Deepseek V3 support added (#176)

Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com>

1 files changed, 96 insertions, 4 deletions

diff --git a/src/llama.cpp b/src/llama.cpp
index b983c84b..29bd14af 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -106,7 +106,7 @@
 
 // bump if necessary
 #define LLAMA_MAX_LAYERS  512
-#define LLAMA_MAX_EXPERTS 160  // DeepSeekV2
+#define LLAMA_MAX_EXPERTS 256  // DeepSeekV2
 
 //
 // helpers
@@ -294,6 +294,8 @@ enum llm_kv {
     LLM_KV_EXPERT_USED_COUNT,
     LLM_KV_EXPERT_SHARED_COUNT,
     LLM_KV_EXPERT_WEIGHTS_SCALE,
+    LLM_KV_EXPERT_WEIGHTS_NORM,
+    LLM_KV_EXPERT_GATING_FUNC,
     LLM_KV_POOLING_TYPE,
     LLM_KV_LOGIT_SCALE,
     LLM_KV_DECODER_START_TOKEN_ID,
@@ -399,6 +401,8 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
     { LLM_KV_EXPERT_USED_COUNT,                 "%s.expert_used_count"                 },
     { LLM_KV_EXPERT_SHARED_COUNT,               "%s.expert_shared_count"               },
     { LLM_KV_EXPERT_WEIGHTS_SCALE,              "%s.expert_weights_scale"              },
+    { LLM_KV_EXPERT_WEIGHTS_NORM,               "%s.expert_weights_norm"               },
+    { LLM_KV_EXPERT_GATING_FUNC,                "%s.expert_gating_func"                },
     { LLM_KV_POOLING_TYPE ,                     "%s.pooling_type"                      },
     { LLM_KV_LOGIT_SCALE,                       "%s.logit_scale"                       },
     { LLM_KV_DECODER_START_TOKEN_ID,            "%s.decoder_start_token_id"            },
@@ -520,6 +524,7 @@ enum llm_tensor {
     LLM_TENSOR_FFN_DOWN_SHEXP,
     LLM_TENSOR_FFN_GATE_SHEXP,
     LLM_TENSOR_FFN_UP_SHEXP,
+    LLM_TENSOR_FFN_EXP_PROBS_B,
     LLM_TENSOR_ATTN_Q_NORM,
     LLM_TENSOR_ATTN_K_NORM,
     LLM_TENSOR_LAYER_OUT_NORM,
@@ -1211,6 +1216,7 @@ static const std::map<llm_arch, std::map<llm_tensor, std::string>> LLM_TENSOR_NA
             { LLM_TENSOR_FFN_GATE_SHEXP,     "blk.%d.ffn_gate_shexp" },
             { LLM_TENSOR_FFN_DOWN_SHEXP,     "blk.%d.ffn_down_shexp" },
             { LLM_TENSOR_FFN_UP_SHEXP,       "blk.%d.ffn_up_shexp" },
+	    { LLM_TENSOR_FFN_EXP_PROBS_B,    "blk.%d.exp_probs_b" },
         },
     },
     {
@@ -2186,6 +2192,7 @@ enum e_model {
     MODEL_70B,
     MODEL_236B,
     MODEL_314B,
+    MODEL_671B,
     MODEL_SMALL,
     MODEL_MEDIUM,
     MODEL_LARGE,
@@ -2203,6 +2210,21 @@ static const size_t kiB = 1024;
 static const size_t MiB = 1024*kiB;
 static const size_t GiB = 1024*MiB;
 
+enum llm_expert_gating_func_type {
+    LLM_EXPERT_GATING_FUNC_SOFTMAX = 1,
+    LLM_EXPERT_GATING_FUNC_SIGMOID = 2,
+};
+
+static const char * llama_expert_gating_func_name(llm_expert_gating_func_type type) {
+    switch (type) {
+        case LLM_EXPERT_GATING_FUNC_SOFTMAX: return "softmax";
+        case LLM_EXPERT_GATING_FUNC_SIGMOID: return "sigmoid";
+        default:                             return "unknown";
+    }
+}
+
+
+
 struct llama_hparams {
     bool vocab_only;
     bool rope_finetuned;
@@ -2232,6 +2254,8 @@ struct llama_hparams {
     uint32_t n_ff_shexp = 0;
     uint32_t n_expert_shared = 0;
     float    expert_weights_scale = 0.0;
+    bool     expert_weights_norm = false;
+    uint32_t expert_gating_func = LLM_EXPERT_GATING_FUNC_SOFTMAX;
 
     float f_norm_eps;
     float f_norm_rms_eps;
@@ -2502,6 +2526,7 @@ struct llama_layer {
     struct ggml_tensor * ffn_down_b = nullptr; // b2
     struct ggml_tensor * ffn_up_b   = nullptr; // b3
     struct ggml_tensor * ffn_act;
+    struct ggml_tensor * ffn_exp_probs_b = nullptr;
 
     // mamba proj
     struct ggml_tensor * ssm_in;
@@ -4677,6 +4702,7 @@ static const char * llama_model_type_name(e_model type) {
         case MODEL_70B:           return "70B";
         case MODEL_236B:          return "236B";
         case MODEL_314B:          return "314B";
+        case MODEL_671B:          return "671B";
         case MODEL_SMALL:         return "0.1B";
         case MODEL_MEDIUM:        return "0.4B";
         case MODEL_LARGE:         return "0.8B";
@@ -5302,11 +5328,19 @@ static void llm_load_hparams(
                 ml.get_key(LLM_KV_EXPERT_FEED_FORWARD_LENGTH, hparams.n_ff_exp);
                 ml.get_key(LLM_KV_EXPERT_SHARED_COUNT, hparams.n_expert_shared);
                 ml.get_key(LLM_KV_EXPERT_WEIGHTS_SCALE, hparams.expert_weights_scale);
+                ml.get_key(LLM_KV_EXPERT_WEIGHTS_NORM, hparams.expert_weights_norm, false);
+                ml.get_key(LLM_KV_EXPERT_GATING_FUNC, hparams.expert_gating_func, false);
+	        if (hparams.expert_gating_func == 0) {
+                    // for compatibility with existing DeepSeek V2 and V2.5 GGUFs
+                    // that have no expert_gating_func model parameter set
+                    hparams.expert_gating_func = LLM_EXPERT_GATING_FUNC_SOFTMAX;
+                }
                 ml.get_key(LLM_KV_ROPE_SCALING_YARN_LOG_MUL, hparams.rope_yarn_log_mul);
 
                 switch (hparams.n_layer) {
                     case 27: model.type = e_model::MODEL_16B; break;
                     case 60: model.type = e_model::MODEL_236B; break;
+                    case 61: model.type = e_model::MODEL_671B; break;
                     default: model.type = e_model::MODEL_UNKNOWN;
                 }
             } break;
@@ -5566,6 +5600,10 @@ static void llm_load_vocab(
                 vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_CODER;
                 vocab.tokenizer_clean_spaces = false;
             } else if (
+                    tokenizer_pre == "deepseek-v3") {
+                vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEEPSEEK3_LLM;
+                vocab.tokenizer_clean_spaces = false;
+            } else if (
                     tokenizer_pre == "falcon") {
                 vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_FALCON;
             } else if (
@@ -6075,6 +6113,8 @@ static void llm_load_print_meta(llama_model_loader & ml, llama_model & model) {
         LLAMA_LOG_INFO("%s: n_ff_exp             = %d\n",     __func__, hparams.n_ff_exp);
         LLAMA_LOG_INFO("%s: n_expert_shared      = %d\n",     __func__, hparams.n_expert_shared);
         LLAMA_LOG_INFO("%s: expert_weights_scale = %.1f\n",   __func__, hparams.expert_weights_scale);
+        LLAMA_LOG_INFO("%s: expert_weights_norm  = %d\n",     __func__, hparams.expert_weights_norm);
+        LLAMA_LOG_INFO("%s: expert_gating_func   = %s\n",     __func__, llama_expert_gating_func_name((enum llm_expert_gating_func_type) hparams.expert_gating_func));
         LLAMA_LOG_INFO("%s: rope_yarn_log_mul    = %.4f\n",   __func__, hparams.rope_yarn_log_mul);
     }
 
@@ -7540,6 +7580,7 @@ static bool llm_load_tensors(
                             layer.ffn_up   = ml.create_tensor(ctx_split, tn(LLM_TENSOR_FFN_UP,   "weight", i), {n_embd,   n_ff});
                         } else {
                             layer.ffn_gate_inp = ml.create_tensor(ctx_layer, tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), {n_embd, n_expert});
+                            layer.ffn_exp_probs_b = ml.create_tensor(ctx_layer, tn(LLM_TENSOR_FFN_EXP_PROBS_B, "bias", i), {n_expert} );
 
                             GGML_ASSERT(n_expert      > 0);
                             GGML_ASSERT(n_expert_used > 0);
@@ -8346,12 +8387,14 @@ static struct ggml_tensor * llm_build_moe_ffn(
          struct ggml_tensor * up_exps,
          struct ggml_tensor * gate_exps,
          struct ggml_tensor * down_exps,
+         struct ggml_tensor * exp_probs_b,
                     int64_t   n_expert,
                     int64_t   n_expert_used,
             llm_ffn_op_type   type_op,
                        bool   norm_w,
                        bool   scale_w,
                       float   w_scale,
+llm_expert_gating_func_type   gating_op,
          const llm_build_cb & cb,
                         int   il) {
     int64_t n_embd = cur->ne[0];
@@ -8360,11 +8403,32 @@ static struct ggml_tensor * llm_build_moe_ffn(
     ggml_tensor * logits = llm_build_lora_mm(lctx, ctx, gate_inp, cur); // [n_expert, n_tokens]
     cb(logits, "ffn_moe_logits", il);
 
-    ggml_tensor * probs = ggml_soft_max(ctx, logits); // [n_expert, n_tokens]
+    //ggml_tensor * probs = ggml_soft_max(ctx, logits); // [n_expert, n_tokens]
+    ggml_tensor * probs = nullptr;
+    switch (gating_op) {
+        case LLM_EXPERT_GATING_FUNC_SOFTMAX:
+            {
+                probs = ggml_soft_max(ctx, logits); // [n_expert, n_tokens]
+            } break;
+        case LLM_EXPERT_GATING_FUNC_SIGMOID:
+            {
+                probs = ggml_sigmoid(ctx, logits); // [n_expert, n_tokens]
+            } break;
+        default:
+            GGML_ABORT("fatal error");
+    }
     cb(probs, "ffn_moe_probs", il);
 
+    // add experts selection bias - introduced in DeepSeek V3
+    // leave probs unbiased as it's later used to get expert weights
+    ggml_tensor * selection_probs = probs;
+    if (exp_probs_b != nullptr) {
+        selection_probs = ggml_add(ctx, probs, exp_probs_b);
+        cb(selection_probs, "ffn_moe_probs_biased", il);
+    }
+
     // select experts
-    ggml_tensor * selected_experts = ggml_top_k(ctx, probs, n_expert_used); // [n_expert_used, n_tokens]
+    ggml_tensor * selected_experts = ggml_top_k(ctx, selection_probs, n_expert_used); // [n_expert_used, n_tokens]
     cb(selected_experts->src[0], "ffn_moe_argsort", il);
     cb(selected_experts, "ffn_moe_topk", il);
 
@@ -9180,9 +9244,11 @@ struct llm_build_context {
                         model.layers[il].ffn_up_exps,
                         model.layers[il].ffn_gate_exps,
                         model.layers[il].ffn_down_exps,
+			nullptr,
                         n_expert, n_expert_used,
                         LLM_FFN_SILU, true,
                         false, 0.0,
+			LLM_EXPERT_GATING_FUNC_SOFTMAX,
                         cb, il);
                 cb(cur, "ffn_moe_out", il);
             }
@@ -9673,9 +9739,11 @@ struct llm_build_context {
                     model.layers[il].ffn_up_exps,
                     model.layers[il].ffn_gate_exps,
                     model.layers[il].ffn_down_exps,
+		    nullptr,
                     n_expert, n_expert_used,
                     LLM_FFN_GELU, true,
                     false, 0.0,
+		    LLM_EXPERT_GATING_FUNC_SOFTMAX,
                     cb, il);
             cb(cur, "ffn_moe_out", il);
 
@@ -9814,9 +9882,11 @@ struct llm_build_context {
                     model.layers[il].ffn_up_exps,
                     model.layers[il].ffn_gate_exps,
                     model.layers[il].ffn_down_exps,
+		    nullptr,
                     n_expert, n_expert_used,
                     LLM_FFN_SILU, true,
                     false, 0.0,
+		    LLM_EXPERT_GATING_FUNC_SOFTMAX,
                     cb, il);
             cb(cur, "ffn_moe_out", il);
 
@@ -10944,9 +11014,11 @@ struct llm_build_context {
                         model.layers[il].ffn_up_exps,
                         model.layers[il].ffn_gate_exps,
                         model.layers[il].ffn_down_exps,
+			nullptr,
                         n_expert, n_expert_used,
                         LLM_FFN_SILU, false,
                         false, 0.0,
+			LLM_EXPERT_GATING_FUNC_SOFTMAX,
                         cb, il);
             cb(cur, "ffn_moe_out", il);
 
@@ -13109,9 +13181,11 @@ struct llm_build_context {
                     model.layers[il].ffn_up_exps,
                     model.layers[il].ffn_gate_exps,
                     model.layers[il].ffn_down_exps,
+		    nullptr,
                     n_expert, n_expert_used,
                     LLM_FFN_SILU, true,
                     false, 0.0,
+		    LLM_EXPERT_GATING_FUNC_SOFTMAX,
                     cb, il);
             cb(cur, "ffn_moe_out", il);
 
@@ -13324,9 +13398,11 @@ struct llm_build_context {
                             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, false,
+                            LLM_FFN_SILU, hparams.expert_weights_norm,
                             true, hparams.expert_weights_scale,
+			    (enum llm_expert_gating_func_type) hparams.expert_gating_func,
                             cb, il);
                 cb(moe_out, "ffn_moe_out", il);
 
@@ -18547,6 +18623,7 @@ struct llama_data_read {
                 read_to(&n_seq_id, sizeof(n_seq_id));
 
                 if (n_seq_id != 0) {
+		    llama_batch_free(batch);
                     LLAMA_LOG_ERROR("%s: invalid seq_id-agnostic kv cell\n", __func__);
                     return false;
                 }
@@ -19732,6 +19809,21 @@ static int32_t llama_chat_apply_template_internal(
         if (add_ass) {
             ss << "Assistant:";
         }
+    } else if (tmpl == "deepseek3" || tmpl_contains(LU8("'<｜Assistant｜>' + message['content'] + '<｜end▁of▁sentence｜>'"))) {
+        // DeepSeek-V3
+        for (auto message : chat) {
+            std::string role(message->role);
+            if (role == "system") {
+                ss << message->content << "\n\n";
+            } else if (role == "user") {
+                ss << LU8("<｜User｜>") << message->content;
+            } else if (role == "assistant") {
+                ss << LU8("<｜Assistant｜>") << message->content << LU8("<｜end▁of▁sentence｜>");
+            }
+        }
+        if (add_ass) {
+            ss << LU8("<｜Assistant｜>");
+        }
     } else {
         // template not supported
         return -1;