Repository navigation
TensorTune v1.1.1 - Smarter Diagnostics & Enhanced Stability
This is TensorTune v1.1.1, a refinement release focusing on improved user guidance for optional dependencies, smarter startup diagnostics in both GUI and CLI, and important bug fixes for a smoother user experience.
✨ Key Highlights in v1.1.1:
Context-Aware Library Warnings (CLI & GUI): Startup logs now provide more relevant status for optional GPU monitoring libraries (PyADLX for AMD, PyZE for Intel, WMI for Windows, Metal for Apple). This significantly reduces irrelevant log noise based on your detected hardware, making it easier to spot actual issues.
CLI Warning Suppression: You can now toggle a setting in the CLI's "Launcher Settings" (w option) to suppress these non-critical optional library warnings after the first run, while still seeing critical alerts.
📚 Detailed Setup Guides for Optional Libraries: To assist users who want to leverage advanced GPU monitoring, we've added:
-
PYADLX_SETUP_GUIDE.md (for manual setup of PyADLX on Windows for AMD GPUs)
-
PYZE_SETUP_GUIDE.md (for PyZE/pyze-l0 for Intel Arc/Xe GPUs)
-
WMI_SETUP_GUIDE.md (for the WMI Python package on Windows)
-
TensorTune's error messages for these libraries will now point you directly to these guides.
🐛 CLI Bug Fix: Resolved a critical SyntaxError that could occur in the CLI during "Launch Best Remembered Config" or "Direct Launch" operations. VRAM decision data is now handled more robustly.
🛠️ GUI Stability: Refined the GUI's startup sequence for better stability and to ensure all necessary data is available for initial logging and UI population.
📄 Documentation & Installation:
- All relevant project files (README.md, tensortune_install.py, tensortune_examples.py, requirements.txt comments) have been updated to reflect version 1.1.1 and the latest features.
- The appdirs library has been added as a default requirement for consistent and correct user-specific configuration/data path management across different operating systems.