Chore: Update model hub v0.4.10

[hahuyhoang411]

Describe Your Changes

Update model hubs with new models, adjust all models from Q5 to Q4 and clean up redundant models.

Fixes Issues

• Closes #Request: support for Cohere Command-R #2405
• Closes #Please add ibm/merlinite-7b model #2272
• Closes #feat: Multilingual interface #2004
• Closes #llava v1 vs v1.1 #2391

Self Checklist

• Add:

  • Command-R 34B: Model is optimized for RAG and tool using
  • Merlinite 7B: New 7B model from IBM -> This model is low quality without a proper
  • Vistral 7B: Vietnamese model
  • Mistral 70B Q4: leak weight
  • Hermes Pro 7B: Update version of Openhermes 7B

• Adjust:

  • yi34B Q5 -> yi34B Q4
  • noromaid v0.1 -> v0.4
  • Openchat: Better description
  • llava 1.5 -> llava 1.6 (for fixing bug)
  • Gemma 7B as the featured model
  • Stable Zephyr 3B as the featured model for low-end machine

• Delete:

  • Tinyllama 1.1B
  • Tulu 70b
  • Capybara 34B
  • Solar 11B
  • Starling 7B

• Folder ID update:

  • stealth 7b
  • trinity 7b

• Version model extension from 1.0.27 to 1.0.28

[louis-menlo]

LGTM