RGES-PIT/Experienced

IMPORTANT UPDATE (July 22, 2026):

We have updated the dataset to resolve a configuration issue in the simulations. To ensure a fair competition for all participants, we cannot currently elaborate on the specific nature of the issue or the fix. Any models that were fit to the previous dataset must be refit to this new data before submission. We sincerely apologize for any inconvenience this may cause.

The data for the beginner tier events (RMDC26_000001-RMDC26_000188) are not affected by this update, and are still valid for submission in the experienced tier of the challenge.

Please also note the following technical details regarding this update:

• The saturation flag for some data has changed. We do not recommend using saturated data.

• We do not recommend using astrometric data if the photometric signal-to-noise ratio in that band is less than five (SNR < 5). Below this SNR threshold, the astrometric simulation produces unphysical values.

Experienced Tier

The experienced tier data can be found in RMDC26_Experienced_Tier_test.parquet. If read in properly, it should have a shape of 103,281,456 rows × 18 columns and contain lightcurves for 2,087 unique events (event names range from RMDC26_000189 to RMDC26_002288, with 13 events excluded). The columns are as follows.

Columns

Column Name	Data Type	Description
name	string	Unique event identifier (e.g., RMDC26_000189). Format: RMDC26_XXXXXX. Range: RMDC26_000189 to RMDC26_002288 (2,087 events total).
l_deg	float64	Galactic longitude of the event in degrees. Range: [0, 360].
b_deg	float64	Galactic latitude of the event in degrees. Range: [-90, 90].
ra_deg	float64	Catalog right ascension of the event in degrees. Range: [0, 360].
dec_deg	float64	Catalog declination of the event in degrees. Range: [-90, 90].
bjd	float64	Barycentric Julian Date of the observation. Approximate range: 2,461,447–2,463,164.
filt	string	Filter used for the observation. Values: F087, F146, F213.
flux_uJy	float64	Measured flux in microjanskys (μJy).
flux_err_uJy	float64	Uncertainty in the measured flux in μJy. Always > 0.
extinction	float64	Extinction in the specific filter's band. Range: [0, ~16.4].
saturation_flag	int64	Saturation flag: 0 = unsaturated, 1 = saturated. ~0.04% of observations are saturated.
measured_ra_deg	float64	Measured right ascension in degrees. Range: [0, 360].
measured_dec_deg	float64	Measured declination in degrees. Range: [-90, 90].
measured_ra_err_deg	float64	Uncertainty in measured RA in degrees. Always > 0.
measured_dec_err_deg	float64	Uncertainty in measured Dec in degrees. Always > 0.
obs_x	float64	Observatory X-coordinate in AU (Cartesian-ecliptic reference frame). Range: [-1.01, 1.02].
obs_y	float64	Observatory Y-coordinate in AU (Cartesian-ecliptic reference frame). Range: [-0.64, 0.62].
obs_z	float64	Observatory Z-coordinate in AU (Cartesian-ecliptic reference frame). Range: [-0.005, 0.005].

Astrometry Note: Astrometric positions are calculated as the flux-weighted average position of the source images and lens flux. Any additional ambient flux is assumed to be isotropic.

Dataset Statistics

• Total rows: 102,885,552
• Unique events: 2,087
• Observations per event: 49,488 (all events have the same number)
• Filters per event: All events have observations in 3 filters (F087, F146, F213)
• Filter distribution: F146 (~93.9%), F087 (~3.0%), F213 (~3.0%)
• Saturated observations: ~0.04% (42,831 out of 103,281,456)

Simulation Details & Disclaimer

Simulation Engine: All events have been generated using GULLS. More information is available at gulls-microlensing.github.io.

Event Complexity:

• Up to two source stars per event.
• Up to four lens bodies per event, which may include up to two stars, two planets, and one moon.
• Orbital motion of sources and lenses
• Orbital parallax

Sample Bias & Selection:

• Anomaly Filtering: This dataset has been filtered to include only events that likely contain an anomaly (e.g., stellar binary, planetary or lunar perturbations). Note that some false-positive anomaly detections may be present.
• Non-Representativeness: Due to this filtering, the sample is not representative of the full population of events expected from the Roman Space Telescope, nor is it representative of the full range of anomaly types expected in the actual mission.
• Stellar Variability: To increase realism, stellar variability due to starspots has been added to a subset of the sources using butterpy.

Disclaimer: This dataset is intended for testing and development purposes and should be used with the understanding of the biases described above.

Uploading (Organization Members Only)

CLI

# Install the Hugging Face CLI
brew install huggingface-cli

# Login with your Hugging Face credentials
hf auth login

# Push your dataset files
hf upload RGES-PIT/Experienced . --repo-type=dataset

Python

from huggingface_hub import HfApi

api = HfApi(token=os.getenv("HF_TOKEN"))
api.upload_folder(
    folder_path="/path/to/local/dataset",
    repo_id="RGES-PIT/Experienced",
    repo_type="dataset",
)

HTTPS

# Make sure git-lfs is installed (https://git-lfs.com)
git lfs install

git remote add origin https://huggingface.co/datasets/RGES-PIT/Experienced

# You'll be prompted for your HF credentials
git push -u origin main

SSH

# Make sure git-lfs is installed (https://git-lfs.com)
git lfs install

git remote add origin git@hf.co:datasets/RGES-PIT/Experienced

# Make sure SSH key is set in your user settings (https://huggingface.co/settings/keys)
git push -u origin main