PyTorch’s popularity spread only three years after its launch: the open-source ML library posted a 194% user growth in the first half of 2019. Since then PyTorch has not looked back.
According to the survey by Stackoverflow, while TensorFlow is the most in-demand library, PyTorch is more preferred. Data from Papers with Code shows that PyTorch is the most preferred library among researchers. When it comes to frameworks (repositories are classified by framework by inspecting the contents of each GitHub repository and checking imports in code), searches from January 2020 to December 2021 show PyTorch accounts for 75% of implementations. Has-much higher than Tensorflow (below 25%), JAX, MXNeT, PaddlePaddle, Torch, Caffe2 and MindSpore.
The growing popularity of PyTorch
Professor Francois Fleuret from the University of Geneva conducted a poll on Twitter earlier this month asking developers to vote for their core framework, and 72% voted for PyTorch.
Back in 2020, Elon Musk replied to Pranay Pothol, a Pune-based engineer, in a tweet thread about AI capabilities at Tesla: Musk said ‘Pytorch is the most used external tool set/library’ Prannoy’s question To build Pytorch NN for?. The response started a debate on the microblogging site and Prannoy was later seen defending Pytorch, saying “Pytorch is faster and sometimes faster than TensorFlow because execution in PyTorch is, by default, asynchronous.”
PyTorch’s API has become a runaway hit among developers. While some users stick to JAX and NumPy, they are open to PyTorch for DL.
Many PyTorch users have actually switched from TensorFlow and find it very user friendly.
Netlyt co-founder and AI engineer Tony Petrov wrote, “I personally use PyTorch for NLP tasks. The reason is not because Keras or Tensorflow are inferior in performance, but because of their supporting libraries.” There is a shortage.”
Mukhtab Mayank, Co-Founder, Parallel Dots, said: “PyTorch is the most productive and easy-to-use framework in my opinion. We know that it is very easy to deploy in production to medium sized deployment in the form of Pytorch library.
PyTorch is an open-source machine learning library based on the Torch library. The system was developed primarily by Facebook’s AI Research Lab (FAIR): Adam Paszke, Saumit Chintalla, Sam Gross and Gregory Chanan worked on building the system and now has core maintainers, and a broader set of developers. which directly merges pull requests and various parts of the core code base itself.
Many researchers and users who don’t focus too much on production-level code prefer PyTorch because the coding on it is straightforward. With Torch Script they get high flexibility in transitioning between eager mode and graph mode to use both processes.
PyTorch is also supported by a thriving community. PyTorch has its own forum, but developers using the code extend to Reddit, Quora, Twitter, and Github. From tutorials, discussions to opinions and advice, the PyTorch community is thriving at a rapid pace. According to the official website, “PyTorch adopts a governance structure with a small group of maintainers driving the overall project direction with a strong bias towards PyTorch’s design philosophy, where design and code contributions are valued. “
A trend report from Google showed that PyTorch has attracted a lot more interest over the past 12 months than TensorFlow.
Significant chunks of deep learning software built on top of PyTorch, including Tesla Autopilot, Huggingface’s Transformer, and even Uber’s Pyro. Pytorch’s future looks bright.