@fugood/llama.node 1.4.11 → 1.4.12

package/src/llama.cpp/src/models/bert.cpp

@@ -142,11 +142,13 @@ llm_build_bert::llm_build_bert(const llama_model & model, const llm_graph_params
                     LLM_FFN_GELU, LLM_FFN_SEQ, il);
             cb(cur, "ffn_out", il);
         } else if (model.arch == LLM_ARCH_JINA_BERT_V2) {
+            const bool up_contains_gate = !model.layers[il].ffn_gate && model.layers[il].ffn_up->ne[1] != hparams.n_ff();
+            auto type_op = up_contains_gate ? LLM_FFN_GEGLU : LLM_FFN_GELU;
             cur = build_ffn(cur,
-                    model.layers[il].ffn_up, NULL, NULL,
+                    model.layers[il].ffn_up, model.layers[il].ffn_up_b, NULL,
                     model.layers[il].ffn_gate, NULL, NULL,
                     model.layers[il].ffn_down, model.layers[il].ffn_down_b, NULL, NULL,
-                    model.layers[il].ffn_gate ? LLM_FFN_GELU : LLM_FFN_GEGLU, LLM_FFN_PAR, il);
+                    type_op, LLM_FFN_PAR, il);
             cb(cur, "ffn_out", il);
         } else {
             cur = build_ffn(cur,

package/src/llama.cpp/src/models/cogvlm.cpp

@@ -3,12 +3,14 @@
 llm_build_cogvlm::llm_build_cogvlm(const llama_model & model, const llm_graph_params & params) :
     llm_graph_context(params) {
     const int64_t n_embd_head = hparams.n_embd_head_v;
-    float         kq_scale    = 1.0f / sqrtf(float(n_embd_head));
+    const float   kq_scale    = 1.0f / sqrtf(float(n_embd_head));
 
     GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
     GGML_ASSERT(n_embd_head == hparams.n_rot);
 
-    ggml_tensor *inpL, *cur;
+    ggml_tensor * inpL;
+    ggml_tensor * cur;
+
     inpL = build_inp_embd(model.tok_embd);
 
     ggml_tensor * inp_pos = build_inp_pos();
@@ -44,7 +46,7 @@ llm_build_cogvlm::llm_build_cogvlm(const llama_model & model, const llm_graph_pa
         }
 
         ggml_tensor * inpSA = inpL;
-        cur                 = build_norm(inpSA, model.layers[il].attn_norm, NULL, LLM_NORM_RMS, il);
+        cur = build_norm(inpSA, model.layers[il].attn_norm, NULL, LLM_NORM_RMS, il);
 
         // build self attention
         {

package/src/llama.cpp/src/models/deepseek2.cpp

@@ -215,7 +215,7 @@ llm_build_deepseek2::llm_build_deepseek2(const llama_model & model, const llm_gr
                 model.layers[il].ffn_exp_probs_b,
                 n_expert, n_expert_used,
                 LLM_FFN_SILU, hparams.expert_weights_norm,
-                true, hparams.expert_weights_scale,
+                hparams.expert_weights_scale, hparams.expert_weights_scale,
                 (llama_expert_gating_func_type) hparams.expert_gating_func,
                 il);
             cb(moe_out, "ffn_moe_out", il);

package/src/llama.cpp/src/models/gemma-embedding.cpp

@@ -1,7 +1,5 @@
 #include "models.h"
 
-
-
 llm_build_gemma_embedding::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;
@@ -12,10 +10,8 @@ llm_build_gemma_embedding::llm_build_gemma_embedding(const llama_model & model,
     inpL = build_inp_embd(model.tok_embd);
 
     // important: do not normalize weights for raw embeddings input (i.e. encoded image emdeddings)
-    if (ubatch.token) {
-        inpL = ggml_scale(ctx0, inpL, sqrtf(n_embd));
-        cb(inpL, "inp_scaled", -1);
-    }
+    inpL = ggml_scale(ctx0, inpL, ubatch.token ? sqrtf(n_embd) : 1.0f);
+    cb(inpL, "inp_scaled", -1);
 
     // inp_pos - contains the positions
     ggml_tensor * inp_pos = build_inp_pos();

package/src/llama.cpp/src/models/gemma3.cpp

@@ -10,10 +10,9 @@ llm_build_gemma3<iswa>::llm_build_gemma3(const llama_model & model, const llm_gr
     inpL = build_inp_embd(model.tok_embd);
 
     // important: do not normalize weights for raw embeddings input (i.e. encoded image emdeddings)
-    if (ubatch.token) {
-        inpL = ggml_scale(ctx0, inpL, sqrtf(n_embd));
-        cb(inpL, "inp_scaled", -1);
-    }
+    inpL = ggml_scale(ctx0, inpL, ubatch.token ? sqrtf(n_embd) : 1.0f);
+    cb(inpL, "inp_scaled", -1);
+
     // inp_pos - contains the positions
     ggml_tensor * inp_pos = build_inp_pos();
 

package/src/llama.cpp/src/models/gemma3n-iswa.cpp

@@ -1,7 +1,5 @@
 #include "models.h"
 
-
-
 llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const llm_graph_params & params) :
     llm_graph_context(params),
     model(model),
@@ -15,10 +13,9 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
     inpL = build_inp_embd(model.tok_embd);
 
     // important: do not normalize weights for raw embeddings input (i.e. encoded image emdeddings)
-    if (ubatch.token) {
-        inpL = ggml_scale(ctx0, inpL, sqrtf(n_embd));
-        cb(inpL, "inp_scaled", -1);
-    }
+    inpL = ggml_scale(ctx0, inpL, ubatch.token ? sqrtf(n_embd) : 1.0f);
+    cb(inpL, "inp_scaled", -1);
+
     // inp_pos - contains the positions
     ggml_tensor * inp_pos = build_inp_pos();
 
@@ -248,7 +245,7 @@ ggml_tensor * llm_build_gemma3n_iswa::view_2d_slice(ggml_tensor * x, int idx) {
 // equivalent to get_per_layer_inputs() in python code
 // output shape: [n_embd_altup, n_layer, n_tokens]
 ggml_tensor * llm_build_gemma3n_iswa::get_per_layer_inputs() {
-    auto          inp = std::make_unique<llm_graph_input_embd>();
+    auto inp = std::make_unique<llm_graph_input_embd>();
     ggml_tensor * inp_per_layer;
     if (ubatch.token) {
         inp->tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, ubatch.n_tokens);

package/src/llama.cpp/src/models/llama.cpp

@@ -1,6 +1,7 @@
 #include "models.h"
 
-llm_build_llama::llm_build_llama(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
+template <bool embed>
+llm_build_llama<embed>::llm_build_llama(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
     const int64_t n_embd_head = hparams.n_embd_head_v;
 
     GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
@@ -14,7 +15,14 @@ llm_build_llama::llm_build_llama(const llama_model & model, const llm_graph_para
     // inp_pos - contains the positions
     ggml_tensor * inp_pos = build_inp_pos();
 
-    auto * inp_attn = build_attn_inp_kv();
+    using inp_attn_type = std::conditional_t<embed, llm_graph_input_attn_no_cache, llm_graph_input_attn_kv>;
+
+    inp_attn_type * inp_attn = nullptr;
+    if constexpr (embed) {
+        inp_attn = build_attn_inp_no_cache();
+    } else {
+        inp_attn = build_attn_inp_kv();
+    }
 
     const float kq_scale = hparams.f_attention_scale == 0.0f ? 1.0f/sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
 
@@ -145,11 +153,16 @@ llm_build_llama::llm_build_llama(const llama_model & model, const llm_graph_para
     cb(cur, "result_norm", -1);
     res->t_embd = cur;
 
-    // lm_head
-    cur = build_lora_mm(model.output, cur);
+    if constexpr (!embed) {
+        // lm_head
+        cur = build_lora_mm(model.output, cur);
 
-    cb(cur, "result_output", -1);
-    res->t_logits = cur;
+        cb(cur, "result_output", -1);
+        res->t_logits = cur;
+    }
 
     ggml_build_forward_expand(gf, cur);
 }
+
+template struct llm_build_llama<false>;
+template struct llm_build_llama<true>;

package/src/llama.cpp/src/models/maincoder.cpp

@@ -0,0 +1,117 @@
+#include "models.h"
+
+llm_build_maincoder::llm_build_maincoder(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
+    const int64_t n_embd_head = hparams.n_embd_head_v;
+
+    GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
+    GGML_ASSERT(n_embd_head == hparams.n_rot);
+
+    ggml_tensor * cur;
+    ggml_tensor * inpL;
+
+    inpL = build_inp_embd(model.tok_embd);
+
+    // inp_pos - contains the positions
+    ggml_tensor * inp_pos = build_inp_pos();
+
+    auto * inp_attn = build_attn_inp_kv();
+
+    ggml_tensor * inp_out_ids = build_inp_out_ids();
+
+    for (int il = 0; il < n_layer; ++il) {
+        ggml_tensor * inpSA = inpL;
+
+        // norm
+        cur = build_norm(inpL,
+                model.layers[il].attn_norm, NULL,
+                LLM_NORM_RMS, il);
+        cb(cur, "attn_norm", il);
+
+        // self-attention
+        {
+            // compute Q and K and RoPE them
+            ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
+            cb(Qcur, "Qcur", il);
+
+            ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
+            cb(Kcur, "Kcur", il);
+
+            ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
+            cb(Vcur, "Vcur", il);
+
+            Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head,    n_tokens);
+            Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
+            Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
+
+            Qcur = ggml_rope_ext(
+                    ctx0, Qcur, inp_pos, nullptr,
+                    n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
+                    ext_factor, attn_factor, beta_fast, beta_slow
+                    );
+
+            Kcur = ggml_rope_ext(
+                    ctx0, Kcur, inp_pos, nullptr,
+                    n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
+                    ext_factor, attn_factor, beta_fast, beta_slow
+                    );
+
+            Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
+            cb(Qcur, "Qcur_normed", il);
+
+            Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
+            cb(Kcur, "Kcur_normed", il);
+
+            cb(Qcur, "Qcur", il);
+            cb(Kcur, "Kcur", il);
+            cb(Vcur, "Vcur", il);
+
+            cur = build_attn(inp_attn,
+                    model.layers[il].wo, model.layers[il].bo,
+                    Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
+        }
+        if (il == n_layer - 1 && inp_out_ids) {
+            cur   = ggml_get_rows(ctx0,   cur, inp_out_ids);
+            inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
+        }
+        ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpSA);
+        cb(ffn_inp, "ffn_inp", il);
+
+        // feed-forward network
+        cur = build_norm(ffn_inp,
+                model.layers[il].ffn_norm, NULL,
+                LLM_NORM_RMS, il);
+        cb(cur, "ffn_norm", il);
+
+        cur = build_ffn(cur,
+                model.layers[il].ffn_up,   NULL, NULL,
+                model.layers[il].ffn_gate, NULL, NULL,
+                model.layers[il].ffn_down, NULL, NULL,
+                NULL,
+                LLM_FFN_SILU, LLM_FFN_PAR, il);
+        cb(cur, "ffn_out", il);
+
+        cur = ggml_add(ctx0, cur, ffn_inp);
+
+        cur = build_cvec(cur, il);
+        cb(cur, "l_out", il);
+
+        // input for next layer
+        inpL = cur;
+    }
+    cur = inpL;
+
+    cur = build_norm(cur,
+            model.output_norm, NULL,
+            LLM_NORM_RMS, -1);
+
+    cb(cur, "result_norm", -1);
+    res->t_embd = cur;
+
+    // lm_head
+    cur = build_lora_mm(model.output, cur);
+
+    cb(cur, "result_output", -1);
+    res->t_logits = cur;
+
+    ggml_build_forward_expand(gf, cur);
+}

package/src/llama.cpp/src/models/mimo2-iswa.cpp

@@ -0,0 +1,123 @@
+
+#include "models.h"
+
+llm_build_mimo2_iswa::llm_build_mimo2_iswa(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
+    ggml_tensor * cur;
+    ggml_tensor * inpL;
+
+    inpL = build_inp_embd(model.tok_embd);
+
+    ggml_tensor * inp_pos = build_inp_pos();
+    auto * inp_attn = build_attn_inp_kv_iswa();
+    ggml_tensor * inp_out_ids = build_inp_out_ids();
+
+    for (int il = 0; il < n_layer; ++il) {
+        ggml_tensor * inpSA = inpL;
+
+        uint32_t n_head_l    = hparams.n_head(il);
+        uint32_t n_head_kv_l = hparams.n_head_kv(il);
+        const float freq_base_l  = model.get_rope_freq_base(cparams, il);
+        const float freq_scale_l = model.get_rope_freq_scale(cparams, il);
+
+        cur = inpL;
+
+        // self_attention
+        {
+            cur = build_norm(inpL, model.layers[il].attn_norm, NULL, LLM_NORM_RMS, il);
+            cb(cur, "attn_norm", il);
+
+            // compute Q and K and RoPE them
+            ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
+            cb(Qcur, "Qcur", il);
+
+            ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
+            cb(Kcur, "Kcur", il);
+
+            ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
+            cb(Vcur, "Vcur", il);
+
+            Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head_k, n_head_l,    n_tokens);
+            Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head_k, n_head_kv_l, n_tokens);
+            Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head_v, n_head_kv_l, n_tokens);
+
+            Qcur = ggml_rope_ext(
+                ctx0, Qcur, inp_pos, nullptr,
+                n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
+                ext_factor, attn_factor, beta_fast, beta_slow
+                );
+
+            Kcur = ggml_rope_ext(
+                ctx0, Kcur, inp_pos, nullptr,
+                n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
+                ext_factor, attn_factor, beta_fast, beta_slow
+                );
+
+            cb(Qcur, "Qcur", il);
+            cb(Kcur, "Kcur", il);
+            cb(Vcur, "Vcur", il);
+
+            ggml_tensor * sinks = model.layers[il].attn_sinks;
+
+            cur = build_attn(inp_attn,
+                    model.layers[il].wo, NULL,
+                    Qcur, Kcur, Vcur, nullptr, sinks, nullptr, 1.0f/sqrtf(float(n_embd_head_k)), il);
+        }
+
+        if (il == n_layer - 1 && inp_out_ids) {
+            cur   = ggml_get_rows(ctx0,   cur, inp_out_ids);
+            inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
+        }
+
+        ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpSA);
+        cb(ffn_inp, "ffn_inp", il);
+
+        cur = build_norm(ffn_inp,
+                model.layers[il].ffn_norm, NULL,
+                LLM_NORM_RMS, il);
+        cb(cur, "ffn_norm", il);
+
+        // feed-forward network
+        if (model.layers[il].ffn_gate_inp == nullptr) {
+            // dense branch
+            cur = build_ffn(cur,
+                    model.layers[il].ffn_up,   model.layers[il].ffn_up_b,   NULL,
+                    model.layers[il].ffn_gate, model.layers[il].ffn_gate_b, NULL,
+                    model.layers[il].ffn_down, model.layers[il].ffn_down_b, NULL,
+                    NULL,
+                    LLM_FFN_SILU, LLM_FFN_PAR, il);
+            cb(cur, "ffn_out", il);
+        } else {
+            // MoE branch
+            cur = build_moe_ffn(cur, model.layers[il].ffn_gate_inp, model.layers[il].ffn_up_exps,
+                                model.layers[il].ffn_gate_exps, model.layers[il].ffn_down_exps,
+                                model.layers[il].ffn_exp_probs_b, n_expert, n_expert_used, LLM_FFN_SILU, true, false,
+                                0.0, LLAMA_EXPERT_GATING_FUNC_TYPE_SIGMOID, il);
+            cb(cur, "ffn_moe_out", il);
+        }
+
+        cur = ggml_add(ctx0, cur, ffn_inp);
+
+        cur = build_cvec(cur, il);
+        cb(cur, "l_out", il);
+
+        // input for next layer
+        inpL = cur;
+    }
+
+    cur = inpL;
+
+    cur = build_norm(cur,
+            model.output_norm, NULL,
+            LLM_NORM_RMS, -1);
+
+    cb(cur, "result_norm", -1);
+    res->t_embd = cur;
+
+    // lm_head
+    cur = build_lora_mm(model.output, cur);
+
+    cb(cur, "result_output", -1);
+    res->t_logits = cur;
+
+    ggml_build_forward_expand(gf, cur);
+}

package/src/llama.cpp/src/models/models.h

@@ -303,6 +303,7 @@ struct llm_build_llada_moe : public llm_graph_context {
     llm_build_llada_moe(const llama_model & model, const llm_graph_params & params);
 };
 
+template <bool embed>
 struct llm_build_llama : public llm_graph_context {
     llm_build_llama(const llama_model & model, const llm_graph_params & params);
 };
@@ -311,10 +312,18 @@ struct llm_build_llama_iswa : public llm_graph_context {
     llm_build_llama_iswa(const llama_model & model, const llm_graph_params & params);
 };
 
+struct llm_build_maincoder : public llm_graph_context {
+    llm_build_maincoder(const llama_model & model, const llm_graph_params & params);
+};
+
 struct llm_build_mamba : public llm_graph_context_mamba {
     llm_build_mamba(const llama_model & model, const llm_graph_params & params);
 };
 
+struct llm_build_mimo2_iswa : public llm_graph_context {
+    llm_build_mimo2_iswa(const llama_model & model, const llm_graph_params & params);
+};
+
 struct llm_build_minicpm3 : public llm_graph_context {
     llm_build_minicpm3(const llama_model & model, const llm_graph_params & params);
 };
@@ -327,6 +336,10 @@ struct llm_build_mistral3 : public llm_graph_context {
     llm_build_mistral3(const llama_model & model, const llm_graph_params & params);
 };
 
+struct llm_build_modern_bert : public llm_graph_context {
+    llm_build_modern_bert(const llama_model & model, const llm_graph_params & params);
+};
+
 struct llm_build_mpt : public llm_graph_context {
     llm_build_mpt(const llama_model & model, const llm_graph_params & params);
 };
@@ -396,6 +409,11 @@ struct llm_build_plamo : public llm_graph_context {
     llm_build_plamo(const llama_model & model, const llm_graph_params & params);
 };
 
+template <bool iswa>
+struct llm_build_plamo3 : public llm_graph_context {
+    llm_build_plamo3(const llama_model & model, const llm_graph_params & params);
+};
+
 struct llm_build_plm : public llm_graph_context {
     llm_build_plm(const llama_model & model, const llm_graph_params & params);
 };

package/src/llama.cpp/src/models/modern-bert.cpp

@@ -0,0 +1,115 @@
+#include "models.h"
+
+llm_build_modern_bert::llm_build_modern_bert(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
+    const int64_t n_embd_head = hparams.n_embd_head_v;
+    const int64_t n_embd_gqa  = hparams.n_embd_v_gqa();
+
+    GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
+
+    ggml_tensor * cur;
+    ggml_tensor * inpL;
+    ggml_tensor * inp_pos = build_inp_pos();
+
+    // construct input embeddings (token, type, position)
+    inpL = build_inp_embd(model.tok_embd);
+    cb(inpL, "inp_embd", -1);
+
+    // embed layer norm
+    inpL = build_norm(inpL, model.tok_norm, nullptr, LLM_NORM, -1);
+    cb(inpL, "inp_norm", -1);
+
+    ggml_tensor * inp_out_ids = build_inp_out_ids();
+
+    auto * inp_attn = build_attn_inp_no_cache();
+
+    for (int il = 0; il < n_layer; ++il) {
+        float freq_base_l = model.get_rope_freq_base(cparams, il);
+
+        cur = inpL;
+
+        // attention layer norm
+        if (model.layers[il].attn_norm) {
+            cur = build_norm(inpL,
+                    model.layers[il].attn_norm, NULL,
+                    LLM_NORM, il);
+            cb(cur, "attn_norm", il);
+        }
+
+        // self attention
+        cur = build_lora_mm(model.layers[il].wqkv, cur);
+        cb(cur, "wqkv", il);
+
+        const size_t type_size = ggml_type_size(cur->type);
+
+        ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head,    n_tokens, n_embd_head*type_size, cur->nb[1], 0*type_size*(n_embd));
+        ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*type_size, cur->nb[1], 1*type_size*(n_embd));
+        ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*type_size, cur->nb[1], 1*type_size*(n_embd + n_embd_gqa));
+
+        // RoPE
+        Qcur = ggml_rope_ext(
+                ctx0, Qcur, inp_pos, nullptr,
+                n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale,
+                ext_factor, attn_factor, beta_fast, beta_slow
+                );
+
+        Kcur = ggml_rope_ext(
+                ctx0, Kcur, inp_pos, nullptr,
+                n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale,
+                ext_factor, attn_factor, beta_fast, beta_slow
+                );
+
+        cb(Qcur, "Qcur", il);
+        cb(Kcur, "Kcur", il);
+        cb(Vcur, "Vcur", il);
+
+        cur = build_attn(inp_attn,
+                    model.layers[il].wo, nullptr,
+                    Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
+        cb(cur, "kqv_out", il);
+
+        if (il == n_layer - 1 && inp_out_ids) {
+            cur  = ggml_get_rows(ctx0,  cur, inp_out_ids);
+            inpL = ggml_get_rows(ctx0, inpL, inp_out_ids);
+        }
+
+        // re-add the layer input
+        ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpL);
+        cb(ffn_inp, "ffn_inp", il);
+
+        // attention layer norm
+        cur = build_norm(ffn_inp,
+                model.layers[il].ffn_norm, NULL,
+                LLM_NORM, il);
+        cb(cur, "ffn_norm", il);
+
+        cur = build_ffn(cur,
+                model.layers[il].ffn_up,   NULL, NULL,
+                NULL,                      NULL, NULL,
+                model.layers[il].ffn_down, NULL, NULL,
+                NULL,
+                LLM_FFN_GEGLU, LLM_FFN_SEQ, il);
+
+        // attentions bypass the intermediate layer
+        cur = ggml_add(ctx0, cur, ffn_inp);
+
+        // input for next layer
+        inpL = cur;
+    }
+
+    cur = inpL;
+
+    cur = build_norm(cur,
+            model.output_norm, NULL,
+            LLM_NORM, -1);
+    cb(cur, "final_norm_out", -1);
+
+    if (hparams.pooling_type == LLAMA_POOLING_TYPE_CLS) {
+        // extracting cls token
+        cur = ggml_view_1d(ctx0, cur, hparams.n_embd, 0);
+        cb(cur, "cls_pooled_embd", -1);
+    }
+
+    cb(cur, "res_embd", -1);
+    res->t_embd = cur;
+    ggml_build_forward_expand(gf, cur);
+}