TensorFlow Reviews & Product Details

TensorFlow is flexible. It provides a platform for building and deploying machine learning models across a wide range of devices and media, and Tensorflow is really scalable, running on a single device to distributed systems with thousands of GPUs Review collected by and hosted on G2.com.

A few things I dislike about TensorFlow are it is resource intensive; TensorFlow is really resource intensive. It requires high computational power and a powerful GPU. the second thing is the learning curve TensorFlow can have a steep learning curve for beginners due to its complexity Review collected by and hosted on G2.com.

Tensorflow is the best library to work with neural networks and building model architecture. The functional API along with other functionalities makes it easy to define any model from easy to complex and train with ease. Review collected by and hosted on G2.com.

Tensorflow needs to add some development in context of memory. In order to deploy any model it takes around 400mb memory for just tensorflow lib. This is the only part which holds me back sometimes. Review collected by and hosted on G2.com.

One of the best features of Tensorflow is its ability to perform multicore training of models. Unlike the old frameworks, TF doesn't rely on single CPU training rather it allows distributed training of models which drastically decreases the training time we have several GBs of images to be trained for diffusion models. Review collected by and hosted on G2.com.

When developers are using Windows for development there are certain issues with the Python pip packages that are part of TF. There is no native support for Decision forests which is one of the most popular packages that is supported by other frameworks. I train la Review collected by and hosted on G2.com.

In tensorflow there have lots of methods for many purpose. And in tf you can do anything about deep learning. Review collected by and hosted on G2.com.

Tensorflow can build own UI for managing models and all. Review collected by and hosted on G2.com.

Tensorflow has several intuitive methods for implementing machine learning algorithms. I personally like to use the image classification section to understand how to detect patterns in supervised data. Review collected by and hosted on G2.com.

I think there is still some room for improvement in terms of readability. In particular, it feels like many of TensorFlow's commands don't follow a "pythonic" pattern. Review collected by and hosted on G2.com.

It provides all the recent algorithms that can run on a Convolutional neural network model. It provides training algorithms, metrics and optimizers for the deep learning algorithm. Review collected by and hosted on G2.com.

Python version of TensorFlow runs only on GPU-based processors. Review collected by and hosted on G2.com.

Easy to get started with. The TensorFlow ecosystem provides support tools to load data efficiently (TF Dataloaders) , build models (Keras), Optimize it (TF Lite), and Deploy and monitor (TFX) and it is production-ready. Review collected by and hosted on G2.com.

One concern I have is inconsistent APIs and functions. Confusion with TF 1 and TF 2. Lots of duplicate and redundant methods. Code customization for research purposes. Review collected by and hosted on G2.com.

Tensorflow is an excellent library for implementing linear algebra equations and algorithms. It also has Keras as its inbuilt module, a perfect module for deep learning and implementing neural network models.

I use it majorly for implementing and training deep learning models. It provides high customizability for defining our loss functions, activation functions, etc. Review collected by and hosted on G2.com.

The Keras interface provided inside TensorFlow is not the same as externally importing Keras. There are a few differences which can someone comfortable with Keras make several mistakes while developing using Tensorflow.

It also sometimes shows some errors which are easy to understand and often not even related to the code, but the running environment/kernel instead. Review collected by and hosted on G2.com.

For people who grew up learning the math of backprop, who enjoy thinking about syntax trees and computation graphs, Tensorflow will allow you to make full use of you that insight. Interesting loss functions like Wasserstein loss (where the gradient itself enters as part of the loss function) enter naturally. Review collected by and hosted on G2.com.

The mix between Tensorflow v1 and v2 code is somewhat difficult to learn, if you only get into it now. Tensorflow v2 is modeled much more on Keras, and is designed for you to particular architectures and pipelines. That's great, but if you then want to mix that with the flexibility of v1, you run into a lot of pain. Review collected by and hosted on G2.com.

There are so many points which I liked aboutTensorFlow i.e. It is fast and it's scalability on the larger dataset, Witht the help of TensorFlow I am able to write customisable evaluation functions. Review collected by and hosted on G2.com.

Overall performance of TensorFlow is good but the documentation of TensorFlow can be improved, Sometime i felt inconsistency in the algorithm which can be further optimised. Review collected by and hosted on G2.com.