- Removed stable.sh and unsloth.sh scripts, replacing them with new implementations.
- Added Dockerfiles for CUDA 12 and 13 environments.
- Introduced docker-compose files for llama.cpp and stable-diffusion.cpp builds.
- Created scripts for building from source, checking Hugging Face updates, and creating build images.
- Implemented a new RPC server script for running the model server.
- Enhanced server management with improved status and logging functionalities.
22 days ago
fix: remove RPC configuration from start_server in server.sh
- feat: add llama-swap functionality with configuration and preset files (f7e0ee6a3d74201154b7bbd244109293f4703ba6)
- fix: update README to clarify usage of llama-swap and authentication details (24d317f7be88de253a3815352a5e076e4fe21381)
- feat: enhance llama-swap configuration with single-model group and member definitions (d8bdd8448f2233bba4f78a9d3eb49416b59de123)
- feat: add LLAMA_CACHE environment variable and update cache directory in llama-swap configuration (8cb1635024a46814e1f24216bae9d0ea301f4914)
- fix: remove unused cache directory option from llama-swap model configurations (eb988b1bf07b51d62aaaa3a18e9fa75911bdc249)
- fix: remove LLAMA_CACHE environment variable from docker run command (ed5f7421d2e1b5804997d7a1f9942ff2b19b02e9)
- feat: add Qwopus3.6-35B-A3B-Coder-MTP model configuration and presets (7e3e2d2686f7a3d77df19f8a505a2fede4cd47fe)
- feat: add cache type configurations for Qwopus model (43f33a4618b21b6b0099721c2c98377698318ee3)
- feat: add cache type configurations…
