vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 17:14:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Expose low_vram for CUDA

Browse source 
Enabling --lowvram instructs the program to not allocate a VRAM scratch buffer for holding temporary results. Reduces VRAM usage at the cost of performance, particularly prompt processing speed. Requires CUDA
This commit is contained in:
YellowRoseCx 2023-06-26 16:47:22 -05:00
parent d2034ced7b
commit 8afa800fb6
3 changed files with 6 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

1
gpttype_adapter.cpp
View file

@ -377,6 +377,7 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
//llama_ctx_paran_parts = -1;
llama_ctx_params.seed = -1;
llama_ctx_params.f16_kv = inputs.f16_kv;
llama_ctx_params.low_vram = inputs.low_vram;
llama_ctx_params.logits_all = false;
llama_ctx_params.use_mmap = inputs.use_mmap;
llama_ctx_params.use_mlock = inputs.use_mlock;