fix: resolve remaining merge conflicts and remove duplicate case statements

Author: mblackbourne

src/llama-model.cpp

@@ -1099,17 +1099,6 @@ void llama_model::load_hparams(llama_model_loader & ml) {
                     ? 1.0f / std::sqrt(float(hparams.n_embd / hparams.n_head(0)))
                     : 1.0f / std::sqrt(float(hparams.n_embd_head_k));
             } break;
-        case LLM_ARCH_GEMMA_EMBEDDING:
-            {
-                // EmbeddingGemma is an embedding model based on GEMMA architecture
-                ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
-                ml.get_key(LLM_KV_POOLING_TYPE, hparams.pooling_type);
-                ml.get_key(LLM_KV_ATTENTION_CAUSAL, hparams.causal_attn, false);
-
-                // Set embedding-specific defaults
-                hparams.causal_attn = false; // Embeddings use bidirectional attention
-                type = LLM_TYPE_300M; // EmbeddingGemma is 300M params
-            } break;
         case LLM_ARCH_GEMMA3:
             {
                 hparams.swa_type = LLAMA_SWA_TYPE_STANDARD;
@@ -3515,36 +3504,6 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
                         layer.ffn_post_norm = create_tensor(tn(LLM_TENSOR_FFN_POST_NORM, "weight", i), {n_embd}, 0);
                     }
                 } break;
-            case LLM_ARCH_GEMMA_EMBEDDING:
-                {
-                    // EmbeddingGemma uses similar structure to GEMMA3
-                    tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
-
-                    // output norm for embeddings
-                    output_norm = create_tensor(tn(LLM_TENSOR_OUTPUT_NORM, "weight"), {n_embd}, 0);
-
-                    // layers (similar to GEMMA3 but for embeddings)
-                    for (int i = 0; i < n_layer; ++i) {
-                        auto & layer = layers[i];
-
-                        layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, 0);
-
-                        layer.wq = create_tensor(tn(LLM_TENSOR_ATTN_Q,   "weight", i), {n_embd, n_embd_head_k * n_head}, 0);
-                        layer.wk = create_tensor(tn(LLM_TENSOR_ATTN_K,   "weight", i), {n_embd, n_embd_k_gqa}, 0);
-                        layer.wv = create_tensor(tn(LLM_TENSOR_ATTN_V,   "weight", i), {n_embd, n_embd_v_gqa}, 0);
-                        layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_k * n_head, n_embd}, 0);
-
-                        layer.attn_post_norm = create_tensor(tn(LLM_TENSOR_ATTN_POST_NORM, "weight", i), {n_embd}, 0);
-                        layer.attn_k_norm    = create_tensor(tn(LLM_TENSOR_ATTN_K_NORM,    "weight", i), {n_embd_head_k}, 0);
-                        layer.attn_q_norm    = create_tensor(tn(LLM_TENSOR_ATTN_Q_NORM,    "weight", i), {n_embd_head_k}, 0);
-
-                        layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, 0);
-                        layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd, n_ff}, 0);
-                        layer.ffn_up   = create_tensor(tn(LLM_TENSOR_FFN_UP,   "weight", i), {n_embd, n_ff}, 0);
-                        layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {n_ff, n_embd}, 0);
-                        layer.ffn_post_norm = create_tensor(tn(LLM_TENSOR_FFN_POST_NORM, "weight", i), {n_embd}, 0);
-                    }
-                } break;
             case LLM_ARCH_GEMMA3:
             case LLM_ARCH_GEMMA_EMBEDDING:
                 {
@@ -10644,8 +10603,6 @@ struct llm_build_gemma3_iswa : public llm_graph_context {
 struct llm_build_gemma_embedding : public llm_graph_context {
     llm_build_gemma_embedding(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
         const int64_t n_embd_head = hparams.n_embd_head_k;
-        const int64_t n_embd_k_gqa = hparams.n_embd_k_gqa();
-        const int64_t n_embd_v_gqa = hparams.n_embd_v_gqa();
 
         ggml_tensor * cur;
         ggml_tensor * inpL;
@@ -19008,18 +18965,14 @@ ggml_cgraph * llama_model::build_graph(const llm_graph_params & params) const {
             {
                 llm = std::make_unique<llm_build_gemma3_iswa>(*this, params);
             } break;
-        case LLM_ARCH_GEMMA_EMBEDDING:
-            {
-                // EmbeddingGemma uses custom embedding builder
-                llm = std::make_unique<llm_build_gemma_embedding>(*this, params);
-            } break;
         case LLM_ARCH_GEMMA3N:
             {
                 llm = std::make_unique<llm_build_gemma3n_iswa>(*this, params);
             } break;
         case LLM_ARCH_GEMMA_EMBEDDING:
             {
-                llm = std::make_unique<llm_build_gemma_embedding_iswa>(*this, params);
+                // EmbeddingGemma uses custom embedding builder
+                llm = std::make_unique<llm_build_gemma_embedding>(*this, params);
             } break;
         case LLM_ARCH_STARCODER2:
             {
@@ -19420,7 +19373,6 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
         case LLM_ARCH_GEMMA:
         case LLM_ARCH_GEMMA2:
         case LLM_ARCH_GEMMA3:
-        case LLM_ARCH_GEMMA_EMBEDDING:
         case LLM_ARCH_GEMMA3N:
         case LLM_ARCH_GEMMA_EMBEDDING:
         case LLM_ARCH_STARCODER2: