Changes

• Fixed Metal K_quants
• UI Improvements

Changes

• Added Metal support for llama inference (not for K_Quant), Metal must be enabled in the model settings.

Changelog

• Add gpt2 inference
• Add replit inference
• Add support for k_quants
• Add chat reload button
• Add gpt_neox updated
• Fixed custom prompt format
  for example for ORCA its: ### User:\n{{prompt}}\n\n### Response:\n
• Fixed RedPajma
• Fixed context params on load
• Fixed memory leak on reload model
• Fixed autoscroll in message view

Install ipa by sideloading method like AltStore, Sideloadly, diawi.com etc.
For MacOS install by copy to Application dir.