Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,93 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# AstroM3Dataset
|
2 |
+
|
3 |
+
## Description
|
4 |
+
|
5 |
+
AstroM3Dataset is a time-series astronomy dataset containing photometry, spectra, and metadata features for variable stars.
|
6 |
+
The dataset includes multiple subsets (`full`, `sub10`, `sub25`, `sub50`) and supports different random seeds (`42`, `66`, `0`, `12`, `123`).
|
7 |
+
Each sample consists of:
|
8 |
+
|
9 |
+
- **Photometry**: Light curve data of shape `(N, 3)` (time, flux, flux\_error).
|
10 |
+
- **Spectra**: Spectra observations of shape `(M, 3)` (wavelength, flux, flux\_error).
|
11 |
+
- **Metadata**: Auxiliary features of shape `(38,)`.
|
12 |
+
- **Label**: The class name as a string.
|
13 |
+
|
14 |
+
## Corresponding paper and code
|
15 |
+
|
16 |
+
- Paper: [AstroM<sup>3</sup>: A self-supervised multimodal model for astronomy](https://arxiv.org/abs/2411.08842)
|
17 |
+
- Code Repository: [GitHub: AstroM<sup>3</sup>](https://github.com/MeriDK/AstroM3/)
|
18 |
+
|
19 |
+
|
20 |
+
## Subsets and Seeds
|
21 |
+
AstroM3Dataset is available in different subset sizes:
|
22 |
+
|
23 |
+
- `full`: Entire dataset
|
24 |
+
- `sub50`: 50% subset
|
25 |
+
- `sub25`: 25% subset
|
26 |
+
- `sub10`: 10% subset
|
27 |
+
|
28 |
+
Each subset is sampled from the respective train, validation, and test splits of the full dataset.
|
29 |
+
For reproducibility, each subset is provided with different random seeds:
|
30 |
+
|
31 |
+
- `42`, `66`, `0`, `12`, `123`
|
32 |
+
|
33 |
+
|
34 |
+
## Data Organization
|
35 |
+
The dataset is organized as follows:
|
36 |
+
```
|
37 |
+
AstroM3Dataset/
|
38 |
+
βββ photometry.zip # Contains all photometry light curves
|
39 |
+
βββ spectra/ # Spectra files organized by class
|
40 |
+
β βββ EA/
|
41 |
+
β β βββ file1.dat
|
42 |
+
β β βββ file2.dat
|
43 |
+
β β βββ ...
|
44 |
+
β βββ EW/
|
45 |
+
β βββ SR/
|
46 |
+
β βββ ...
|
47 |
+
βββ splits/ # Train/val/test splits for each subset and seed
|
48 |
+
β βββ full/
|
49 |
+
β β βββ 42/
|
50 |
+
β β β βββ train.csv
|
51 |
+
β β β βββ val.csv
|
52 |
+
β β β βββ test.csv
|
53 |
+
β β β βββ info.json # Contains feature descriptions and preprocessing info
|
54 |
+
β β βββ 66/
|
55 |
+
β β βββ 0/
|
56 |
+
β β βββ 12/
|
57 |
+
β β βββ 123/
|
58 |
+
β βββ sub10/
|
59 |
+
β βββ sub25/
|
60 |
+
β βββ sub50/
|
61 |
+
βββ AstroM3Dataset.py # Hugging Face dataset script
|
62 |
+
```
|
63 |
+
|
64 |
+
## Usage
|
65 |
+
To load the dataset using the Hugging Face `datasets` library:
|
66 |
+
|
67 |
+
```python
|
68 |
+
from datasets import load_dataset
|
69 |
+
|
70 |
+
# Load the default full dataset with seed 42
|
71 |
+
dataset = load_dataset("MeriDK/AstroM3Dataset")
|
72 |
+
```
|
73 |
+
|
74 |
+
The default configuration is **full_42** (entire dataset with seed 42).
|
75 |
+
To load a specific subset and seed, use {subset}_{seed} as the name:
|
76 |
+
|
77 |
+
```python
|
78 |
+
from datasets import load_dataset
|
79 |
+
|
80 |
+
# Load the 25% subset sampled using seed 123
|
81 |
+
dataset = load_dataset("MeriDK/AstroM3Dataset", name="sub25_123")
|
82 |
+
```
|
83 |
+
|
84 |
+
## Citation
|
85 |
+
If you find this dataset usefull, please cite:
|
86 |
+
```bibtex
|
87 |
+
@article{rizhko2024astrom,
|
88 |
+
title={AstroM $\^{} 3$: A self-supervised multimodal model for astronomy},
|
89 |
+
author={Rizhko, Mariia and Bloom, Joshua S},
|
90 |
+
journal={arXiv preprint arXiv:2411.08842},
|
91 |
+
year={2024}
|
92 |
+
}
|
93 |
+
```
|