LLM inference in C/C++
C++ ggml
model : skip gate_up_exps if TENSOR_SKIP is set (#29014)
David Friehsb49650adb31f
diff --git a/src/llama-model.cpp b/src/llama-model.cppindex 3607bacd6e..de3b2e38ff 100644--- a/src/llama-model.cpp+++ b/src/llama-model.cpp@@ -3251,6 +3251,15 @@ ggml_tensor * llama_model_base::create_tensor(const LLM_TN_IMPL & tn, const std:}void llama_model_base::create_tensor_gate_up_exps(llama_layer & layer, int bid, int64_t n_embd_, int64_t n_ff_, int64_t n_expert_, int flags) {+ if (flags & TENSOR_SKIP) {+ const int skip = TENSOR_NOT_REQUIRED | TENSOR_SKIP;++ create_tensor(tn(LLM_TENSOR_FFN_GATE_UP_EXPS, "weight", bid), {n_embd_, n_ff_ * 2, n_expert_}, skip | TENSOR_SKIP_IF_VIRTUAL);+ create_tensor(tn(LLM_TENSOR_FFN_GATE_EXPS, "weight", bid), {n_embd_, n_ff_, n_expert_}, skip);+ create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS, "weight", bid), {n_embd_, n_ff_, n_expert_}, skip);+ return;+ }+layer.ffn_gate_up_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_UP_EXPS, "weight", bid), {n_embd_, n_ff_ * 2, n_expert_}, TENSOR_NOT_REQUIRED);if (layer.ffn_gate_up_exps == nullptr) {layer.ffn_gate_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_EXPS, "weight", bid), {n_embd_, n_ff_, n_expert_}, flags);