new | ask | show | jobs
7speter 2 days ago  [ - ]

You can offload the vision projector to CPU/sysRAM

slim 2 days ago  [ - ]

Awesome. How would you setup that ?

skirmish 2 days ago  [ - ]

  --no-mmproj-offload
See the documentation: https://github.com/ggml-org/llama.cpp/blob/master/docs/multi...