MCPcopy Create free account
hub / github.com/Tiiny-AI/PowerInfer / n_gqa

Method n_gqa

smallthinker/src/llama-hparams.cpp:51–60  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

49}
50
51uint32_t llama_hparams::n_gqa(uint32_t il) const {
52 const uint32_t n_head = this->n_head(il);
53 const uint32_t n_head_kv = this->n_head_kv(il);
54
55 if (n_head_kv == 0) {
56 return 0;
57 }
58
59 return n_head/n_head_kv;
60}
61
62uint32_t llama_hparams::n_embd_k_gqa(uint32_t il) const {
63 const uint32_t n_head_kv = this->n_head_kv(il);

Callers 2

print_infoMethod · 0.45
llama_tensor_get_typeFunction · 0.45

Calls 2

n_headMethod · 0.95
n_head_kvMethod · 0.95

Tested by

no test coverage detected