vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

format

Browse source
This commit is contained in:
liyuhang 2025-01-29 08:26:13 +08:00
parent 1099ef271e
commit 96bde6fd51
2 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

2
src/llama-model.cpp
View file

@ -3081,7 +3081,7 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
layer.wqkv = create_tensor(tn(LLM_TENSOR_ATTN_QKV, "weight", i), {n_embd, n_embd + 2*n_embd_gqa}, llama_model_loader::TENSOR_NOT_REQUIRED);
layer.bqkv = create_tensor(tn(LLM_TENSOR_ATTN_QKV, "bias", i), {n_embd + 2*n_embd_gqa}, llama_model_loader::TENSOR_NOT_REQUIRED);
if(layer.wqkv == nullptr){
if(layer.wqkv == nullptr) {
layer.wq = create_tensor(tn(LLM_TENSOR_ATTN_Q, "weight", i), {n_embd, n_embd_head_k * n_head}, 0);
layer.wk = create_tensor(tn(LLM_TENSOR_ATTN_K, "weight", i), {n_embd, n_embd_k_gqa}, 0);
layer.wv = create_tensor(tn(LLM_TENSOR_ATTN_V, "weight", i), {n_embd, n_embd_v_gqa}, 0);

2
src/llama.cpp
View file

@ -7215,7 +7215,7 @@ struct llm_build_context {
struct ggml_tensor * Qcur = nullptr;
struct ggml_tensor * Kcur = nullptr;
struct ggml_tensor * Vcur = nullptr;
if(model.type == LLM_TYPE_1_5B|| model.type == LLM_TYPE_4B || model.type == LLM_TYPE_9B){
if(model.type == LLM_TYPE_1_5B|| model.type == LLM_TYPE_4B || model.type == LLM_TYPE_9B) {
Qcur = llm_build_lora_mm(lctx, ctx0, model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {