count task. 337 records; the answer is a single letter:
option
meaning
records
A
healthy (no broken bar)
120
B
two broken bars
80
C
three broken bars
50
D
four broken bars
87
reasoning is filled on every record; the twin repo IEEE-BRB-mcq is identical except that field is empty.
The figures are the same files as IEEE-BRB (the binary… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/IEEE-BRB-mcq-annotated.IEEE-BRB-mcq-annotated-loose
IEEE-BRB-mcq-annotated-loose
Stator-current spectra of a mains-fed induction motor around the supply line, as a
multiple-choice count task. 454 records; the answer is a single letter:
option
meaning
records
A
healthy (no broken bar)
120
B
two broken bars
111
C
three broken bars
110
D
four broken bars
113
Temporary comparison copy. Same build as IEEE-BRB-mcq-annotated but without the distance-from-boundary bar (MARGIN_DB = 0.00 dB instead of the published… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/IEEE-BRB-mcq-annotated-loose.