MyCaffe 1.12.0.60

MyCaffe AI Platform (CUDA 11.8.0, cuDNN 8.8.0) with GPT and ChatGPT technology - version 1.12.0.56 ready!

MyCaffe now supports Encoder/Decoder Transformer Models like those used in ChatGPT! The MyCaffe AI Platform provides an easy AI solution for multiple AI disciplines, including:

• ChatGPT Models • GPT Models • Transformer Models • Object detection with Single-Shot Multi Box (SSD) • Onnx AI Model Support (import and export) • Classification with AlexNet, ResNet, VGG, NoisyNet, and Inception models • Classification with SiameseNet • Classification with TripletNet • Auto Encoders and DANN • Reinforcement Learning with Policy Gradient and Deep Q-Learning • Recurrent Learning with CharNet • Neural Style Transfer • Seq2Seq Models

Speed up AI training with the MyCaffe in-memory database that caches full datasets or drip-fed datasets into your local RAM on one side while feeding the training process on the other with label balanced data. Easily Train on multiple-GPUs with NCCL.

CUDA 11.8.0.522, cuDNN 8.8.0.121, nvapi 515, Windows 10-22H2/Windows 11-22H2, Driver 516.94

MyCaffe[1] (a complete C# re-write of CAFFE[2]) now supports Visual Studio 2022 and CUDA 11.8.0/cuDNN 8.8.0 and Windows 11.

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

IMPORTANT NOTES: If you receive a build error triggered by the 'xcopy' post-build event, change the Post-build event command line to: copy "$(SolutionDir)packages\MyCaffe.1.12.0.60\NativeBinaries\x64*.*" "$(TargetDir)"

REQUIRED SOFTWARE to use MyCaffe: 1.) Download and install full version of Microsoft SQL Express 2016 (or later). NOTE: The full version of SQL Express must be installed as opposed to the light version included in Visual Studio. Microsoft SQL Express can be downloaded from https://www.microsoft.com/en-us/sql-server/sql-server-downloads

2.) Download and install the MyCaffe Test Application for easy access to required NVIDIA DLL's. The MyCaffe Test Application which you can download from the MyCaffe GitHub site.

This release of the MyCaffe AI Platform and Test Applications has the following new additions: • CUDA 11.8.0.522/cuDNN 8.8.0.121/nvapi 515/driver 516.94 • Windows 11 22H2 • Windows 10 22H2, OS Build 19045.2604, SDK 10.0.19041.0 • Added GELU support to TransformerBlockLayer. • Added GELU_BERT support to TransformerBlockLayer. • Added Activation Graphing support to MyCaffe Auto Tests. • Upgraded to Google ProtoBuf 3.22.0 • Upgraded to NewtonJson 13.0.2 • Added ‘i4’, ‘i8’, and ‘f8’ type support to Blob LoadFromNumpy • Added new MultiheadAttentionLayer. • Added CudaDnn mask_batch support. • Added Blob.Compare support. • Added EncoderBlock support to TransformerBlockLayer • Added DecoderBlock support to TransformerBlockLayer • Added new TokenizedDataPairsLayer. • Added new TokenizedDataPairsLayerPy. • Added new algorithm support to SoftmaxLayer including LOG • Added new NLLLossLayer. • Added new CudaDnn channel_duplicate. • Added new CudaDnn matmul • Added new CudaDnn transpose_hw • Added new CudaDnn add3 • Added new CudaDnn channel_mean • Added support to CudaDnn channel_max/min to return indexes. • Added gpu support to ArgmaxLayer • Added Blob.SaveToImage. • Added support for loading very large numpy arrays.

The following bug fixes are in this release: • Improved internal blobs for better memory sharing. • Improved overall GPU memory use. • Fixed bug in Blob.CopyFrom wrt shape comparison. • Fixed bug in CudaDnn layer where large indexes are now properly passed. • Fixed bug in CausalSelfAttentionLayer related to batch training. • Optimized memory use on GPT and Encoder/Decoder models.

Easily run Encoder/Decoder for Language Translation[3], minGPT[4], Single-Shot Multi-Box Nets[5][6], import/export ONNX AI Models, run Triplet Nets[7][8], run Siamese Nets[9][10], Neural Style, train Deep Q-Learning or Policy Gradient models to beat Pong or Cart-Pole, or create the CIFAR-10 and MNIST datasets using the MyCaffe Test Application which you can download from the MyCaffe GitHub site.

Visually design, develop, test and train these models with the SignalPop AI Desigenr 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.

Happy ‘deep’ learning!

[1] MyCaffe: A Complete C# Re-Write of Caffe with Reinforcement Learning by D. Brown, 2018.

[2] Caffe: Convolutional Architecture for Fast Feature Embedding by Yangqing Jia, Evan Shelhamer, Jeff Donahue, Sergey Karayev, Jonathan Long, Ross Girshick, Sergio Guadarrama, Trevor Darrell, 2014, arXiv:1408.5093

[3] [GitHub: devjwsong/transformer-translator-pytorch] by Jaewoo (Kyle) Song, 2021

[4] GitHub: karpathy/minGPT by Andrej Karpathy, 2022

[5] SSD: Single Shot MultiBox Detector by Wei Liu, Dragomir Anguelov, Dumitru Erhan, Christian Szegedy, Scott Reed, Cheng-Yang Fu, Alexander C. Berg, 2016.

[6] GitHub: SSD: Single Shot MultiBox Detector, by weiliu89/caffe, 2016

[7] Deep metric learning using Triplet network by Elad Hoffer and Nir Ailon, 2018, arXiv:1412.6622

[8] In Defense of the Triplet Loss for Person Re-Identification by Alexander Hermans, Lucas Beyer and Bastian Liebe, 2017, arXiv:1703.07737v2

[9] Siamese Network Training with Caffe by Berkeley Artificial Intelligence (BAIR)

[10] Siamese Neural Network for One-shot Image Recognition by G. Koch, R. Zemel and R. Salakhutdinov, ICML 2015 Deep Learning Workshop, 2015.