Sunday, December 23, 2018

[TensorFlow] My example of using SavedModelBuilder to do inference in TensorFlow

The purpose of this post is to show my example of SavedModelBuilder to do inference in TensorFlow. From my experiment, this approach can save a model with the signature that has input and output node name. And SavedModelBuilder can restore the graph based on the previously saved model pb file and the signature definition. Once, the restore is done, the inference task can be executed directly without GPU device needed if the training task is on GPU device.

Friday, December 21, 2018

[Reinforcement Learning] Get started to learn Actor Critic for reinforcement learning

Actor-Critic is basically combined with Policy Gradient (Actor)  and Function Approximation (Critic) based algorithm together. Actor is based on the probability given by policy to act and Critic judges the performance of Actor and gives the score. So, Actor will improve its probability given by policy based on Critic's judge and score. The following diagram is the concept:

Thursday, December 13, 2018

Wednesday, December 12, 2018

[Reinforcement Learning] Using dynamic programming to solve a simple GridWorld with 4X4

I borrow the example and its source code from here which is a dynamic programming to solve a simple GridWorld with 4X4 and put my explanation for the calculation of value function. Hope that will help to understand dynamic programming and Markov Reward Process(MRP) more quickly.

Wednesday, November 14, 2018

[RNN] What are the difference of input and output's tensor shape in dynamic_rnn and static_rnn using TensorFlow

When studying RNN, my first issue encountered in my program is about the shape of input and output tensors. Shape is a very important information to connect between layers. Here I just directly point out what are differences in input/output shape of static RNN and dynamic RNN.
P.S: If you use Keras to write your RNN model, you won't need to deal with these details.

Monday, November 12, 2018

[TensorFlow] The explanation of average gradients by example in data parallelism

When studying some examples of training model using Multi-GPUs ( in data parallelism ), the average gradients function always exists in some kind of ways, and here is a simple version as follows:

Tuesday, October 30, 2018

[TensorFlow] Train in Tensorflow and do inference with the trained model

If you want to train your model in Tensorflow and do inference with the trained model, you can refer to this post.

1. Train your model

I will use the simple CNN model in my previous post:
[ONNX] Train in Tensorflow and export to ONNX (Part II)

So, after training, you will get these files:
├── checkpoint
├── graph.pbtxt
├── my_mnist_model.index
└── my_mnist_model.meta

Wednesday, October 24, 2018

[LLVM] LLVM studying list for newbie

If you are an LLVM newbie and are interested in LLVM like me, you may take a look at my LLVM studying list. It takes time for me to search the related resources and documents. So, I think it will help somehow. By the way, most of my list items are written in Chinese so that those who are native Engish speakers may not suit for this.

Tuesday, October 23, 2018

[TensorFlow] Does it help the processing time and transmission time if increasing CUDA Steam number in TensorFlow?

Before starting to increase the CUDA Steam number in TensorFlow, I want to recap some ideas about the Executor module. When TensorFlow session runs, it will build Executor. Meanwhile, if you enable CUDA in TensorFlow build configuration, the Executor will add visible GPU devices and create TF device object (GPUDevice object) mapping to physical GPU device. There are 4 kinds of streams inside GPUDevice:

  • CUDA stream 
  • Host_to_Device stream
  • Device_to_Host stream
  • Device_to_Device stream

Wednesday, October 17, 2018

[TensorFlow Grappler] How to do the topological sorting in TensorFlow Grappler?

If you try to implement some optimizers in TensorFlow Grappler, you must have to know how to deal with the directed computation graph. One of the most important tools/knowledges is topological sorting.
The definition from Wiki: Topological sorting
"In the field of computer science, a topological sort or topological ordering of a directed graph is a linear ordering of its vertices such that for every directed edge uv from vertex u to vertex v, u comes before v in the ordering."

[Tool] To draw a sequence diagram using online tool sequencediagram

This website provides an online free tool for users to draw the sequence diagram as follows:

Basically, you can follow the instructions at the left top corner button. Check it out.
Here is my example of the sequence diagram about tracing some source codes of XLA AOT in TensorFlow.

[TensorFlow Grappler] The ways to traverse all nodes' input and output in the graph using C++ in TensorFlow Grappler

Here I want to introduce 2 ways to traverse all nodes' input and output in the graph using C++ in Grappler.
P.S: you have to be able to get GrapplerItem and GraphDef objects in your code.

First, check my example node name in Tensorboard as follows:

Monday, October 1, 2018

[NUMACTL] How to use numactl in practice?

I recently attended the Intel AI workshop and they gave an advice of using NUMACTL to improve the performance of training and inferencing in Deep Learning with Intel Caffe. Here I post some related information as follows:

Sunday, September 16, 2018

[TFLMS] Large Model Support in TensorFlow by Graph Rewriting

This post just introduces this paper "Large Model Support in TensorFlow by Graph Rewriting" and it is published as a pull request in the TensorFlow repository for contributing to the TensorFlow community. With TFLMS, we were able to train ResNet-50 and 3DUnet with 4.7x and 2x larger batch size, respectively. Quite amazing...

Friday, September 7, 2018

Monday, September 3, 2018

[XLA related] Sort out my thought and notes about XLA related

This post could be a little bit unstructured because it's for my reference in notes.
I recently found that there are several slides in SlideShare which contain very good information and source code analysis/study about XLA related as follows:

Wednesday, August 29, 2018

[TensorFlow] My simple way to profile TensorFlow and dump variables and GPU memory

As we know that if we want to profile Tensorflow graph and know what operations take more time and what less. This can be done with Tensorflow timeline module like this:
( I ignore the part of the model to simplify my example code )
run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()
with tf.Session(config=config) as sess:
    for epoch in range(n_epochs):
        for iteration in range(10):
  , feed_dict={X: picture, y:picture_label}, 
                     options=run_options, run_metadata=run_metadata)
            fetched_timeline = timeline.Timeline(run_metadata.step_stats)
            chrome_trace = fetched_timeline.generate_chrome_trace_format()
            with open('timeline_step_%d.json' % iteration, 'w') as f:

Monday, August 20, 2018

[ONNX] Train in Tensorflow and export to ONNX (Part II)

If you read the previous post as the link below, you probably may ask a question: If the input TF graph for freezing is not a binary format, what do we do?

Let us recall the previous example below. The file "graph.proto" is the binary format of the protobuf file for TensorFlow graph generated from the following function:
  with open("graph.proto", "wb") as file:
    graph = tf.get_default_graph().as_graph_def(add_shapes=True)

Thursday, August 16, 2018

[TensorFlow] Rewriter_Config and Memory Optimization Passes

In the previous post as the below link, I mentioned that the default value of rewrite_config seems to change a little bit.

To clarify my doubt, I check the TensorFlow's and arrange the mapping table:

[TensorFlow] How to print the timestamp of a node/operation of computation graph in run-time?

When some people first time tries to debug or print out information of some result from a node/operation in the computation graph in TensorFlow, they maybe confuse about how to do it. Fortunately, someone in Google gave a great explanation of the print function:
After reading it, you should understand how tf.Print() function works and to use it.

Wednesday, August 8, 2018

[ONNX] Use ONNX_TF and nGraph_ONNX to do inference/prediction with ONNX model

Here I try to use the pre-trained model from ONNX model zoo, which the models are already converted from some deep learning framework. So I download the Resnet50 model from the following URL and untar it:

tar -xzvf resnet50.tar.gz 
P.S: pre-trained ONNX models:

Then, I can do the inference/prediction using this ONNX model in two ways:

[ONNX] Train in Tensorflow and export to ONNX (Part I)

From my point of view, ONNX is a model description spec and ONNX model needs Deep Learning framework or backend tool/compiler which supports it to run.
The advantage of ONNX as I know is about portable and exchangeable between DL frameworks.
Here I will use this tutorial to convert TensorFlow's model to ONNX model by myself.

Monday, July 30, 2018

[Fun] compress and composite dataset to one image file

If you want to compress and composite the images in a dataset, like MNIST or CIFAR-10, you probably can refer to my example.

People need to see it to believe it. Here you go:

Tuesday, July 17, 2018

[Confusion Matrix] How to calculate confusion matrix, precision and recall list from scratch

I directly give an example which is with 10 categories, such as CIFAR-10 and MNIST. It explains how to calculate the confusion matrix, precision and recall list from scratch in Python. My data is generated at random. You should replace by yours. Here it goes:

Saturday, July 14, 2018

[Qt5] How to develop Qt5 GUI with TensorFlow C++ library?

Here I give a simple and complete example of how to develop Qt5 GUI with TensorFlow C++ library on Linux platform. Please check out my GitHub's repository as follow:

Monday, July 9, 2018

[TensorFlow] How to implement LMDBDataset in API?

I have finished implementing the LMDBDataset in API.  It could be not the bug-free component, but at least it's my first time to try to implement C++ and Python function in TensorFlow. The API architecture looks like this:

Wednesday, July 4, 2018

[TensorFlow] How to build your C++ program or application with TensorFlow library using CMake

When you want to build your  C++ program or application using TensorFlow library or functions, you probably will encounter some header file missed issues or linking problems. Here is the step list that I have verified and it works well.

1. Prepare TensorFlow ( v1.10) and its third party's library
$ git clone --recursive
$ cd tensorflow/contrib/makefile
$ ./

Tuesday, June 26, 2018

[XLA JIT] How to turn on XLA JIT compilation at multiple GPUs training

Before I discuss this question, let's recall how to turn on  XLA JIT compilation and use it in TensorFlow python API.

1. Session
Turning on JIT compilation at the session level will result in all possible operators being greedily compiled into XLA computations. Each XLA computation will be compiled into one or more kernels for the underlying device.

Sunday, June 24, 2018

Thursday, June 21, 2018

[TensorFlow] How to get CPU configuration flags (such as SSE4.1, SSE4.2, and AVX...) in a bash script for building TensorFlow from source

The AVX and SSE4.2 and others are offered by Intel CPU. (AVX and SSE4.2 are CPU infrastructures for faster matrix computations) Did you wonder what CPU configuration flags (such as SSE4.1, SSE4.2, and AVX...) you should use on your machine when building Tensorflow from source? If so, here is a quick solution for you.

Wednesday, June 20, 2018

[TensorFlow 記憶體優化實驗] Compare the memory options in Grappler Memory Optimizer

As we know that in Tensorflow, there is an optimization module called "Grappler". It provides many kinds of optimization functionalities, such as: Layout, Memory, ModelPruner, and so on... In this experiment, we can see the effect of some memory options enabled in a simple CNN model using MNIST dataset.

Thursday, June 14, 2018

[XLA 研究] How to use XLA AOT compilation in TensorFlow

This document is going to explain how to use AOT compilation in TensorFlow. We will use the tool: tfcompile, which is a standalone tool that ahead-of-time (AOT) compiles TensorFlow graphs into executable code. It can reduce the total binary size, and also avoid some runtime overheads. A typical use-case of tfcompile is to compile an inference graph into executable code for mobile devices. The following steps are as follows:

1. Build tool: tfcompile
> bazel build --config=opt --config=cuda //tensorflow/compiler/aot:tfcompile

Friday, June 8, 2018

[XLA 研究] Take a glance to see the graph changes in XLA JIT compilation

In the preamble of this article, to understand XLA JIT is pretty hard because you probably need to understand TensorFlow Graph, Executor,  LLVM, and math... I have been through this painful study work somehow so that I hope my experience can help for those who are interested in XLA but have not get understood yet.

Thursday, June 7, 2018

[TX2 研究] My first try on Jetson TX2

I got a Jetson TX2 several days ago from my friend and it looks like following pictures. I setup it using Nivida's installing tool: JetPack-L4T-3.2 version ( During the installation, I indeed encounter some issues with not abling to setup IP address on TX2, and I resolved it. If anyone still has this issue, let me know and I will post another article to explain the resolving steps.