ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-05-01 03:41:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

Just in case

Browse Source
This commit is contained in:
Kawrakow
2026-02-21 18:14:14 +00:00
parent d93fe37ba7
commit 52285a5ee7
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/llama-model.h
View File

@@ -422,7 +422,7 @@ struct llama_model {
size_t max_nodes(int n_tokens) const { size_t max_nodes(int n_tokens) const {
auto n_tensors = tensors_by_name.size(); auto n_tensors = tensors_by_name.size();
if (split_mode == LLAMA_SPLIT_MODE_GRAPH) n_tensors *= devices.size(); if (split_mode == LLAMA_SPLIT_MODE_GRAPH && !devices.empty()) n_tensors *= devices.size();
if (arch == LLM_ARCH_QWEN3NEXT || arch == LLM_ARCH_QWEN35MOE) { if (arch == LLM_ARCH_QWEN3NEXT || arch == LLM_ARCH_QWEN35MOE) {
return std::max<size_t>(n_tokens * 40, 32u * n_tensors); return std::max<size_t>(n_tokens * 40, 32u * n_tensors);
} }