There is a lot to get excited about in Ollama

---

---

AI Summary

• AMA Version 0.1.2 Release Summary
  • Two updates released in one week
    • Majority of new features in 0.1.2.1
    • Version 0.1.2.2 focuses on refinements
• New Features
  • Introduction of new models:
    • Quen
    • Duct
    • Be
    • No Sequel
    • Stable Code
    • New Hermes
    • Two Mixol
    • Stable LM2
  • Official TypeScript and JavaScript library announced
  • New official Python library
• CPU and AVX Instructions
  • Original requirement for AVX instructions
  • AVX used for vector calculations, common in CPUs for over a decade
  • Support extended to older CPUs without AVX and newer CPUs with AVX2
• GPU Compatibility and Error Handling
  • Improved GPU recognition; fallback to CPU if GPU not detected
  • Better support for NVIDIA GPUs in WSL
  • Resolved issues with AMA hanging after multiple requests
  • Enhanced error messages for internet connectivity issues
• User Experience Improvements
  • Messages directive in model file for chat endpoint
    • Allows inclusion of user and assistant roles with questions or answers
  • Configuration settings for models can be adjusted in AMA REPL
    • New /save command to serialize settings to a new model
    • /load command to switch models or clear context
  • Correct display of settings with /show parameters command
• Community Engagement
  • Encouragement to join Discord community at discord.gg/olama
  • Encouragement to share favorite techno evangelist video
• Closing Thoughts
  • Version 0.1.2 updates considered beneficial for all users
  • Specific interest in the messages in the model file feature
  • Call to action: like, subscribe, and thanks for watching
  • Farewell with music

[Music]
Goodbye