Activity for Conscious Artificial Intelligence
-
Joao Paulo Schwarz Schuler committed [r2094] on Current Source Code
mode objfpc
-
adding option to change activation function in attention and transformer
-
simple image classifier is back using SGD
-
adding example of image classifier with attention mechanism
-
fixing attention with images
-
Using Adam with the simple image classifier
-
Coding denoiser
-
Attention on images
-
fixes comments
-
fixes comments
-
fixes comments
-
code portability
-
removing MTP
-
removing MTP
-
removing MTP
-
removing MTP
-
removing MTP
-
removing MTPCPU dependency
-
removing MTPCPU dependency
-
tokenizers and transformers are ready to use
-
experimental tokenizer is ready to use
-
samplers on pixels
-
coding experimental tokenizer
-
coding experimental tokenizer
-
coding experimental tokenizer
-
coding tokenizer
-
coding tokenizer
-
coding tokenizer
-
coding experimental encoder
-
coding experimental encoder
-
coding experimental encoder
-
coding experimental encoder
-
coding experimental encoder
-
coding experimental encoder
-
coding experimental encoder
-
coding experimental encoder
-
coding experimental encoder
-
coding CAI encoding
-
coding CAI encoding
-
coding CAI encoding
-
coding CAI encoding
-
Coding experimental BPE
-
minor speed gains
-
minor speed gains
-
minor speed gains
-
fixes memory usage in grouped convolutions
-
Adding AddCompressedTransformerBlockCAI
-
Adding AddCompressedTransformerBlockCAI
-
Adding GenerateStringFromCasualCharNN
-
Adding GenerateStringFromCasualCharNN
-
Adding StringToArrayOfInteger
-
fixing NLP examples
-
Adding TVolume.OneHotEncodingAtEnd
-
Adding TVolume.OneHotEncodingAtEnd
-
updating CAI transformers
-
updating CAI transformers
-
updating CAI transformers
-
updating CAI transformers
-
Experimental CAI transformer
-
Experimental CAI transformer
-
Experimental CAI transformer
-
Experimental CAI transformer
-
Experimental CAI transformer
-
Experimental CAI transformer
-
Experimental CAI transformer
-
Adding TNNetSignedSquareRootN
-
coding CAI transformer
-
coding CAI transformer
-
add TNNetSignedSquareRoot1
-
grouped transformer
-
Coding transformer blocks
-
Normalization on CAI transformer block
-
TNNetChannelNorm fix
-
adding TNNetChannelNorm
-
fixes bug 137
-
fixes TNNetChannelZeroCenter.BackpropagateNoTest
-
CAI Transformer numerical stability
-
numerical stability
-
Code portability with Delphi
-
Code portability with Delphi
-
Adding TNNetDeepConcat.Replicate
-
Delphi compatibility
-
Grouped Transformer Decoder
-
1000+ layers deep support
-
1000+ layers deep support
-
trying to converge 1000+ deep models
-
trying to converge 1000+ deep models
-
trying to converge 1000+ deep models
-
trying to converge 1000+ deep models
-
speeding up backpropagation on 1000+ deep models
-
attempt to stabilize 1000+ layers deep
-
experimenting new CAI Norm
-
experimenting new CAI Norm
-
experimenting new CAI Norm
-
experiment to avoid overflow error in 1000+ layers deep models
-
trying to find overflow error
-
experimental pointwise softmax
-
speeding up GetMaxAbs
-
experiment for numerical stability
-
experiment for numerical stability
1 >