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

Method build_norm_gated

subprojects/llama.cpp/src/models/qwen3next.cpp:439–448  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

437}
438
439ggml_tensor * llm_build_qwen3next::build_norm_gated(
440 ggml_tensor * input,
441 ggml_tensor * weights,
442 ggml_tensor * gate,
443 int layer) {
444 ggml_tensor * normalized = build_norm(input, weights, nullptr, LLM_NORM_RMS, layer);
445 ggml_tensor * gated_silu = ggml_silu(ctx0, gate);
446
447 return ggml_mul(ctx0, normalized, gated_silu);
448}
449
450ggml_tensor * llm_build_qwen3next::build_layer_attn(
451 llm_graph_input_attn_kv * inp,

Callers

nothing calls this directly

Calls 2

ggml_siluFunction · 0.85
ggml_mulFunction · 0.85

Tested by

no test coverage detected