Update README.md

README.md

@@ -5,9 +5,9 @@
 
 Netron is a viewer for neural network, deep learning and machine learning models.
 
-Netron supports ONNX, TensorFlow Lite, Core ML, Keras, Caffe, Darknet, PyTorch, TensorFlow.js, Safetensors and NumPy.
+Netron supports ONNX, TensorFlow Lite, PyTorch, torch.export, ExecuTorch, Core ML, Keras, Caffe, Darknet, TensorFlow.js, Safetensors and NumPy.
 
-Netron has experimental support for TorchScript, torch.export, ExecuTorch, TensorFlow, OpenVINO, RKNN, ncnn, MNN, PaddlePaddle, GGUF and scikit-learn.
+Netron has experimental support for TorchScript, MLIR, TensorFlow, OpenVINO, RKNN, ncnn, MNN, PaddlePaddle, GGUF and scikit-learn.
 
 <p align='center'><a href='https://www.lutzroeder.com/ai'><img src='.github/screenshot.png' width='800'></a></p>
 
@@ -32,5 +32,8 @@ Sample model files to download or open using the browser version:
  * **TensorFlow Lite**: [yamnet](https://huggingface.co/thelou1s/yamnet/resolve/main/lite-model_yamnet_tflite_1.tflite) [[open](https://netron.app?url=https://huggingface.co/thelou1s/yamnet/blob/main/lite-model_yamnet_tflite_1.tflite)]
  * **TensorFlow**: [chessbot](https://github.com/srom/chessbot/raw/master/model/chessbot.pb) [[open](https://netron.app?url=https://github.com/srom/chessbot/raw/master/model/chessbot.pb)]
  * **Keras**: [mobilenet](https://github.com/aio-libs/aiohttp-demos/raw/master/demos/imagetagger/tests/data/mobilenet.h5) [[open](https://netron.app?url=https://github.com/aio-libs/aiohttp-demos/raw/master/demos/imagetagger/tests/data/mobilenet.h5)]
+
+* **MLIR**: [edge_detection](https://github.com/iree-org/iree/raw/main/tests/e2e/stablehlo_models/edge_detection.mlir) [[open](https://netron.app?url=https://github.com/iree-org/iree/blob/main/tests/e2e/stablehlo_models/edge_detection.mlir)]
+
  * **Core ML**: [exermote](https://github.com/Lausbert/Exermote/raw/master/ExermoteInference/ExermoteCoreML/ExermoteCoreML/Model/Exermote.mlmodel) [[open](https://netron.app?url=https://github.com/Lausbert/Exermote/raw/master/ExermoteInference/ExermoteCoreML/ExermoteCoreML/Model/Exermote.mlmodel)]
  * **Darknet**: [yolo](https://github.com/AlexeyAB/darknet/raw/master/cfg/yolo.cfg) [[open](https://netron.app?url=https://github.com/AlexeyAB/darknet/raw/master/cfg/yolo.cfg)]

edge_detection.mlir

@@ -0,0 +1,24 @@
+// RUN: iree-run-mlir --Xcompiler,iree-input-type=stablehlo --Xcompiler,iree-hal-target-device=local --Xcompiler,iree-hal-local-target-device-backends=vmvx %s --input=1x128x128x1xf32 | FileCheck %s
+// RUN: iree-run-mlir --Xcompiler,iree-input-type=stablehlo --Xcompiler,iree-hal-target-device=local --Xcompiler,iree-hal-local-target-device-backends=llvm-cpu %s --input=1x128x128x1xf32 | FileCheck %s
+
+// Image edge detection module generated by.
+// https://github.com/iree-org/iree/blob/main/samples/colab/tensorflow_edge_detection.ipynb.
+//
+// Input : a single 128x128 pixel image as a tensor<1x128x128x1xf32>, with pixels in [0.0, 1.0]
+// Output: a single image in the same format after running edge detection
+
+module {
+  // CHECK-LABEL: EXEC @edge_detect_sobel_operator
+  func.func @edge_detect_sobel_operator(%arg0: tensor<1x128x128x1xf32>) -> tensor<1x128x128x1xf32> {
+    %0 = stablehlo.constant dense<[[[[-1.000000e+00]], [[0.000000e+00]], [[1.000000e+00]]], [[[-2.000000e+00]], [[0.000000e+00]], [[2.000000e+00]]], [[[-1.000000e+00]], [[0.000000e+00]], [[1.000000e+00]]]]> : tensor<3x3x1x1xf32>
+    %1 = stablehlo.constant dense<[[[[1.000000e+00]], [[2.000000e+00]], [[1.000000e+00]]], [[[0.000000e+00]], [[0.000000e+00]], [[0.000000e+00]]], [[[-1.000000e+00]], [[-2.000000e+00]], [[-1.000000e+00]]]]> : tensor<3x3x1x1xf32>
+    %2 = stablehlo.convolution(%arg0, %0) dim_numbers = [b, 0, 1, f]x[0, 1, i, o]->[b, 0, 1, f], window = {stride = [1, 1], pad = [[1, 1], [1, 1]], rhs_dilate = [1, 1]} {batch_group_count = 1 : i64, feature_group_count = 1 : i64} : (tensor<1x128x128x1xf32>, tensor<3x3x1x1xf32>) -> tensor<1x128x128x1xf32>
+    %3 = stablehlo.multiply %2, %2 : tensor<1x128x128x1xf32>
+    %4 = stablehlo.convolution(%arg0, %1) dim_numbers = [b, 0, 1, f]x[0, 1, i, o]->[b, 0, 1, f], window = {stride = [1, 1], pad = [[1, 1], [1, 1]], rhs_dilate = [1, 1]} {batch_group_count = 1 : i64, feature_group_count = 1 : i64} : (tensor<1x128x128x1xf32>, tensor<3x3x1x1xf32>) -> tensor<1x128x128x1xf32>
+    %5 = stablehlo.multiply %4, %4 : tensor<1x128x128x1xf32>
+    %6 = stablehlo.add %3, %5 : tensor<1x128x128x1xf32>
+    %7 = stablehlo.sqrt %6 : tensor<1x128x128x1xf32>
+    return %7 : tensor<1x128x128x1xf32>
+  }
+  // CHECK: 1x128x128x1xf32=
+}