Autogguf

Added
- Dynamic KV Overrides (see wiki: AutoGGUF/wiki/Dynamic-KV-Overrides)
- Quantization commands are now printed and logged

Added
- LoRA Conversion:
- New section for converting HuggingFace PEFT LoRA adapters to GGML/GGUF
- Output type selection (GGML or GGUF)
- Base model selection for GGUF output
- LoRA adapter list with individual scaling factors
- Export LoRA section for merging adapters into base model
- UI Improvements:
- Updated task names in task list
- IMatrix generation check
- Larger window size
- Added exe favicon
- Localization:
- French and Simplified Chinese support for LoRA and "Refresh Models" strings
- Code and Build:
- Code organization improvements
- Added build script
- .gitignore file
- Misc:
- Currently includes src folder with conversion tools
- No console window popup

Added
- AUTOGGUF_CHECK_BACKEND environment variable to disable backend check on start

Changed
- --onefile build with PyInstaller, _internal directory is no longer required

Added
- Support for new llama-imatrix parameters:
- Context size (--ctx-size) input
- Threads (--threads) control
- New parameters to IMatrix section layout
- Slider-spinbox combination for thread count selection
- QSpinBox for output frequency input (1-100 range with percentage suffix)

Changed
- Converted context size input to a QSpinBox
- Updated generate_imatrix() method to use new UI element values
- Improved error handling in preset loading
- Enhanced localization support for new UI elements

Fixed
- Error when loading presets containing KV overrides

Removed
- Duplicated functions

Added
- Refresh Models button
- Linux build (built on Ubuntu 24.04 LTS)

Fixed
- iostream llama.cpp issue, quantized_models directory created on launch

Added
- More robust logging (find logs at latest_<timestamp>.log in logs folder)
- Localizations with support for 28 languages (machine translated using Gemini Experimental 0801)