vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 01:24:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'master' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/build.yml
#	.github/workflows/editorconfig.yml
#	.gitignore
#	CMakeLists.txt
#	README.md
This commit is contained in:
Concedo 2024-01-31 18:53:38 +08:00
commit 15deabd200
15 changed files with 1060 additions and 1371 deletions
Download patch file Download diff file Expand all files Collapse all files

9
llama.cpp
View file

@ -2760,10 +2760,10 @@ static std::string llama_model_ftype_name(llama_ftype ftype) {
case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "Q5_K - Small";
case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "Q5_K - Medium";
case LLAMA_FTYPE_MOSTLY_Q6_K: return "Q6_K";
case LLAMA_FTYPE_MOSTLY_IQ2_XXS:return "IQ2_XSS - 2.0625 bpw";
case LLAMA_FTYPE_MOSTLY_IQ2_XXS:return "IQ2_XXS - 2.0625 bpw";
case LLAMA_FTYPE_MOSTLY_IQ2_XS: return "IQ2_XS - 2.3125 bpw";
case LLAMA_FTYPE_MOSTLY_Q3_K_XS:return "Q3_K - Extra small";
case LLAMA_FTYPE_MOSTLY_IQ3_XXS:return "IQ3_XSS - 3.0625 bpw";
case LLAMA_FTYPE_MOSTLY_IQ3_XXS:return "IQ3_XXS - 3.0625 bpw";
default: return "unknown, may not work";
}
@ -6954,11 +6954,6 @@ static int llama_decode_internal(
n_threads = std::min(4, n_threads);
}
const bool fully_offloaded = model.n_gpu_layers >= (int) hparams.n_layer + 1;
if ((ggml_cpu_has_cublas() || ggml_cpu_has_vulkan()) && fully_offloaded) {
n_threads = 1;
}
#ifdef GGML_USE_MPI
const int64_t n_layer = hparams.n_layer;
ggml_mpi_graph_compute_pre(lctx.ctx_mpi, gf, n_layer);