Hi, and thanks for releasing these models.
This follows up on the [report on the SigLIP2-Base repo](https://huggingface.co/AvitoTech/SigLIP2-Base-for-animal-identification/discussions).
- **`AutoModel.from_pretrained` did not work.** Every tensor was stored under a `clip.` prefix left over from the training wrapper, which does not match `SiglipModel.base_model_prefix`, so all 408 parameters were randomly initialised and only a warning was emitted. The tensors are now stored under their canonical names.
- **`config.json`.** `text_config.hidden_act`: `gelu_pytorch_tanh`; `vision_config.hidden_act`: `gelu_pytorch_tanh`.
- **Image processor.** Added `preprocessor_config.json`, copied unchanged from `google/siglip-base-patch16-224`, so the repo is self-contained and the card no longer has to send users to another repository for preprocessing.
- **License.** Added `license: apache-2.0` and a `LICENSE` file (see below).
### Verification
The reference is the wrapper this checkpoint was tra…
