Bug fixes for the 2023.1 AI Kit release (#1336)

* Intel Python Numpy Numba_dpes kNN sample (#1292)

* *.py and *.ipynb files with implementation
* README.md and sample.json files with documentation
* License and thir party programs

* Adding PyTorch Training Optimizations with AMX BF16 oneAPI sample (#1293)

* add IntelPytorch Quantization code samples (#1301)

* add IntelPytorch Quantization code samples

* fix the spelling error in the README file

* use john's README with grammar fix and title change

* Rename third-party-grograms.txt to third-party-programs.txt

Co-authored-by: Jimmy Wei <[email protected]>

* AMX bfloat16 mixed precision learning TensorFlow Transformer sample (#1317)

* [New Sample] Intel Extension for TensorFlow Getting Started (#1313)

* first draft

* Update README.md

* remove redunant file

* [New Sample] [oneDNN] Benchdnn tutorial  (#1315)

* New Sample: benchDNN tutorial

* Update readme: new sample

* Rename sample to benchdnn_tutorial

* Name fix

* Add files via upload (#1320)

* [New Sample] oneCCL Bindings for PyTorch Getting Started (#1316)

* Update README.md

* [New Sample] oneCCL Bindings for PyTorch Getting Started

* Update README.md

* add torch-ccl version check

* [New Sample] Intel Extension for PyTorch Getting Started (#1314)

* add new ipex GSG notebook for dGPU

* Update sample.json

for expertise field

* Update requirements.txt

Update package versions to comply with Snyk  tool

* Updated title field in sample.json in TF Transformer AMX bfloat16 Mixed Precision sample to fit within character length range (#1327)

* add arch checker class (#1332)

* change gpu.patch to convert the code samples from cpu to gpu correctly (#1334)

* Fixes for spelling in AMX bfloat16 transformer sample and printing error in python code in numpy vs numba sample (#1335)

* 2023.1 ai kit itex get started example fix (#1338)

* Fix the typo

* Update ResNet50_Inference.ipynb

* fix resnet inference demo link (#1339)

---------

Co-authored-by: krzeszew <[email protected]>
Co-authored-by: alexsin368 <[email protected]>
Co-authored-by: ZhaoqiongZ <[email protected]>
Co-authored-by: Louie Tsai <[email protected]>
Co-authored-by: Orel Yehuda <[email protected]>
Co-authored-by: yuning <[email protected]>
Co-authored-by: Wang, Kai Lawrence <[email protected]>

Author: 8 people

AI-and-Analytics/Features-and-Functionality/IntelPython_Numpy_Numba_dpex_kNN/IntelPython_Numpy_Numba_dpex_kNN.py

@@ -217,7 +217,7 @@ def knn_numba(X_train, y_train, X_test, k):
 predictions = knn_numba(X_train.values, y_train.values, X_test.values, 3)
 true_values = y_test.to_numpy()
 accuracy = np.mean(predictions == true_values)
-print('Numba accuracy:' + accuracy)
+print('Numba accuracy:', accuracy)
 
 
 # ## Numba_dpex k-NN
@@ -335,7 +335,7 @@ def knn_numba_dpex(train, train_labels, test, k, predictions, votes_to_classes_l
 
 true_values = y_test.to_numpy()
 accuracy = np.mean(predictions_numba == true_values)
-print('Numba_dpex accuracy:' + accuracy)
+print('Numba_dpex accuracy:', accuracy)
 
 
 # In[ ]:

AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_Transformer_AMX_bfloat16_MixedPrecision/IntelTensorFlow_Transformer_AMX_bfloat16_MixedPrecision.ipynb

@@ -20,7 +20,7 @@
    "id": "1bdd8c5b-209e-4a7d-9d98-2b552c987039",
    "metadata": {},
    "source": [
-    "# Tensorflow Transformer with AMX bfoat16 Mixed Precision Learning\n",
+    "# Tensorflow Transformer with AMX bfloat16 Mixed Precision Learning\n",
     "\n",
     "In this example we will be learning Transformer block for text classification using **IMBD dataset**. And then we will modify the code to use mixed precision learning with **bfloat16**. The example based on the [Text classification with Transformer Keras code example](https://keras.io/examples/nlp/text_classification_with_transformer/).\n",
     "\n",

AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_Transformer_AMX_bfloat16_MixedPrecision/README.md

@@ -1,4 +1,4 @@
-# `TensorFlow (TF) Transformer with Intel® Advanced Matrix Extensions (Intel® AMX) bfoat16 Mixed Precision Learning` 
+# `TensorFlow (TF) Transformer with Intel® Advanced Matrix Extensions (Intel® AMX) bfloat16 Mixed Precision Learning` 
 
 This sample code demonstrates optimizing a TensorFlow model with Intel® Advanced Matrix Extensions (Intel® AMX) using bfloat16 (Brain Floating Point) on  4th Gen Intel® Xeon® Scalable Processors (Sapphire Rapids).
 

AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_Transformer_AMX_bfloat16_MixedPrecision/sample.json

@@ -1,6 +1,6 @@
 {
   "guid": "60A68888-6099-414E-999B-EDC7310A01EA",
-  "name": "TensorFlow Transformer with Advanced Matrix Extensions bfoat16 Mixed Precision Learning",
+  "name": "TensorFlow Transformer with Advanced Matrix Extensions bfloat16 Mixed Precision Learning",
   "categories": ["Toolkit/oneAPI AI And Analytics/AI Getting Started Samples"],
   "description": "This sample code demonstrates optimizing a TensorFlow model with Intel® Advanced Matrix Extensions (Intel® AMX) using bfloat16 (Brain Floating Point) on Sapphire Rapids",
   "builder": ["cli"],

AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/ResNet50_Inference.ipynb

@@ -65,7 +65,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!wget https://raw.githubusercontent.com/intel/intel-extension-for-pytorch/master/examples/resnet50.py"
+    "!wget https://raw.githubusercontent.com/intel/intel-extension-for-pytorch/master/examples/cpu/inference/resnet50_general_inference_script.py"
    ]
   },
   {
@@ -111,7 +111,7 @@
     "source $ONEAPI_INSTALL/setvars.sh --force > /dev/null 2>&1\n",
     "source activate pytorch\n",
     "echo \"########## Executing the run\"\n",
-    "DNNL_VERBOSE=1 python resnet50.py > infer_rn50_cpu.csv\n",
+    "DNNL_VERBOSE=1 python resnet50_general_inference_script.py > infer_rn50_cpu.csv\n",
     "echo \"########## Done with the run\""
    ]
   },
@@ -175,7 +175,7 @@
     "source $ONEAPI_INSTALL/setvars.sh --force > /dev/null 2>&1\n",
     "source activate pytorch\n",
     "echo \"########## Executing the run\"\n",
-    "DNNL_VERBOSE=1 python resnet50.py > infer_rn50_gpu.csv\n",
+    "DNNL_VERBOSE=1 python resnet50_general_inference_script_gpu.py > infer_rn50_gpu.csv\n",
     "echo \"########## Done with the run\""
    ]
   },

AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/codes_for_ipynb/gpu.patch

@@ -1,13 +1,18 @@
---- resnet50.py	2023-01-24 14:40:21.570401244 +0900
-+++ resnet50_xpu.py	2023-01-24 14:43:31.170393594 +0900
-@@ -23,8 +23,8 @@
+diff --git a/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/resnet50_general_inference_script.py b/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/resnet50_general_inference_script_gpu.py
+similarity index 92%
+rename from AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/resnet50_general_inference_script.py
+rename to AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/resnet50_general_inference_script_gpu.py
+index dae594af..edd0fcb3 100644
+--- a/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/resnet50_general_inference_script.py
++++ b/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/resnet50_general_inference_script_gpu.py
+@@ -23,8 +23,8 @@ def main(args):
 
    import intel_extension_for_pytorch as ipex
 
 -  model = model.to(memory_format=torch.channels_last)
 -  data = data.to(memory_format=torch.channels_last)
-+  model = model.to("xpu", memory_format=torch.channels_last)
-+  data = data.to("xpu", memory_format=torch.channels_last)
++  model = model.to("xpu",memory_format=torch.channels_last)
++  data = data.to("xpu",memory_format=torch.channels_last)
 
    if args.dtype == 'float32':
      model = ipex.optimize(model, dtype=torch.float32)

AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/codes_for_py/gpu.patch

@@ -0,0 +1,32 @@
+diff --git a/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/Intel_Extension_For_PyTorch_Hello_World.py b/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/Intel_Extension_For_PyTorch_Hello_World.py
+index 00eb371b..a3ded045 100755
+--- a/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/Intel_Extension_For_PyTorch_Hello_World.py
++++ b/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted/Intel_Extension_For_PyTorch_Hello_World.py
+@@ -75,7 +75,7 @@ def main():
+     3. crite: Criterion function to minimize loss
+     '''
+     model = TestModel()
+-    model = model.to(memory_format=torch.channels_last)
++    model = model.to("xpu", memory_format=torch.channels_last)
+     optim = torch.optim.SGD(model.parameters(), lr=0.01)
+     crite = nn.MSELoss(reduction='sum')
+ 
+@@ -104,7 +104,8 @@ def main():
+         '''
+         model.train()
+         for batch_index, (data, y_ans) in enumerate(trainLoader):
+-            data = data.to(memory_format=torch.channels_last)
++            data = data.to("xpu", memory_format=torch.channels_last)
++            y_ans = y_ans.to("xpu", memory_format=torch.channels_last)
+             optim.zero_grad()
+             y = model(data)
+             loss = crite(y, y_ans)
+@@ -116,7 +117,7 @@ def main():
+         '''
+         model.eval()
+         for batch_index, data in enumerate(testLoader):
+-            data = data.to(memory_format=torch.channels_last)
++            data = data.to("xpu", memory_format=torch.channels_last)
+             y = model(data)
+ 
+ if __name__ == '__main__':

AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_TensorFlow_GettingStarted/README.md

@@ -38,7 +38,7 @@ Please follow bellow steps to setup GPU environment.
 3. Activate the created conda env:  ```$source activate user-tensorflow-gpu ```
 4. Install the required packages:  ```(user-tensorflow-gpu) $pip install tensorflow_hub ipykernel ```
 5. Deactivate conda env:  ```(user-tensorflow-gpu)$conda deactivate ```
-6. Register the kernel to Jupyter NB: ``` $~/.conda/envs/user-tensorflowgpu/bin/python  -m ipykernel install --user --name=user-tensorflow-gpu ```  
+6. Register the kernel to Jupyter NB: ``` $~/.conda/envs/user-tensorflow-gpu/bin/python  -m ipykernel install --user --name=user-tensorflow-gpu ```  
 
 Once users finish GPU environment setup, please do the same steps but remove "-gpu" from above steps. 
 In the end, you will have two new conda environments which are user-tensorflow-gpu and user-tensorflow

AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_TensorFlow_GettingStarted/ResNet50_Inference.ipynb

@@ -51,7 +51,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "%env ONEAPI_INSTALL=~/intel/oneapi"
+    "%env ONEAPI_INSTALL=/opt/intel/oneapi"
    ]
   },
   {
@@ -124,7 +124,7 @@
     "%%writefile run.sh\n",
     "#!/bin/bash\n",
     "source $ONEAPI_INSTALL/setvars.sh --force > /dev/null 2>&1\n",
-    "source activate tensorflow-gpu\n",
+    "source activate user-tensorflow-gpu\n",
     "echo \"########## Executing the run\"\n",
     "DNNL_VERBOSE=1 python infer_resnet50.py > infer_rn50_gpu.csv\n",
     "echo \"########## Done with the run\""
@@ -156,7 +156,7 @@
    "metadata": {},
    "source": [
     "#### Run on CPU via Intel TensorFlow\n",
-    "Users also can run the same infer_resnet50.py on CPU with intel tensorflow or stock tensorflow."
+    "Users also can run the same infer_resnet50.py on CPU with intel tensorflow or stock tensorflow. Please switch to the user-tensorflow jupyter kernel and execute again from prerequisites for CPU run"
    ]
   },
   {
@@ -168,7 +168,7 @@
     "%%writefile run.sh\n",
     "#!/bin/bash\n",
     "source $ONEAPI_INSTALL/setvars.sh --force > /dev/null 2>&1\n",
-    "source activate tensorflow\n",
+    "source activate user-tensorflow\n",
     "echo \"########## Executing the run\"\n",
     "DNNL_VERBOSE=1 python infer_resnet50.py > infer_rn50_cpu.csv\n",
     "echo \"########## Done with the run\""
@@ -269,7 +269,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "FdIndex=2"
+    "FdIndex=0"
    ]
   },
   {
@@ -325,7 +325,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "onednn.breakdown(data,\"arch\",\"time\")"
+    "onednn.breakdown(exec_data,\"arch\",\"time\")"
    ]
   },
   {
@@ -382,7 +382,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "os.chdir(initial_cwd)\n",
     "print('[CODE_SAMPLE_COMPLETED_SUCCESFULLY]')"
    ]
   }

AI-and-Analytics/version_check.py

@@ -11,6 +11,29 @@
 modin_found = util.find_spec("modin") is not None
 torchccl_found = util.find_spec("oneccl_bindings_for_pytorch") is not None
 
+class arch_checker:
+
+    def __init__(self):
+        cpuinfo_found = util.find_spec("cpuinfo") is not None
+        if cpuinfo_found == False:
+            self.arch = 'None'
+            print("please install py-cpuinfo")
+            return
+        from cpuinfo import get_cpu_info
+        info = get_cpu_info()
+        flags = info['flags']
+        arch_list = ['SPR', 'CPX',"ICX|CLX", "SKX", "BDW|CORE|ATOM"]
+        isa_list = [['amx_bf16', 'amx_int8', 'amx_tile'],['avx512_bf16'],['avx512_vnni'],['avx512'],['avx2']]
+        index = len(arch_list) - 1
+        for flag in flags:
+            for idx, isa_sublist in enumerate(isa_list):
+                for isa in isa_sublist:
+                    if isa in flag:
+                        if idx < index:
+                            index = idx
+        self.arch = arch_list[index]
+        return
+
 if tensorflow_found == True:
 
     import tensorflow as tf
@@ -85,3 +108,7 @@ def get_mkl_enabled_flag():
 if torchccl_found == True:
     import oneccl_bindings_for_pytorch as torchccl
     print("oneCCL Bindings version {}".format(torchccl.__version__))
+
+
+checker = arch_checker()
+print("Arch : ", checker.arch)