refactor: Remove n_embd_k/v_s from unified cache

No longer needed now that unified isn't also supporting recurrent

#13979 (comment)

Branch: HybridRecurrentCache

Author: gabe-l-hart

src/llama-kv-cache-unified.cpp

@@ -68,8 +68,8 @@ llama_kv_cache_unified::llama_kv_cache_unified(
             continue;
         }
 
-        const uint32_t n_embd_k_gqa = hparams.n_embd_k_gqa(il) + hparams.n_embd_k_s(il);
-        const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il) + hparams.n_embd_v_s(il);
+        const uint32_t n_embd_k_gqa = hparams.n_embd_k_gqa(il);
+        const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il);
 
         const char * dev_name = "CPU";
 
@@ -1410,7 +1410,7 @@ void llama_kv_cache_unified::state_write_data(llama_io_write_i & io, const std::
     for (const auto & layer : layers) {
         const uint32_t il = layer.il;
 
-        const uint32_t n_embd_k_gqa = hparams.n_embd_k_gqa(il) + hparams.n_embd_k_s(il);
+        const uint32_t n_embd_k_gqa = hparams.n_embd_k_gqa(il);
 
         // Write key type
         const int32_t k_type_i = (int32_t)layer.k->type;
@@ -1432,7 +1432,7 @@ void llama_kv_cache_unified::state_write_data(llama_io_write_i & io, const std::
         for (const auto & layer : layers) {
             const uint32_t il = layer.il;
 
-            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il) + hparams.n_embd_v_s(il);
+            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il);
 
             // Write value type
             const int32_t v_type_i = (int32_t)layer.v->type;
@@ -1456,7 +1456,7 @@ void llama_kv_cache_unified::state_write_data(llama_io_write_i & io, const std::
         for (const auto & layer : layers) {
             const uint32_t il = layer.il;
 
-            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il) + hparams.n_embd_v_s(il);
+            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il);
 
             // Write value type
             const int32_t v_type_i = (int32_t)layer.v->type;
@@ -1599,7 +1599,7 @@ bool llama_kv_cache_unified::state_read_data(llama_io_read_i & io, uint32_t cell
     for (const auto & layer : layers) {
         const uint32_t il = layer.il;
 
-        const uint32_t n_embd_k_gqa = hparams.n_embd_k_gqa(il) + hparams.n_embd_k_s(il);
+        const uint32_t n_embd_k_gqa = hparams.n_embd_k_gqa(il);
 
         // Read type of key
         int32_t k_type_i_ref;
@@ -1629,7 +1629,7 @@ bool llama_kv_cache_unified::state_read_data(llama_io_read_i & io, uint32_t cell
         for (const auto & layer : layers) {
             const uint32_t il = layer.il;
 
-            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il) + hparams.n_embd_v_s(il);
+            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il);
 
             // Read type of value
             int32_t v_type_i_ref;
@@ -1659,7 +1659,7 @@ bool llama_kv_cache_unified::state_read_data(llama_io_read_i & io, uint32_t cell
         for (const auto & layer : layers) {
             const uint32_t il = layer.il;
 
-            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il) + hparams.n_embd_v_s(il);
+            const uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(il);
 
             // Read type of value
             int32_t v_type_i_ref;