commit bf7edb619cd5cf0ae84342299abe7c27f3ba6e7d
author mathad01 <matthew.haddon@arm.com> Tue Apr 20 16:12:45 2021 +0100
committer Jan Eilers <jan.eilers@arm.com> Thu Apr 22 08:21:43 2021 +0000
tree aace117c1b6de99b257552be74b8d8dc2b39718d
parent e11e63d749b0909f13f9a39c8d34ef5523255170

IVGCVSW-5418 ExecuteNetwork test for MobileBERT

 * Removed check in TfLiteParser and Delegate that requires both weights
and biases to be constant or non-constant simultaneously
 * Updated TfLiteParser FullyConnected layer test to properly use
non-constant weights
 * MobileBERT Float32 model now runs on TfLiteParser

Signed-off-by: mathad01 <matthew.haddon@arm.com>
Change-Id: I1d75eea466caa90cd695ad353160362df2f69483

delegate/src/FullyConnected.hpp

diff --git a/delegate/src/FullyConnected.hpp b/delegate/src/FullyConnected.hpp
index 2b45c48..e94304f 100644
--- a/delegate/src/FullyConnected.hpp
+++ b/delegate/src/FullyConnected.hpp
@@ -77,15 +77,6 @@
         {
             return kTfLiteError;
         }
-        if ((isConstantWeights && !tflite::IsConstantTensor(&tfLiteBiasTensor))
-            || (!isConstantWeights && tflite::IsConstantTensor(&tfLiteBiasTensor)))
-        {
-            TF_LITE_MAYBE_KERNEL_LOG(
-                tfLiteContext,
-                "TfLiteArmnnDelegate: Weights and bias are not compatible"
-                " in operator #%d node #%d: ", operatorCode, nodeIndex);
-            return kTfLiteError;
-        }
         biasTensorInfo = GetTensorInfoForTfLiteTensor(tfLiteBiasTensor);
     }
     else

src/armnnTfLiteParser/TfLiteParser.cpp

diff --git a/src/armnnTfLiteParser/TfLiteParser.cpp b/src/armnnTfLiteParser/TfLiteParser.cpp
index 9b1fa90..5070d5b 100644
--- a/src/armnnTfLiteParser/TfLiteParser.cpp
+++ b/src/armnnTfLiteParser/TfLiteParser.cpp
@@ -2374,15 +2374,6 @@
 
     desc.m_ConstantWeights = IsConstTensor(inputs[1]);
 
-    // Either both weights and biases need to be inputs or both weights and biases need to be constant
-    if (inputs.size() == 3 && desc.m_ConstantWeights != IsConstTensor(inputs[2]))
-    {
-        throw ParseException(
-                fmt::format("Weights and bias are not compatible."
-                            "Node {}",
-                            CHECK_LOCATION().AsString()));
-    }
-
     auto inputTensorIndexes = AsUnsignedVector(GetInputTensorIds(m_Model, subgraphIndex, operatorIndex));
     std::vector<unsigned int> tensorIndexesToRegister = {inputTensorIndexes[0]};
     if (desc.m_ConstantWeights)
@@ -3600,7 +3591,15 @@
 bool TfLiteParserImpl::IsConstTensor(TensorRawPtr tensorPtr)
 {
     CHECK_TENSOR_PTR(tensorPtr);
-    return !tensorPtr->is_variable;
+    bool isConst = true;
+
+    auto buffer = GetBuffer(m_Model, tensorPtr->buffer);
+    if (buffer->data.size() == 0)
+    {
+        isConst = false;
+    }
+
+    return isConst;
 }
 
 

src/armnnTfLiteParser/test/FullyConnected.cpp

diff --git a/src/armnnTfLiteParser/test/FullyConnected.cpp b/src/armnnTfLiteParser/test/FullyConnected.cpp
index 333e17f..1ce1b2f 100644
--- a/src/armnnTfLiteParser/test/FullyConnected.cpp
+++ b/src/armnnTfLiteParser/test/FullyConnected.cpp
@@ -224,7 +224,7 @@
                       "is_variable": true
                     }, )";
 
-            biasBuffer = R"(,{ "data": [ 10, 0, 0, 0 ] } )";
+            biasBuffer = R"(,{ "data": [] } )";
             outputs = "3";
         }
         m_JsonString = R"(
@@ -250,7 +250,6 @@
                         "details_type": 0,
                         "quantized_dimension": 0
                       },
-                      "is_variable": false
                     },
                     {
                       "shape": )" + filterShape + R"(,
@@ -263,7 +262,6 @@
                         "details_type": 0,
                         "quantized_dimension": 0
                       },
-                      "is_variable": true
                     },
                     )" + biasTensor + R"(
                     {
@@ -281,7 +279,6 @@
                         "details_type": 0,
                         "quantized_dimension": 0
                       },
-                      "is_variable": false
                     }
                   ],
                   "inputs": )" + inputTensors + R"(,
@@ -309,7 +306,7 @@
                   "data": []
                 },
                 {
-                  "data": [ 2, 3, 4, 5 ]
+                  "data": []
                 }
                 )" + biasBuffer + R"(
               ]