update marker of examples.md (#1458)

Author: jingxu10

docs/tutorials/examples.md

@@ -35,7 +35,7 @@ output = model(data)
 
 #### Complete - Float32
 
-[//]: # (train_single_fp32_complete)
+[//]: # (marker_train_single_fp32_complete)
 ```
 import torch
 import torchvision
@@ -79,11 +79,11 @@ torch.save({
      'optimizer_state_dict': optimizer.state_dict(),
      }, 'checkpoint.pth')
 ```
-[//]: # (train_single_fp32_complete)
+[//]: # (marker_train_single_fp32_complete)
 
 #### Complete - BFloat16
 
-[//]: # (train_single_bf16_complete)
+[//]: # (marker_train_single_bf16_complete)
 ```
 import torch
 import torchvision
@@ -128,15 +128,15 @@ torch.save({
      'optimizer_state_dict': optimizer.state_dict(),
      }, 'checkpoint.pth')
 ```
-[//]: # (train_single_bf16_complete)
+[//]: # (marker_train_single_bf16_complete)
 
 ### Distributed Training
 
 Distributed training with PyTorch DDP is accelerated by oneAPI Collective Communications Library Bindings for Pytorch\* (oneCCL Bindings for Pytorch\*). The extension supports FP32 and BF16 data types. More detailed information and examples are available at its [Github repo](https://github.com/intel/torch-ccl).
 
 **Note:** When performing distributed training with BF16 data type, use oneCCL Bindings for Pytorch\*. Due to a PyTorch limitation, distributed training with BF16 data type with Intel® Extension for PyTorch\* is not supported.
 
-[//]: # (train_ddp_complete)
+[//]: # (marker_train_ddp_complete)
 ```
 import os
 import torch
@@ -194,7 +194,7 @@ torch.save({
      'optimizer_state_dict': optimizer.state_dict(),
      }, 'checkpoint.pth')
 ```
-[//]: # (train_ddp_complete)
+[//]: # (marker_train_ddp_complete)
 
 ## Inference
 
@@ -206,7 +206,7 @@ The `optimize` function of Intel® Extension for PyTorch\* applies optimizations
 
 ##### Resnet50
 
-[//]: # (inf_rn50_imp_fp32)
+[//]: # (marker_inf_rn50_imp_fp32)
 ```
 import torch
 import torchvision.models as models
@@ -223,11 +223,11 @@ model = ipex.optimize(model)
 with torch.no_grad():
   model(data)
 ```
-[//]: # (inf_rn50_imp_fp32)
+[//]: # (marker_inf_rn50_imp_fp32)
 
 ##### BERT
 
-[//]: # (inf_bert_imp_fp32)
+[//]: # (marker_inf_bert_imp_fp32)
 ```
 import torch
 from transformers import BertModel
@@ -248,15 +248,15 @@ model = ipex.optimize(model)
 with torch.no_grad():
   model(data)
 ```
-[//]: # (inf_bert_imp_fp32)
+[//]: # (marker_inf_bert_imp_fp32)
 
 #### TorchScript Mode
 
 We recommend you take advantage of Intel® Extension for PyTorch\* with [TorchScript](https://pytorch.org/docs/stable/jit.html) for further optimizations.
 
 ##### Resnet50
 
-[//]: # (inf_rn50_ts_fp32)
+[//]: # (marker_inf_rn50_ts_fp32)
 ```
 import torch
 import torchvision.models as models
@@ -277,11 +277,11 @@ with torch.no_grad():
 
   model(data)
 ```
-[//]: # (inf_rn50_ts_fp32)
+[//]: # (marker_inf_rn50_ts_fp32)
 
 ##### BERT
 
-[//]: # (inf_bert_ts_fp32)
+[//]: # (marker_inf_bert_ts_fp32)
 ```
 import torch
 from transformers import BertModel
@@ -306,13 +306,13 @@ with torch.no_grad():
 
   model(data)
 ```
-[//]: # (inf_bert_ts_fp32)
+[//]: # (marker_inf_bert_ts_fp32)
 
 #### TorchDynamo Mode (Experimental, _NEW feature from 2.0.0_)
 
 ##### Resnet50
 
-[//]: # (inf_rn50_dynamo_fp32)
+[//]: # (marker_inf_rn50_dynamo_fp32)
 ```
 import torch
 import torchvision.models as models
@@ -330,11 +330,11 @@ model = torch.compile(model, backend="ipex")
 with torch.no_grad():
     model(data)
 ```
-[//]: # (inf_rn50_dynamo_fp32)
+[//]: # (marker_inf_rn50_dynamo_fp32)
 
 ##### BERT
 
-[//]: # (inf_bert_dynamo_fp32)
+[//]: # (marker_inf_bert_dynamo_fp32)
 ```
 import torch
 from transformers import BertModel
@@ -356,7 +356,7 @@ model = torch.compile(model, backend="ipex")
 with torch.no_grad():
     model(data)
 ```
-[//]: # (inf_bert_dynamo_fp32)
+[//]: # (marker_inf_bert_dynamo_fp32)
 
 ### BFloat16
 
@@ -367,7 +367,7 @@ We recommend using Auto Mixed Precision (AMP) with BFloat16 data type.
 
 ##### Resnet50
 
-[//]: # (inf_rn50_imp_bf16)
+[//]: # (marker_inf_rn50_imp_bf16)
 ```
 import torch
 import torchvision.models as models
@@ -385,11 +385,11 @@ with torch.no_grad():
   with torch.cpu.amp.autocast():
     model(data)
 ```
-[//]: # (inf_rn50_imp_bf16)
+[//]: # (marker_inf_rn50_imp_bf16)
 
 ##### BERT
 
-[//]: # (inf_bert_imp_bf16)
+[//]: # (marker_inf_bert_imp_bf16)
 ```
 import torch
 from transformers import BertModel
@@ -411,15 +411,15 @@ with torch.no_grad():
   with torch.cpu.amp.autocast():
     model(data)
 ```
-[//]: # (inf_bert_imp_bf16)
+[//]: # (marker_inf_bert_imp_bf16)
 
 #### TorchScript Mode
 
 We recommend you take advantage of Intel® Extension for PyTorch\* with [TorchScript](https://pytorch.org/docs/stable/jit.html) for further optimizations.
 
 ##### Resnet50
 
-[//]: # (inf_rn50_ts_bf16)
+[//]: # (marker_inf_rn50_ts_bf16)
 ```
 import torch
 import torchvision.models as models
@@ -440,11 +440,11 @@ with torch.no_grad():
 
     model(data)
 ```
-[//]: # (inf_rn50_ts_bf16)
+[//]: # (marker_inf_rn50_ts_bf16)
 
 ##### BERT
 
-[//]: # (inf_bert_ts_f16)
+[//]: # (marker_inf_bert_ts_f16)
 ```
 import torch
 from transformers import BertModel
@@ -470,7 +470,7 @@ with torch.no_grad():
 
     model(data)
 ```
-[//]: # (inf_bert_ts_f16)
+[//]: # (marker_inf_bert_ts_f16)
 
 ### INT8
 
@@ -491,7 +491,7 @@ Please follow the steps below to perform static calibration:
 7. Save the INT8 model into a `pt` file.
 
 
-[//]: # (int8_static)
+[//]: # (marker_int8_static)
 ```
 import os
 import torch
@@ -521,7 +521,7 @@ with torch.no_grad():
 
 traced_model.save("quantized_model.pt")
 ```
-[//]: # (int8_static)
+[//]: # (marker_int8_static)
 
 ##### Dynamic Quantization
 
@@ -535,7 +535,7 @@ Please follow the steps below to perform static calibration:
 6. Run inference to perform dynamic quantization.
 7. Save the INT8 model into a `pt` file.
 
-[//]: # (int8_dynamic)
+[//]: # (marker_int8_dynamic)
 ```
 import os
 import torch
@@ -563,7 +563,7 @@ with torch.no_grad():
 
 traced_model.save("quantized_model.pt")
 ```
-[//]: # (int8_dynamic)
+[//]: # (marker_int8_dynamic)
 
 #### Deployment
 
@@ -575,7 +575,7 @@ Follow the steps below:
 2. Load the INT8 model from the saved file.
 3. Run inference.
 
-[//]: # (int8_deploy)
+[//]: # (marker_int8_deploy)
 ```
 import torch
 #################### code changes ####################
@@ -590,7 +590,7 @@ data = torch.rand(<shape>)
 with torch.no_grad():
   model(data)
 ```
-[//]: # (int8_deploy)
+[//]: # (marker_int8_deploy)
 
 oneDNN provides [oneDNN Graph Compiler](https://github.com/oneapi-src/oneDNN/tree/dev-graph-preview4/doc#onednn-graph-compiler) as a prototype feature that could boost performance for selective topologies. No code change is required. Install <a class="reference external" href="installation.md#installation_onednn_graph_compiler">a binary</a> with this feature enabled. We verified this feature with `Bert-large`, `bert-base-cased`, `roberta-base`, `xlm-roberta-base`, `google-electra-base-generator` and `google-electra-base-discriminator`.
 
@@ -604,7 +604,7 @@ The example code below works for all data types.
 
 **example-app.cpp**
 
-[//]: # (cppsdk_sample)
+[//]: # (marker_cppsdk_sample)
 ```cpp
 #include <torch/script.h>
 #include <iostream>
@@ -626,11 +626,11 @@ int main(int argc, const char* argv[]) {
     return 0;
 }
 ```
-[//]: # (cppsdk_sample)
+[//]: # (marker_cppsdk_sample)
 
 **CMakeLists.txt**
 
-[//]: # (cppsdk_cmake)
+[//]: # (marker_cppsdk_cmake)
 ```cmake
 cmake_minimum_required(VERSION 3.0 FATAL_ERROR)
 project(example-app)
@@ -642,7 +642,7 @@ target_link_libraries(example-app "${TORCH_LIBRARIES}")
 
 set_property(TARGET example-app PROPERTY CXX_STANDARD 14)
 ```
-[//]: # (cppsdk_cmake)
+[//]: # (marker_cppsdk_cmake)
 
 **Command for compilation**