ModelScope library is targeted to support training, evaluation and inference for the state of the art models provided by Mind and further support third-party models provided by users outside alibaba.

In order to enable ModelScope users to use the various models provided by ModelScope quickly and conveniently, we provide a set of complete Python library, which includes the implementation of ModelScope official models, inference, finetuning and evaluation support for those models such as preprocessor and evaluation metrics. We also provide easy-to-use APIs and rich usage examples. By calling the library, users can write just a few lines of code to complete tasks such as model inference, training, and evaluation, and can also quickly carry out secondary development on this basis to realize their own innovative ideas.

At present, the algorithm models provided by library cover four main AI fields of image, natural language processing, speech, and multi-modality, and dozens of application scenarios and tasks.

Please refer to installation.

You can refer to quick_start for quick start.

We also provide other documentations including:

• Introduction to tasks
• Use pipeline for model inference
• Finetune example
• Preprocessing of data
• Evaluation metrics

This project is licensed under the Apache License (Version 2.0).