From 948559260ae6cd79a60cdd999c16394ac342fb21 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sigbj=C3=B8rn=20Skj=C3=A6ret?= <sigbjorn.skjaeret@scala.com>
Date: Fri, 7 Jun 2024 12:27:57 +0200
Subject: [PATCH] Add Qwen2MoE 57B-A14B

---
 llama.cpp | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/llama.cpp b/llama.cpp
index 8b675ea99..7a901adb7 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1811,6 +1811,7 @@ enum e_model {
     MODEL_8x22B,
     MODEL_16x12B,
     MODEL_10B_128x3_66B,
+    MODEL_57B_A14B,
 };
 
 static const size_t kiB = 1024;
@@ -3933,6 +3934,7 @@ static const char * llama_model_type_name(e_model type) {
         case MODEL_8x22B:         return "8x22B";
         case MODEL_16x12B:        return "16x12B";
         case MODEL_10B_128x3_66B: return "10B+128x3.66B";
+        case MODEL_57B_A14B:      return "57B.A14B";
         default:                  return "?B";
     }
 }
@@ -4250,6 +4252,7 @@ static void llm_load_hparams(
                 ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
                 switch (hparams.n_layer) {
                     case 24: model.type = e_model::MODEL_A2_7B; break;
+                    case 28: model.type = e_model::MODEL_57B_A14B; break;
                     default: model.type = e_model::MODEL_UNKNOWN;
                 }
             } break;