Skip to content

LLM inference in C/C++

C++ ggml

Fix stale tensor-split params for draft models (#24814)

Al G2da668617612
diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index e07f6e9863..8d68cff45b 100644
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -1012,9 +1012,17 @@ struct llama_model::impl {
std::vector<layer_dev> dev_layer;
bool has_tensor_overrides;
+
+ std::vector<float> tensor_split_owned;
};
llama_model::llama_model(const llama_model_params & params) : params(params), pimpl(std::make_unique<impl>()) {
+ if (params.tensor_split != nullptr) {
+ // llama_model_params stores tensor_split as a borrowed pointer, but the model
+ // may need it later for tensor-parallel KV-cache split metadata.
+ pimpl->tensor_split_owned.assign(params.tensor_split, params.tensor_split + llama_max_devices());
+ this->params.tensor_split = pimpl->tensor_split_owned.data();
+ }
pimpl->has_tensor_overrides = params.tensor_buft_overrides && params.tensor_buft_overrides[0].pattern;
}