example_bundle / README.txt
RozanskiT's picture
Upload 6 files
8b93567 verified
Astro Emulators Toolkit Bundle
Summary:
model: mlp
task: regression
fit_method: gradient
x_names: ('teff', 'logg', 'feh')
y_names: ('flux',)
Domain:
input_domain: {'kind': 'box_v1', 'space': 'runtime_tree_v1', 'storage': {'format': 'safetensors_v1', 'filename': 'input_domain.safetensors', 'layout': 'tree_minmax_v1'}}
Provenance:
toolkit_version: 0.1.0
saved_at_utc: 2026-03-13T02:55:33.972802+00:00
python_version: 3.12.11
git_sha: 72e167d2c8d68abc1e2c6c4b2503e06f3a375537
spec:
spec_version: 2
x:
names:
- teff
- logg
- feh
meaning:
- effective temperature
- surface gravity
- metallicity [Fe/H]
units:
- K
- dex
- dex
y:
names:
- flux
meaning:
- continuum-normalized flux vector on the shared wavelength grid
units:
- dimensionless
wavelength_angstrom:
array(shape=(500,), dtype=float64)
input_domain:
kind: box_v1
space: runtime_tree_v1
storage:
format: safetensors_v1
filename: input_domain.safetensors
layout: tree_minmax_v1
reference_scaling:
kind: affine_v1
parametrization: minmax
space: runtime_tree_v1
storage:
format: safetensors_v1
filename: reference_scaling.safetensors
layout: tree_minmax_v1
notes: Inputs are min-max scaled before training using the fixed bounds stored in this bundle. Output reference scaling is identity.
Note: this bundle is the canonical emulator artifact. Physical-space composition is external.