ConfigILM

ConfigILM is an open-source Python library for rapid iterative development of image-language models in pytorch. It provides a convenient implementation for seamlessly combining models from two popular pytorch libraries, timm and huggingface . This allows a variety of configurations of models without additional implementation effort. At the same time, the interface simplifies the exchange of components of the model and thus offers development possibilities for novel models. In addition, the package provides pre-built and throughput-optimized pytorch dataloaders, allowing developed models to be tested directly in various application areas such as remote sensing (RS) or common objects in context (COCO). The documentation contains installation instructions, tutorial examples, and a complete discussion of the interface to the library. ConfigILM is released under the MIT License, thus encouraging its use in academic and commercial environments. Source code and documentation are available at https://github.com/lhackel-tub/ConfigILM