Current lemmatization models make many mistakes, especially in complex languages such as German. They can be updated to the Large model, or better yet, to Transformer. For Transformer, PyTorch must be added. Docker does not work well with GPUs, so a CPU version is needed. This will increase the time it takes to import texts, and the import waiting time will need to be increased to avoid import errors.
Current lemmatization models make many mistakes, especially in complex languages such as German. They can be updated to the Large model, or better yet, to Transformer. For Transformer, PyTorch must be added. Docker does not work well with GPUs, so a CPU version is needed. This will increase the time it takes to import texts, and the import waiting time will need to be increased to avoid import errors.