v0.7.0

FerrisMind/omnivoice-rsv0.7.0Mar 28, 2026by paulosalem

AI Summary

This release introduces support for vision modality in LLM interactions and migrates the LLM API caching mechanism from pickle to JSON.

Key Highlights

  • Added support for vision modality (image input/output)
  • Migrated LLM API caching from pickle to JSON
  • Added example notebook for Vision for Product, Diagnosis, and Appreciation Feedback

Breaking Changes

  • Migrated LLM API caching from pickle to JSON (may be incompatible with old cache files)

New Features

  • Vision modality support
  • JSON-based LLM API caching
  • New vision example notebook

Full Release Notes

  - Take a look at example [Vision for Product, Diagnosis and Appreciation Feedback (image modality)](./examples/Vision%20for%20Product%2C%20Diagnosis%20and%20Appreciation%20Feedback%20%28image%20modality%29.ipynb) notebook.
  - LLM API caching now uses JSON instead of pickle.