llama : refactor src/llama.cpp (#10902)

* llama : scatter llama.cpp into multiple modules (wip) * llama : control-vector -> adapter * llama : arch * llama : mmap ggml-ci * ci : remove BUILD_SHARED_LIBS=OFF ggml-ci * llama : arch (cont) ggml-ci * llama : chat ggml-ci * llama : model ggml-ci * llama : hparams ggml-ci * llama : adapter ggml-ci * examples : fix ggml-ci * rebase ggml-ci * minor * llama : kv cache ggml-ci * llama : impl ggml-ci * llama : batch ggml-ci * cont ggml-ci * llama : context ggml-ci * minor * llama : context (cont) ggml-ci * llama : model loader ggml-ci * common : update lora ggml-ci * llama : quant ggml-ci * llama : quant (cont) ggml-ci * minor [no ci]
2025-01-03 10:18:53 +02:00 · 2025-01-03 10:18:53 +02:00 · f66f582927
commit f66f582927
parent 2f0ee84b9b
61 changed files with 12193 additions and 11649 deletions
--- a/src/llama-hparams.cpp
+++ b/src/llama-hparams.cpp
@ -0,0 +1,71 @@
+#include "llama-hparams.h"
+
+#include "ggml.h"
+
+uint32_t llama_hparams::n_head(uint32_t il) const {
+    if (il < n_layer) {
+        return n_head_arr[il];
+    }
+
+    GGML_ABORT("fatal error");
+}
+
+uint32_t llama_hparams::n_head_kv(uint32_t il) const {
+    if (il < n_layer) {
+        return n_head_kv_arr[il];
+    }
+
+    GGML_ABORT("fatal error");
+}
+
+uint32_t llama_hparams::n_ff(uint32_t il) const {
+    if (il < n_layer) {
+        return n_ff_arr[il];
+    }
+
+    GGML_ABORT("fatal error");
+}
+
+uint32_t llama_hparams::n_gqa(uint32_t il) const {
+    const uint32_t n_head    = this->n_head(il);
+    const uint32_t n_head_kv = this->n_head_kv(il);
+
+    if (n_head_kv == 0) {
+        return 0;
+    }
+
+    return n_head/n_head_kv;
+}
+
+uint32_t llama_hparams::n_embd_k_gqa(uint32_t il) const {
+    const uint32_t n_head_kv = this->n_head_kv(il);
+
+    return n_embd_head_k * n_head_kv;
+}
+
+uint32_t llama_hparams::n_embd_v_gqa(uint32_t il) const {
+    const uint32_t n_head_kv = this->n_head_kv(il);
+
+    return n_embd_head_v * n_head_kv;
+}
+
+uint32_t llama_hparams::n_embd_k_s() const {
+    if (wkv_head_size != 0) {
+        // for RWKV models
+        return 2 * n_embd;
+    }
+
+    // TODO: maybe support other convolution strides than 1
+    // NOTE: since the first column of the conv_state is shifted out each time, it's not actually needed
+    return (ssm_d_conv > 0 ? ssm_d_conv - 1 : 0) * ssm_d_inner;
+}
+
+uint32_t llama_hparams::n_embd_v_s() const {
+    if (wkv_head_size != 0) {
+        // corresponds to RWKV's wkv_states size
+        return n_embd * wkv_head_size;
+    }
+
+    // corresponds to Mamba's ssm_states size
+    return ssm_d_state * ssm_d_inner;
+}