Llama-assistant

https://github.com/user-attachments/assets/077c8692-e148-46aa-a659-7171a1bbc165

🔧 Changes
- [x] Utilize llama cpp KV cache mechanism to make faster inference. See (1)
- [x] Summarize the chat history when it is about to exceeds the context length
- [x] Recursive check and update missing setting from te DEFAULT CONFIG
- [x] Add validators (type, min, max value) for input fields in the setting dialog

(1) llama cpp's KV cache check prefix of your chat history to reuse the K-V cache. For example:
- Generated sequence so far = "ABCDEF"
- If we modify the chat history somehow like: "ABCDXT". Then it matches prefix and reuses the cache for "ABCD" and newly computes the Key and Value vectors for "XT", then generates new responses.
-> So we need to make the most of this mechanism by keep the history prefix as fixed as possible


🔧 Changes
- 💬 Supported continue conversation.
- 🔍 Added RAG support with [LlamaIndex](https://github.com/run-llama/llama_index).
- ⚙️ Added model settings.
- 📝 Added markdown support.
- ⌛ Added loading text animation while downloading the models and generating answers.
- 🔄 Fixed chatbox scrolling issue.


Thank gallegi for adding these features! 👍


🔧 Changes
- 🔄 Replace Whisper implementation with [pywhispercpp](https://github.com/vietanhdev/pywhispercpp).
- 🖥️ Add built versions:
- 🍎 MacOS
- 🪟 Windows
- 🐧 Linux


🔧 Changes
- 🎙️🔥 **Add offline STT support:** WhisperCPP. The base model is downloaded from Hugging Face. Your audio is transcribed locally on your machine. 🔥


🔧 Changes
- Add binary build for MacOS with PyInstaller.


🐛 Bugfixes
- 🖱️ Fixed: wrong cursor position when inserting text.
- 💥 Fixed: crashing when changing shortcut.
- ⌨️ Fixed: wrong shortcut keys in macOS.
- 🙈 Hide "Copy Result" or "Clear" after clearing the chat.
- 🔄 Handle the last response correctly.

🔧 Changes
- 🧹 Clear results + input when clicking "Clear".
- ⌨️ Type "clear" or "cls" to clear.
- 🚫 Prevent action buttons when no data is input.