vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-12 14:11:27 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 3

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/docker.yml
#	docs/ops.md
#	docs/ops/Metal.csv
#	ggml/CMakeLists.txt
#	ggml/src/ggml-sycl/CMakeLists.txt
#	grammars/README.md
#	models/templates/llama-cpp-deepseek-r1.jinja
#	scripts/sync-ggml.last
#	tests/test-chat.cpp
This commit is contained in:
Concedo 2026-01-01 15:34:10 +08:00
commit 54e419f587
28 changed files with 391 additions and 76 deletions
Download patch file Download diff file Expand all files Collapse all files

4
src/llama-context.cpp
View file

@ -1452,7 +1452,9 @@ uint32_t llama_context::graph_max_nodes(uint32_t n_tokens) const {
if (model.arch == LLM_ARCH_QWEN3NEXT) {
return std::max<uint32_t>(n_tokens * 40, 32u * model.n_tensors());
}
return std::max<uint32_t>(1024u, 8u*model.n_tensors());
uint32_t res = std::max<uint32_t>(1024u, 8u*model.n_tensors());
res += model.n_lora_nodes;
return res;
}
llm_graph_result * llama_context::get_gf_res_reserve() const {