Commits

Commits on Nov 10, 2015

Update readme
Hello1024
committed

Commits on Nov 2, 2015

Commits on Oct 27, 2015

Commits on Oct 26, 2015

Update Readme.md
Hello1024
committed

Commits on Oct 25, 2015

Add distributed compute ability with shared tensors. Provides near-linear speedup
Hello1024
committed

Commits on Oct 16, 2015

Readme tweaks and clarifications
karpathy
committed

Commits on Oct 11, 2015

speeding up the code by about 15% with faster indexing. I did not imagine this would have such a large effect. Thank you @vseledkin for pointing this out
karpathy
committed

Commits on Sep 21, 2015

adding clarification justifying not normalizing gradient by seqlength
karpathy
committed

Commits on Sep 20, 2015

changing default LSTM initialization to use biases of 1.0 for the forget gates, which is a common trick and encourages remembering in the beginning of training. I ran 2 medium-sized experiments and…
karpathy
committed

Commits on Aug 27, 2015

Merge pull request karpathy#92 from ajkl/patch-1
karpathy
committed

Commits on Aug 26, 2015

adding torch/nn to dependencies
ajkl
committed

Commits on Aug 20, 2015

Commits on Aug 10, 2015

Commits on Aug 8, 2015

fixing a bug introduced in previous commit. We have to use doubles not floats when we convert GPU models to CPU, because CPU trained models use doubles. This introduced an error where a CPU checkpo…
karpathy
committed

Commits on Aug 5, 2015

quick patch for converting GPU checkpoints to CPU checkpoints
karpathy
committed

Commits on Jul 27, 2015

Commits on Jul 25, 2015

Commits on Jul 20, 2015

clean unused code
donglixp
committed