minjiechen commited on
Commit
c16729c
·
verified ·
1 Parent(s): 18cac84

Publish MagNet dataset library index

Browse files
Files changed (2) hide show
  1. LICENSE +13 -0
  2. README.md +113 -0
LICENSE ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ MagNet Dataset License Notice
2
+
3
+ The four MagNet dataset releases in this repository are licensed under the
4
+ Creative Commons Attribution 4.0 International License (CC BY 4.0).
5
+
6
+ Official legal code:
7
+ https://creativecommons.org/licenses/by/4.0/legalcode
8
+
9
+ Attribution should identify the Princeton MagNet dataset and cite:
10
+ Li et al., “How MagNet: Machine Learning Framework for Modeling Power
11
+ Magnetic Material Characteristics,” IEEE Transactions on Power Electronics,
12
+ 2023.
13
+ https://doi.org/10.1109/TPEL.2023.3309232
README.md ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ pretty_name: Princeton MagNet Magnetic Material Data
3
+ license: cc-by-4.0
4
+ task_categories:
5
+ - tabular-regression
6
+ - time-series-forecasting
7
+ tags:
8
+ - magnetics
9
+ - power-electronics
10
+ - material-science
11
+ - core-loss
12
+ - hysteresis
13
+ size_categories:
14
+ - 100K<n<1M
15
+ configs:
16
+ - config_name: steady-state-scalar
17
+ default: true
18
+ data_files:
19
+ - split: full
20
+ path: releases/magnet-steady-state-training-ready/data/scalar/full/*.parquet
21
+ - config_name: steady-state-sequence
22
+ data_files:
23
+ - split: full
24
+ path: releases/magnet-steady-state-training-ready/data/sequence/full/*.parquet
25
+ - config_name: transient-sequences
26
+ data_files:
27
+ - split: full
28
+ path: releases/magnet-transient-training-ready/data/sequences/full/**/*.parquet
29
+ ---
30
+
31
+ # Princeton MagNet data
32
+
33
+ This is the public, static data library for Princeton MagNet. It contains four
34
+ versioned releases in one dataset repository: steady-state raw and
35
+ training-ready data, and transient raw and training-ready data. The three
36
+ training-ready Parquet views are available as named configurations; the raw
37
+ MATLAB and ZIP files are archival downloads. The dataset is not a leaderboard
38
+ and contains no challenge scores, participant submissions, or team predictions.
39
+
40
+ | Release | Repository folder | Measurements | Materials |
41
+ | --- | --- | ---: | ---: |
42
+ | Steady-state raw | [`releases/magnet-steady-state-raw`](releases/magnet-steady-state-raw/README.md) | 812,492 Combined; 775,865 Cycle | 15 |
43
+ | Steady-state training-ready | [`releases/magnet-steady-state-training-ready`](releases/magnet-steady-state-training-ready/README.md) | 775,865 | 15 |
44
+ | Transient raw | [`releases/magnet-transient-raw`](releases/magnet-transient-raw/README.md) | 122,399 B/H/temperature sequences | 15 |
45
+ | Transient training-ready | [`releases/magnet-transient-training-ready`](releases/magnet-transient-training-ready/README.md) | 122,399 | 15 |
46
+
47
+ The manifest in each folder gives exact file sizes, SHA-256 checksums, source
48
+ lineage, schema version, and validation status. Read the corresponding release
49
+ card before using a file; raw and processed releases have different contracts.
50
+ All releases currently expose one `full` split rather than an arbitrary
51
+ benchmark train/test split.
52
+
53
+ ## Materials
54
+
55
+ Steady-state: `3C90`, `3C92`, `3C94`, `3C95`, `3E6`, `3F4`, `77`, `78`,
56
+ `79`, `ML95S`, `N27`, `N30`, `N49`, `N87`, `T37`.
57
+
58
+ Transient: `3C90`, `3C92`, `3C94`, `3C95`, `3E6`, `3F4`, `77`, `78`,
59
+ `FEC007`, `FEC014`, `N27`, `N30`, `N49`, `N87`, `T37`.
60
+
61
+ ## Use the training-ready data
62
+
63
+ ```python
64
+ from datasets import load_dataset
65
+
66
+ scalar = load_dataset(
67
+ "minjiechen/MagNet", name="steady-state-scalar", split="full"
68
+ )
69
+ waveforms = load_dataset(
70
+ "minjiechen/MagNet", name="steady-state-sequence", split="full"
71
+ )
72
+ transient = load_dataset(
73
+ "minjiechen/MagNet", name="transient-sequences", split="full"
74
+ )
75
+ ```
76
+
77
+ The two steady-state configurations join one-to-one on `measurement_id`.
78
+ Steady-state triangular waveforms retain positive and negative duty fractions;
79
+ sinusoidal duty is null. The transient release retains variable-length B/H
80
+ sequences and a 16 MHz sampling contract. Model users should review quality
81
+ flags, operating ranges, and split-leakage guidance in the individual cards.
82
+
83
+ To download an archival file, use its exact path from the release manifest
84
+ with `hf_hub_download(repo_id="minjiechen/MagNet", repo_type="dataset",
85
+ filename="releases/<release-id>/<path>")`.
86
+
87
+ ## Provenance and limitations
88
+
89
+ The selected T37 steady-state Combined file embeds 2023-09-11 while the Cycle
90
+ file embeds 2023-08-29. The project owner attests that both derive from the
91
+ August 29, 2023 processing and approved publication with this documented
92
+ exception. The original MAT bytes retain both embedded dates; independent
93
+ Cycle-to-Combined derivation has not been verified. See the steady-state cards
94
+ and source catalogs for detail.
95
+
96
+ The ten transient source ZIPs had contributor names and local paths removed
97
+ from Excel package metadata before publication. Scientific worksheet and CSV
98
+ content was preserved; published checksums and the sanitization audit are in
99
+ the transient raw release. This release is not a claim of physical measurement
100
+ accuracy, safety certification, or universal model performance.
101
+
102
+ ## License, citation, and maintainer
103
+
104
+ All four releases are distributed under [Creative Commons Attribution 4.0
105
+ International](https://creativecommons.org/licenses/by/4.0/legalcode)
106
+ (CC BY 4.0). Each release includes its own `LICENSE` file.
107
+
108
+ Please cite Li et al., “How MagNet: Machine Learning Framework for Modeling
109
+ Power Magnetic Material Characteristics,” *IEEE Transactions on Power
110
+ Electronics*, 2023,
111
+ [doi:10.1109/TPEL.2023.3309232](https://doi.org/10.1109/TPEL.2023.3309232).
112
+
113
+ Maintainer: Princeton Power Electronics Research Lab / Minjie Chen.