Sign in
Categories
Migration Mapping Assistant Your Saved List Partners Sell in AWS Marketplace Amazon Web Services Home Help

Reviews from AWS Marketplace

0 AWS reviews
  • 5 star
    0
  • 4 star
    0
  • 3 star
    0
  • 2 star
    0
  • 1 star
    0

External reviews

37 reviews
from G2

External reviews are not included in the AWS star rating for the product.


    Ahmad A.

Amazing library if you are expert in machine learning,

  • November 04, 2018
  • Review provided by G2

What do you like best?
It is amazingly fast, It works in parallel, and supports GPU
What do you dislike?
The idea of Tensors is not very well explained in the official website, and that makes the user to panic if they do not understand the most basic idea of it after an hour of digging
What problems are you solving with the product? What benefits have you realized?
I am a researcher and an engineer, I combine my Machine learning knowledge with the use of TensorFlow to apply it to real life engineering applications
Recommendations to others considering the product:
Do not get disapointed if you are difficulties mastering this library,
once you know how to correctly use it, the creating fantastic machine learning models will be fun and easy


    Andrew C.

TensorFlow Worth the Learning Curve

  • October 12, 2018
  • Review verified by G2

What do you like best?
I like how easy TensorFlow makes building ML models without sacrificing low-level implementation capabilities. It includes a wide variety of prebuilt models and model subblocks that can be plugged together using simple python scripts. Tensorflow handles the implementation details seamlessly allows you to abstract away the underlying hardware, be they GPU's, CPU's or TPU's. We don't have to think about what kind of convolution algorithm we're using unless we absolutely want to. The data ingestion pipeline makes handling hundreds of GB of data a simple task. No more loading everything into RAM or worrying about file access and formatting. It does come with a price and it isn't as intuitive as it could be but it is well worth learning if you are serious about applied machine learning or just experimenting.
What do you dislike?
I dislike the define-and-run model of TensowFlow. It is unintuitive and occasionally lends itself to clunky solutions. It differs from the define-by-run model of the other major ML frameworks which is a barrier to access for many. I also dislike the structure of variables as tensors. It is often unclear whether your variables need to be tensors or plain python types. Once you get the hang of using TensorFlow it becomes obvious but something as simple as variable definitions shouldn't be so opaque.
What problems are you solving with the product? What benefits have you realized?
We are building models using Tensorflow that can learn from our datasets to accurately classify samples. Previously, building these models required highly domain specific knowledge were built in an ad hoc way for each class of data. Tensorflow allows us to build one model that can be far more easily adapted and changed.
Recommendations to others considering the product:
It is worth the learning curve. Google has a fantastic introductory series on both Machine Learning and TensorFlw specifically that I highly recommend.


    Farming

best one out there

  • October 06, 2018
  • Review provided by G2

What do you like best?
the Tensor flow API is the best and the model created on the desktop can be used any where.
And the availability of pre trained models is anothe rbest part.
What do you dislike?
Little hard for the non coding person to train and create models.
What problems are you solving with the product? What benefits have you realized?
ML and predictions
Recommendations to others considering the product:
Should develop a UI for deploying new models and training them.


    Computer Software

A powerful deep learning library, with certain rough edges

  • July 25, 2018
  • Review provided by G2

What do you like best?
Automatic differentiation and support for backpropagation through many useful mathematical operations. Tensorboard interface for monitoring and visualisation.
What do you dislike?
The programming model is somewhat cumbersome, and reliant on global state behind the scenes. For any task there seems to be multiple incompatible ways of achieving it, with varying degrees of documentation. The API is a mess, with many different high level interfaces. There is no standardised workflow, which makes mixing and matching models from different sources very difficult.
What problems are you solving with the product? What benefits have you realized?
Training deep learning models for video analysis. After the initial hurdles, it does the job.
Recommendations to others considering the product:
Carefully consider the alternatives, such as PyTorch which can be easier for development by specifically targeting Python and the Pythonic way of programming (although potentially at the expense of flexibility).


    Information Services

Powerful machine learning library!

  • June 14, 2018
  • Review provided by G2

What do you like best?
Tensorflow is an extremely powerful library whose users can use it for classification, regression, or any number of complex neural network models, such as GANs and CNNs. It can also be used as a backend to Keras or other higher level libraries.
What do you dislike?
The API leaves something to be desired, as it can get quite complex and messy as you write more code. It is pertinent that organization of the code is better handled. As well, for very large applications, it may be slower than competing libraries such as Torch.
What problems are you solving with the product? What benefits have you realized?
We use it to create classification and regression pipelines for our clients, such as with agricultural yield prediction.
Recommendations to others considering the product:
Learn deep learning first as the API is extremely specific to those in the field of AI.


    Derek P.

A short path into artificial intelligence

  • April 26, 2018
  • Review provided by G2

What do you like best?
TensorFlow is a platform that is very easy to learn yet hard to master. Due to a heavy amounts of code samples it is very easy to dive head first into Artificial Intelligence
What do you dislike?
Though very thorough TensorFlow documentation can feel very overwhelming.
What problems are you solving with the product? What benefits have you realized?
I have incorporated tensorflow into my image classifing solutions. Due to the compatibility for android and IOS devices tensorflow is incorporated into many of our apps
Recommendations to others considering the product:
Dont rush into it, take it slow and under the information


    Information Technology and Services

TensorFlow with Keras over Spark is a great solution

  • March 20, 2018
  • Review provided by G2

What do you like best?
Easy of creating models and parameterizing them
What do you dislike?
Need experienced programmers to set it up and train the models
What problems are you solving with the product? What benefits have you realized?
Predictive Analytics
Resource optimization
Recommendations to others considering the product:
Include Keras to further automate


    Naveen K.

Simple, Fast and easy

  • February 02, 2018
  • Review provided by G2

What do you like best?
Efficiency and the ease of deployment during projects
What do you dislike?
Boot up time and the sometimes the clumsiness
What problems are you solving with the product? What benefits have you realized?
Data Management


    Krishnan V.

Really good for AI/ML based use cases

  • February 02, 2018
  • Review provided by G2

What do you like best?
Helps setting up the neural network well - strongly recommend it

We have been using internally and evaluating how to use it.
What do you dislike?
Don't really dislike anything in it - maybe more flexibility
What problems are you solving with the product? What benefits have you realized?
- Better time to market
- Easier to deploy solutions in the market
- Develop newer capabilities
Recommendations to others considering the product:
Decide what are the use cases before taking up the platform. Fairly intuitive and easy to use


    Information Technology and Services

Simple way to build complex models

  • February 02, 2018
  • Review provided by G2

What do you like best?
I like how easy tensor flow makes it to build a simple neural network. You can have a model up and running in minutes, but tensorflow still provides advanced users with the ability to customize models a lot.
What do you dislike?
I wish that the documentation for tensorflow was more detailed. Sometimes I have a hard time finding answers to some questions that I have about tensorflow.
What problems are you solving with the product? What benefits have you realized?
I am trying to make more accurate predictive models in the credit industry. Tensorflow has helped to improve the accuracy of existing models built using other software by a few percent.