| ---
|
| license: bsd-3-clause
|
| tags:
|
| - audio-text
|
| - balanced
|
| - domain-specific
|
| - geology
|
| - huggingface
|
| - kfold-5
|
| - randaugment
|
| - semi-auto
|
| - strict
|
| ---
|
|
|
| # prepare.py
|
|
|
| ## Dataset Summary
|
|
|
| A **geology** dataset with **audio text** modality, stored in **huggingface** format.
|
|
|
| ## Preprocessing & Augmentation
|
|
|
| - **Preprocessing**: domain specific
|
| - **Augmentation**: randaugment
|
|
|
| ## Splits & Sampling
|
|
|
| - **Split strategy**: kfold 5
|
| - **Sampling**: balanced
|
|
|
| ## Quality & Labeling
|
|
|
| - **Quality filtering**: strict
|
| - **Labeling**: semi auto
|
|
|
| ## Files
|
|
|
| - `prepare.py` — main artifact of this repository
|
|
|
| ## License
|
|
|
| See the license field above.
|
|
|
