From d218c61e6d33bd06f0ddb4c98a0b688f64c864bd Mon Sep 17 00:00:00 2001
From: yumengbo <yumeng.bo@intel.com>
Date: Thu, 19 Dec 2024 03:37:38 +0800
Subject: [PATCH] Support Softmax op

---
 ggml/src/ggml-openvino.cpp              | 17 +++++++++++++++++
 ggml/src/ggml-openvino/ggml-decoder.cpp |  6 ++++--
 2 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/ggml/src/ggml-openvino.cpp b/ggml/src/ggml-openvino.cpp
index 1fede40c4a..771ca86d02 100644
--- a/ggml/src/ggml-openvino.cpp
+++ b/ggml/src/ggml-openvino.cpp
@@ -642,11 +642,28 @@ static const std::set<std::string>& openvino_ops = []() -> const std::set<std::s
             return true;
         case GGML_OP_MUL_MAT:
             return false;
+        case GGML_OP_SOFT_MAX:
+            return true;
         case GGML_OP_UNARY:
             switch (ggml_get_unary_op(op))
             {
                 case GGML_UNARY_OP_SILU:
                     return true;
+                case GGML_UNARY_OP_ABS: 
+                case GGML_UNARY_OP_SGN:
+                case GGML_UNARY_OP_NEG:
+                case GGML_UNARY_OP_STEP:
+                case GGML_UNARY_OP_TANH:
+                case GGML_UNARY_OP_ELU:
+                case GGML_UNARY_OP_RELU:
+                case GGML_UNARY_OP_SIGMOID:
+                case GGML_UNARY_OP_GELU:
+                case GGML_UNARY_OP_GELU_QUICK:
+                case GGML_UNARY_OP_HARDSWISH:
+                case GGML_UNARY_OP_HARDSIGMOID:
+                case GGML_UNARY_OP_EXP:
+                case GGML_UNARY_OP_COUNT:
+                    return false;
             }
             return false; 
         default:
diff --git a/ggml/src/ggml-openvino/ggml-decoder.cpp b/ggml/src/ggml-openvino/ggml-decoder.cpp
index 353961d45e..ee156bb995 100644
--- a/ggml/src/ggml-openvino/ggml-decoder.cpp
+++ b/ggml/src/ggml-openvino/ggml-decoder.cpp
@@ -47,11 +47,13 @@ void GgmlOvDecoder::set_input_output(ggml_tensor* node, std::map<std::string, gg
         case GGML_OP_SOFT_MAX:
         {
             inputs[node->src[0]->name] = node->src[0];
-            inputs[node->src[1]->name] = node->src[1];
             outputs[node->name] = node;
             m_input_names.push_back(node->src[0]->name);
-            m_input_names.push_back(node->src[1]->name);
             m_output_names.push_back(node->name);
+            if (node->src[1]) {
+                inputs[node->src[1]->name] = node->src[1];
+                m_input_names.push_back(node->src[1]->name);
+            }
             break;
         } 
         // OPs with 3 inputs: