MyCaffe 0.10.1.145-beta1

MyCaffeControl

A complete C# re-write of Berkeley's open source Convolutional Architecture for Fast Feature Encoding (CAFFE) for Windows C# Developers with full On-line Help, now with Policy Gradient Reinforcement Learning, cuDNN LSTM Recurrent Learning, and Neural Style Transfer support!

This is a prerelease version of MyCaffe.
There is a newer prerelease version of this package available.
See the version list below for details.
Install-Package MyCaffe -Version 0.10.1.145-beta1
dotnet add package MyCaffe --version 0.10.1.145-beta1
<PackageReference Include="MyCaffe" Version="0.10.1.145-beta1" />
For projects that support PackageReference, copy this XML node into the project file to reference the package.
paket add MyCaffe --version 0.10.1.145-beta1
The NuGet Team does not provide support for this client. Please contact its maintainers for support.

CUDA 10.1.168, cuDNN 7.6, nvapi 410, Native Caffe up to 10/24/2018, Windows 10-1809, Driver 430.86

MyCaffe now supports dual RNN/RL training that allows multi-pass training where the first pass involves RNN training and the second pass involves RL training that uses the
already trained RNN side of the model.

IMPORTANT NOTE: When using TCC mode, we recommend that ALL headless GPU’s are placed in TCC mode for we have experienced stability issues when using a mix of TCC and WDM modes with headless GPU’s.

REQUIRED SOFTWARE:
1.) Install NVIDIA CUDA 10.1.168 which you can download from https://developer.nvidia.com/cuda-downloads
2.) Install NVIDIA cuDNN 7.6 which you can download from https://developer.nvidia.com/cudnn
3.) Download and install Microsoft SQL Express 2016 (or later).

This release of the MyCaffe AI Platform and Test Applications has the following new additions:

  • CUDA 10.1.168/cuDNN 7.6 supported (with driver 430.86).
  • NVIDIA Quadro RTX 8000 now supported.
  • New dual Recurrent/Reinforcement trainer.
  • New STEP supported added to RNN.SIMPLE and PG.ST trainers.
  • Added support for single and multiple labels with recurrent learning.
  • Added episode caching and elite selection to PG.MT trainer.
  • Added auto-tests for LSTM Sin-Wave and Random-Wave.
  • Added new OnHotLayer to convert values into one-hot vectors.
  • Added __half memory size support in numerous layers (CONVOLUTION, POOLING, RELU, TANH, SIGMOID and INPUT)
  • Increased minimum compute from 3.5 up to 5.3 (Maxwell GPU) for __half support.
  • Optimized memory use, reducing usage with some larger models.

The following bug fixes are in this release:

  • Fixed bugs related to LSTM layers connected to Data layers.
  • Fixed bugs related to freeze learning.
  • Fixed bugs related in loss layers causing run networks to fail during creation.
  • Fixed bugs in test application related to selected GPU not being used.
  • Fixed bugs related to cuBlas errors, now correct cuBlas errors are returned.

Easily create the CIFAR-10 and MNIST datasets using the MyCaffe Test Application which you can download from the MyCaffe GitHub site.

Create and train the Neural Style Transfer, Recurrent Learning, Policy Gradient Reinforcement Learning, Auto-Encoder, DANN and ResNet models by following step-by-step instructions in the SignalPop Tutorials. And, to see other cool examples that show what MyCaffe can do, see the SignalPop Examples.

If you would like to visually design, develop, test and debug your models, see the SignalPop AI Designer specifically designed to enhance your MyCaffe deep learning.

Also, check out the SignalPop Universal Miner that not only keeps your GPU's cool as you train, but also gives you detailed information on each of your GPU's (such as temperature, fan speed, overclock, and usage), and allows you to easily mine Ethereum. When not training AI, put those GPU's to use making some Ether - never let a good GPU go to waste!

Happy ‘deep’ learning!

CUDA 10.1.168, cuDNN 7.6, nvapi 410, Native Caffe up to 10/24/2018, Windows 10-1809, Driver 430.86

MyCaffe now supports dual RNN/RL training that allows multi-pass training where the first pass involves RNN training and the second pass involves RL training that uses the
already trained RNN side of the model.

IMPORTANT NOTE: When using TCC mode, we recommend that ALL headless GPU’s are placed in TCC mode for we have experienced stability issues when using a mix of TCC and WDM modes with headless GPU’s.

REQUIRED SOFTWARE:
1.) Install NVIDIA CUDA 10.1.168 which you can download from https://developer.nvidia.com/cuda-downloads
2.) Install NVIDIA cuDNN 7.6 which you can download from https://developer.nvidia.com/cudnn
3.) Download and install Microsoft SQL Express 2016 (or later).

This release of the MyCaffe AI Platform and Test Applications has the following new additions:

  • CUDA 10.1.168/cuDNN 7.6 supported (with driver 430.86).
  • NVIDIA Quadro RTX 8000 now supported.
  • New dual Recurrent/Reinforcement trainer.
  • New STEP supported added to RNN.SIMPLE and PG.ST trainers.
  • Added support for single and multiple labels with recurrent learning.
  • Added episode caching and elite selection to PG.MT trainer.
  • Added auto-tests for LSTM Sin-Wave and Random-Wave.
  • Added new OnHotLayer to convert values into one-hot vectors.
  • Added __half memory size support in numerous layers (CONVOLUTION, POOLING, RELU, TANH, SIGMOID and INPUT)
  • Increased minimum compute from 3.5 up to 5.3 (Maxwell GPU) for __half support.
  • Optimized memory use, reducing usage with some larger models.

The following bug fixes are in this release:

  • Fixed bugs related to LSTM layers connected to Data layers.
  • Fixed bugs related to freeze learning.
  • Fixed bugs related in loss layers causing run networks to fail during creation.
  • Fixed bugs in test application related to selected GPU not being used.
  • Fixed bugs related to cuBlas errors, now correct cuBlas errors are returned.

Easily create the CIFAR-10 and MNIST datasets using the MyCaffe Test Application which you can download from the MyCaffe GitHub site.

Create and train the Neural Style Transfer, Recurrent Learning, Policy Gradient Reinforcement Learning, Auto-Encoder, DANN and ResNet models by following step-by-step instructions in the SignalPop Tutorials. And, to see other cool examples that show what MyCaffe can do, see the SignalPop Examples.

If you would like to visually design, develop, test and debug your models, see the SignalPop AI Designer specifically designed to enhance your MyCaffe deep learning.

Also, check out the SignalPop Universal Miner that not only keeps your GPU's cool as you train, but also gives you detailed information on each of your GPU's (such as temperature, fan speed, overclock, and usage), and allows you to easily mine Ethereum. When not training AI, put those GPU's to use making some Ether - never let a good GPU go to waste!

Happy ‘deep’ learning!

Release Notes

MyCaffe AI Platform

Version History

Version Downloads Last updated
0.10.1.169-beta1 37 7/8/2019
0.10.1.145-beta1 74 5/31/2019
0.10.1.48-beta1 88 4/18/2019
0.10.1.21-beta1 89 3/5/2019
0.10.0.190-beta1 151 1/15/2019
0.10.0.140-beta1 106 11/29/2018
0.10.0.122-beta1 125 11/15/2018
0.10.0.75-beta1 126 10/7/2018