[Testdata] Remove unnecessary model testdata

This commit removes unnecessary model testdata.
This was already moved to another repo.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>

diff --git a/README.md b/README.md
index fb30c9f..0369b90 100644 (file)
--- a/README.md
+++ b/README.md
@@ -14,8 +14,7 @@ NPU_SystemService/    Root directory
 |-- tests/
 |   |-- unittests     Source files for Unit Tests
 |   |-- apptests      Source files for App Tests
-|   |-- utils         Source files for Test Utils
-|   `-- models        NPU Model binaries from second/third-party repos
+|   `-- utils         Source files for Test Utils
 |-- tools/            Useful tools or scripts
 |-- packaging/        Tizen Packaging Files (i.e., FastModel Simulator)
 |-- debian/           Debian Packaging Files (i.e., Ubuntu 16.04)

diff --git a/tests/models/README.md b/tests/models/README.md
deleted file mode 100644 (file)
index 63aa021..0000000
--- a/tests/models/README.md
+++ /dev/null
@@ -1,31 +0,0 @@
-## tvn models from [NPU_Compiler](https://github.sec.samsung.net/AIP/NPU_Compiler)
-
-The below describes a short summary of tvn models, generated by `./NPU_Compiler/tools/vcc/tests/explain.sh`.
-Note that it includes valid tvn models only.
-
-```
-NAME           SUMMARY
-000            Identity
-001            MaxPool
-NET_0000       MaxPool-MaxPool
-NET_0001       Placeholder-MaxPool-Conv2D-FakeQuant
-NET_0002       Placeholder-MaxPool--Conv2D------------Add-FakeQuant
-NET_0003       Convolution with excessive memory requirement (over 1MB)
-NET_0004       Padded Conv-BiasAdd-Relu with excessive memory requirement (over 1MB)
-NET_0006       First two residual blocks of RefineDet
-NET_0008       MaxPool-MaxPool network with large(1x40x20x512) input
-NET_0009       MaxPool network with large(1x320x320x64) input
-NET_0011       First Conv2DBackpropInput with constant network for input_sizes
-UNI_CONV_000   Conv-FakeQuantWithMinMaxVars
-UNI_CONV_001   Conv2D-BiasAdd
-UNI_CONV_002   Conv2D-BiasAdd-Relu
-UNI_CONV_003   Conv2D-FakeQuantWithMinMaxVars with non-zero weight
-UNI_CONV_004   Pad-Conv2D-BiasAdd-Relu-FakeQuantiWithMinMaxVars with non-zero weight
-UNI_CONV_005   Conv OFM is not a multiple of 64
-UNI_CONVE_000  Conv2D-BiasAdd-Add-Relu
-UNI_MAXP_000   MaxPool
-UNI_MAXP_001   Pad-MaxPool
-UNI_MAXP_002   Small MaxPool (input 2x2x64, filter 2x2, output 1x1x64)
-UNI_TCNV_000   Conv2DBackpropInput
-UNI_TCNV_001   Conv2DBackpropInput + BiasAdd
-```

diff --git a/tests/models/tvn_models.tar.gz b/tests/models/tvn_models.tar.gz
deleted file mode 100644 (file)
index dd6ba27..0000000
Binary files a/tests/models/tvn_models.tar.gz and /dev/null differ