Home

##Docs & Installation

Full documentation including installation and usage instructions and more is available from: http://neon.nervanasys.com/docs/latest/index.html

Model Zoo

We have many common and state of the art deep learning models ready for use with neon. They can be found in our model zoo.

Resources

• Benchmarks by Facebook and Baidu
• neon examples
• neon tutorials and docs
• neon course
• video tutorial
  • iPython Notebooks used in the tutorial
  • Slides from meet-ups & presentations
• Neon users group
• Blog posts
  • Models
    • Deep Reinforcement Learning with neon: Train neural networks to play Atari games (simple_dqn)
    • Using neon with OpenAI Gym
    • Scene recognition using neon
  • Customer use-cases
    • Agricultural Robotics: Blue River Technology: Convolutional Autoencoders for localizing stem-ground contact points in precision agriculture
    • Autonomous Driving: Serving the Automotive Industry with the Nervana Platform
    • Hyperparameter optimization of deep networks using SigOpt, neon and the Nervana Cloud
  • Speed
    • Neon with Multi-GPU on Nervana cloud: Train neural nets 7.2x faster with multiple GPUs
    • Neon on Kepler/AWS vs Titan/Nervana Cloud: Train neural nets 70x faster on the Nervana cloud vs. AWS
    • Neon with Winograd achieves 2-3x speedups over cuDNN v4
    • Persistent RNN kernels
• Kaggle competitions using neon
  • Whale challenge
  • BCI challenge
• Demos of neon capabilities including speech, video and reinforcement learning
• Convert Caffe models to neon caffe2neon