TUI for llama.cpp management
Lazyllama is a utility i am building to make llama.cpp easy to manage and use. still in its early beta, i wanted to go for an opencode-like look with OpenTUI.
One can modify each individual parameter of the model running, and have a small dashboard next to the chat showing context, hardware and other useful info.
Inference is 15% faster than ollama.
The installer auto detects the best backend for your device and, after confirming, asks for a model directory in which one should move models for them to be used.
I also added some documentation in the homepage for users to read, so that every parameter choice is informed and, hopefully, they learn something new.
Sadly because of school i didnt manage to push much but agentic coding really did make a difference. If it wasn't for Space Bunny Free, lazyllama would've never seen the light of day.