Iva-mt

The second release of the iva_mt library, version 0.2.0 introduces the option to load the local model (stored in .tgz archive).

Key Features:
- Local Model Loading: iva_mt now supports loading models directly from local .tgz archives, streamlining the setup process for users with pre-trained models on their machines. This feature enhances the library's usability in environments with limited internet access or when working with proprietary models.

[Installation Instructions and Documentation](https://github.com/cartesinus/iva_mt#readme)

Full Changelog: https://github.com/cartesinus/iva_mt/commits/0.2.0

The initial release of `iva_mt` version 0.1.0! Made for developers and researchers working on virtual assistants and looking for ways to translate their training sets with diverse and high-quality translations.

Key Features:
- **Diverse Verb Variant Generation:** Utilizes VerbNet to produce a variety of verb forms, enhancing the linguistic richness of translation outputs.
- **Conditional Beam Search:** Implements advanced beam search techniques to optimize translation quality and relevance.
- **m2m100 Model Integration:** Leverages state-of-the-art m2m100 models for high-quality machine translation across multiple languages.

[Installation Instructions and Documentation](https://github.com/cartesinus/multiverb_iva_mt#readme)

**Full Changelog**: https://github.com/cartesinus/iva_mt/commits/0.1.0