4 ms·
You can have multiple models served now with loading/unloading with just the server binary. https://github.com/ggml-org/llama.cpp/blob/master/tools/server/READ
by dminik 6mo ago
You can have multiple models served now with loading/unloading with just the server binary.
https://github.com/ggml-org/llama.cpp/blob/master/tools/server/README.md#model-presets https://github.com/ggml-org/llama.cpp/blob/master/tools/serv...
- speedgoose 6mo agoIt only lacks the automatic FIFO loading/unloading then. Maybe it will be there in a few weeks.