| --- |
| license: apache-2.0 |
| language: |
| - en |
| pipeline_tag: image-classification |
| library_name: transformers |
| tags: |
| - notebook |
| - colab |
| - siglip2 |
| - image-to-text |
| --- |
| |
| <div style=" |
| background: rgba(255, 193, 61, 0.15); |
| padding: 16px; |
| border-radius: 6px; |
| border: 1px solid rgba(255, 165, 0, 0.3); |
| margin: 16px 0; |
| "> |
| Finetune SigLIP2 Image Classification (Notebook) |
| </div> |
|
|
|
|
| This notebook demonstrates how to fine-tune SigLIP 2, a robust multilingual vision-language model, for single-label image classification tasks. The fine-tuning process incorporates advanced techniques such as captioning-based pretraining, self-distillation, and masked prediction, unified within a streamlined training pipeline. The workflow supports datasets in both structured and unstructured forms, making it adaptable to various domains and resource levels. |
|
|
| --- |
|
|
| | Notebook Name | Description | Notebook Link | |
| |-------------------------------------|--------------------------------------------------|----------------| |
| | notebook-siglip2-finetune-type1 | Train/Test Splits | [⬇️Download](https://huggingface.co/prithivMLmods/FineTuning-SigLIP2-Notebook/blob/main/Finetune-SigLIP2-Image-Classification/1.SigLIP2_Finetune_ImageClassification_TrainTest_Splits.ipynb) | |
| | notebook-siglip2-finetune-type2 | Only Train Split | [⬇️Download](https://huggingface.co/prithivMLmods/FineTuning-SigLIP2-Notebook/blob/main/Finetune-SigLIP2-Image-Classification/2.SigLIP2_Finetune_ImageClassification_OnlyTrain_Splits.ipynb) | |
|
|
| > [!warning] |
| To avoid notebook loading errors, please download and use the notebook. |
| --- |
|
|
| The notebook outlines two data handling scenarios. In the first, datasets include predefined train and test splits, enabling conventional supervised learning and generalization evaluation. In the second scenario, only a training split is available; in such cases, the training set is either partially reserved for validation or reused entirely for evaluation. This flexibility supports experimentation in constrained or domain-specific settings, where standard test annotations may not exist. |
|
|
| ``` |
| last updated : jul 2025 |
| ``` |
| --- |
|
|
| <div style=" |
| background: rgba(255, 193, 61, 0.15); |
| padding: 16px; |
| border-radius: 6px; |
| border: 1px solid rgba(255, 165, 0, 0.3); |
| margin: 16px 0; |
| "> |
|
|
|
|
| | **Type 1: Train/Test Splits** | **Type 2: Only Train Split** | |
| |------------------------------|------------------------------| |
| |  |  | |
|
|
| </div> |
|
|
| --- |
|
|
| | Platform | Link | |
| |----------|------| |
| | Huggingface Blog | [](https://huggingface.co/blog/prithivMLmods/siglip2-finetune-image-classification) | |
| | GitHub Repository | [](https://github.com/PRITHIVSAKTHIUR/FineTuning-SigLIP-2) | |
|
|
