MCPcopy Create free account
hub / github.com/appdevforall/CodeOnTheGo / n_head_kv

Method n_head_kv

subprojects/llama.cpp/src/llama-hparams.cpp:38–44  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

36}
37
38uint32_t llama_hparams::n_head_kv(uint32_t il) const {
39 if (il < n_layer) {
40 return n_head_kv_arr[il];
41 }
42
43 GGML_ABORT("fatal error");
44}
45
46uint32_t llama_hparams::n_ff(uint32_t il) const {
47 if (il < n_layer) {

Callers 15

n_gqaMethod · 0.95
n_embd_k_gqaMethod · 0.95
n_embd_v_gqaMethod · 0.95
load_hparamsMethod · 0.80
load_tensorsMethod · 0.80
print_infoMethod · 0.80
llama_model_n_head_kvFunction · 0.80
llm_graph_contextMethod · 0.80
get_kMethod · 0.80
get_vMethod · 0.80
build_graph_shiftMethod · 0.80
build_attention_layerMethod · 0.80

Calls

no outgoing calls

Tested by

no test coverage detected