---
dataset_info:
  features:
  - name: uid
    dtype: string
  - name: target_id
    dtype: int64
  - name: label
    dtype: string
  - name: fold
    dtype: int8
  - name: paper_score
    dtype: float32
  - name: global_flux_view_fluxnorm
    list: float32
    length: 301
  - name: global_flux_view_fluxnorm_var
    list: float32
    length: 301
  - name: flux_trend_global_norm
    list: float32
    length: 301
  - name: flux_trend_global_norm_var
    list: float32
    length: 301
  - name: pgram_smooth_norm
    list: float32
    length: 674
  - name: pgram_tpm_smooth_norm
    list: float32
    length: 674
  - name: local_centr_view
    list: float32
    length: 31
  - name: local_centr_view_var
    list: float32
    length: 31
  - name: local_momentum_dump_view
    list: float32
    length: 31
  - name: local_momentum_dump_view_var
    list: float32
    length: 31
  - name: local_flux_view_fluxnorm
    list: float32
    length: 31
  - name: local_flux_view_fluxnorm_var
    list: float32
    length: 31
  - name: local_weak_secondary_view_selfnorm
    list: float32
    length: 31
  - name: local_weak_secondary_view_selfnorm_var
    list: float32
    length: 31
  - name: local_flux_odd_view_fluxnorm
    list: float32
    length: 31
  - name: local_flux_odd_view_fluxnorm_var
    list: float32
    length: 31
  - name: local_flux_even_view_fluxnorm
    list: float32
    length: 31
  - name: local_flux_even_view_fluxnorm_var
    list: float32
    length: 31
  - name: unfolded_local_flux_view_fluxnorm
    dtype:
      array2_d:
        shape:
        - 20
        - 31
        dtype: float32
  - name: diff_imgs
    dtype:
      array3_d:
        shape:
        - 5
        - 33
        - 33
        dtype: float32
  - name: oot_imgs
    dtype:
      array3_d:
        shape:
        - 5
        - 33
        - 33
        dtype: float32
  - name: target_imgs
    dtype:
      array3_d:
        shape:
        - 5
        - 33
        - 33
        dtype: float32
  - name: quality
    list: float32
    length: 5
  - name: boot_fap
    dtype: float32
  - name: flux_even_local_stat_abs_min
    dtype: float32
  - name: flux_global_stat_abs_min
    dtype: float32
  - name: flux_local_stat_abs_min
    dtype: float32
  - name: flux_odd_local_stat_abs_min
    dtype: float32
  - name: flux_trend_global_stat_max
    dtype: float32
  - name: flux_trend_global_stat_min
    dtype: float32
  - name: flux_weak_secondary_local_stat_abs_min
    dtype: float32
  - name: mag_shift
    dtype: float32
  - name: pgram_smooth_max_power
    dtype: float32
  - name: pgram_tpm_smooth_max_power
    dtype: float32
  - name: ruwe
    dtype: float32
  - name: tce_albedo_stat
    dtype: float32
  - name: tce_cap_stat
    dtype: float32
  - name: tce_dikco_msky_err
    dtype: float32
  - name: tce_dikco_msky
    dtype: float32
  - name: tce_hap_stat
    dtype: float32
  - name: tce_max_mult_ev
    dtype: float32
  - name: tce_max_sngle_ev
    dtype: float32
  - name: tce_maxmes
    dtype: float32
  - name: tce_model_chisq
    dtype: float32
  - name: tce_num_transits
    dtype: float32
  - name: tce_num_transits_obs
    dtype: float32
  - name: tce_period
    dtype: float32
  - name: tce_prad
    dtype: float32
  - name: tce_ptemp_stat
    dtype: float32
  - name: tce_robstat
    dtype: float32
  - name: tce_sdens
    dtype: float32
  - name: tce_slogg
    dtype: float32
  - name: tce_smass
    dtype: float32
  - name: tce_smet
    dtype: float32
  - name: tce_sradius
    dtype: float32
  - name: tce_steff
    dtype: float32
  splits:
  - name: labeled
    num_bytes: 4678333145
    num_examples: 57162
  - name: unlabeled
    num_bytes: 12077505435
    num_examples: 147567
  download_size: 5654031391
  dataset_size: 16755838580
configs:
- config_name: default
  data_files:
  - split: labeled
    path: data/labeled-*
  - split: unlabeled
    path: data/unlabeled-*
license: cc-by-4.0
size_categories:
- 100K<n<1M
pretty_name: ExoMiner++ TESS 2-min TCEs
tags:
- astronomy
- exoplanets
- tess
---
# ExoMiner++ TESS 2-min TCEs (Hugging Face format)

The ExoMiner++ dataset of TESS SPOC 2-min TCEs, sectors 1-67, converted from the
original TFRecords to Hugging Face `datasets`.

- **Source:** Martinho, M., & Valizadegan, H. (2025). *ExoMiner++: Enhanced Transit
  Classification and a New Vetting Catalog for 2-Minute TESS Data* [Dataset].
  Zenodo. https://doi.org/10.5281/zenodo.15466293
- **Paper:** https://doi.org/10.3847/1538-3881/ae03a4 (arXiv:2502.09790)
- **License:** [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/), as the original.

## Changes from the original

- The TFRecords were decoded and stored as Arrow/Parquet. Feature values are unchanged.
- Each TCE was joined with its catalog row (`label`, `fold`, `paper_score`).
- The data is split into `labeled` (57,162 TCEs) and `unlabeled` (147,567 TCEs).

Built with `scripts/build_dataset.py` from [TODO].

This is not an official NASA release and is not endorsed by NASA.

## Columns

- `uid`, `target_id`: TCE and TIC identifiers.
- `label`: KP, CP, BD, EB, FP, NTP, NEB or NPC for labeled TCEs; UNK for unlabeled ones.
- `fold`: the paper's cross-validation fold 0-9; -1 for unlabeled TCEs.
- `paper_score`: the ExoMiner++ score from the original catalog.
- All other columns are the raw, unnormalized TFRecord features: scalars, 1-D views
  (31, 301 or 674 bins), the unfolded flux view (20 x 31) and the difference images
  (5 x 33 x 33).

## Usage

```python
import datasets
labeled = datasets.load_dataset("29thDay/exominer-plusplus", split="labeled")
labeled = labeled.with_format("numpy")
```