Instructions to use paph3285/geoCLIP-tiny-resisc45 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use paph3285/geoCLIP-tiny-resisc45 with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("paph3285/geoCLIP-tiny-resisc45", dtype="auto") - Notebooks
- Google Colab
- Kaggle
π GeoCLIP β Tiny CLIP Trained on RESISC45 (Satellite Scenes)
Fine-tuned Tiny CLIP (β 19 M params) from scratch on RESISC45 satellite scenes.
Pairs: (image, "A satellite image of {label}")
Training Summary
| Parameter | Value |
|---|---|
| Steps | 300 (tiny subset) |
| Optimizer | AdamW (3e-4 lr) |
| Loss β | 1.49 (final) |
| Retrieval Top-5 β | β 15 % |
Visuals
Generated 2025-10-08 02:58 UTC
Inference Providers NEW
This model isn't deployed by any Inference Provider. π Ask for provider support


# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("paph3285/geoCLIP-tiny-resisc45", dtype="auto")