Age | Commit message (Collapse) | Author | |
---|---|---|---|
2016-06-04 | Merge pull request #169 from SeanNaren/R5R5 | Soumith Chintala | |
Fixed error with batchFirst assertion | |||
2016-05-26 | readme change | Soumith Chintala | |
2016-05-22 | Merge pull request #188 from szagoruyko/R5 | Sergey Zagoruyko | |
Fix avg pooling back-compatibility | |||
2016-05-22 | add avg pooling back compat test | Sergey Zagoruyko | |
2016-05-22 | Revert "R5 initial port" | Sergey Zagoruyko | |
This reverts commit ee6a36f0a366db777308b3bd60021d917761c758. | |||
2016-05-17 | Merge pull request #184 from borisfom/R5_up | Soumith Chintala | |
fix for V5 GA RNN APIs | |||
2016-05-17 | fix for V5 GA RNN APIs | Natalia Gimelshein | |
2016-05-17 | Merge pull request #183 from soumith/R5_ | Soumith Chintala | |
cudnn r5 final version checks | |||
2016-05-17 | cudnn r5 final version checks | soumith | |
2016-04-29 | Revert "Added extracting of bias and weights of each layer" | SeanNaren | |
This reverts commit b71a40186b5dec9ac18097663409c4098ea660d7. | |||
2016-04-28 | Added extracting of bias and weights of each layer | SeanNaren | |
2016-04-25 | Do not reset dropout descriptor if sequence length changes | ngimel | |
2016-04-23 | Merge pull request #171 from szagoruyko/R5-gradInput-move | Sergey Zagoruyko | |
[R5] Move gradInput:resizeAs(input) everywhere | |||
2016-04-20 | move gradInput resize to updateGradInput | Sergey Zagoruyko | |
2016-04-20 | Updated assertion for accGradParams | SeanNaren | |
2016-04-20 | Removed x calculations, updated makeContiguous function | SeanNaren | |
2016-04-20 | Updated gradOutput assertion | SeanNaren | |
2016-04-19 | Fixed error with batchFirst assertion | SeanNaren | |
2016-04-18 | Merge pull request #168 from borisfom/R5-int | Soumith Chintala | |
Rebased the rest of R5 changes. Added more RNN modules. | |||
2016-04-18 | Added tests, modified README and added RNN modules | SeanNaren | |
2016-04-18 | Add utility functions and correct tensor dimensions. | Anthony Sandrin | |
Tensor dimensions needed to be reversed since RNN function expect stride 1 in the first dimension. Remove experimental RNN scripts. Make inputSize a parameter to the RNN layer and use strings for enum values. * add missing cutorch.sychronize() to SpatialConvolution benchmark * rename bench to benchSpatial * add benchVolumetric Set weights in __init() Remove unnecessary input argument to resetIODescriptors - added new convolution modes - cleaned up benchVolumetric Add missing synchronization test reverted to non-verbose and random optimizer Return output from RNN:updateOutput and zero cell output. cuDNN does not set cell output for RNN_RELU so the output was garbage. Check for nil cx/hx when not training. Correct self.RNNDesc to self.rnnDesc | |||
2016-04-18 | Update TemporalConvolution.lua | ngimel | |
Cleaner way of what fbesse was doing (thans fbesse!) and typo fix | |||
2016-04-18 | Initial work for cudnn RNN api integration | Anthony Sandrin | |
Added cudnnFind auto-tuning Change RNN layer api and improve descriptor/tensor resizing conditions Implement updateGradInput and accGradParameters | |||
2016-04-18 | Fixing backward pass of TemporalConvolution | Frederic Besse | |
2016-04-18 | R5 initial port | Boris Fomitchev | |
Initial adjustment of internall cudnn calls Got rid of _v3 functions | |||
2016-04-18 | Created branch for CUDNN5 work | Boris Fomitchev | |
2016-04-16 | Merge pull request #167 from szagoruyko/bn-convert-2 | Soumith Chintala | |
[R5] cudnn.convert for BatchNorm | |||
2016-04-16 | cudnn.convert for BN | Sergey Zagoruyko | |
2016-04-15 | add read for BatchNorm | Sergey Zagoruyko | |
2016-04-14 | Merge pull request #165 from szagoruyko/r5-bn | Sergey Zagoruyko | |
[R5] Fix running_var in BatchNorm | |||
2016-04-14 | fix running_var meaning in BN | Sergey Zagoruyko | |
2016-04-13 | R5 rebase | Sergey Zagoruyko | |
2016-04-13 | Merge pull request #162 from szagoruyko/utils-clear | Sergey Zagoruyko | |
Use nn.utils.clear on temporary buffers | |||
2016-04-11 | Merge pull request #163 from hughperkins/notes-on-benchmark-fastest | Soumith Chintala | |
add notes on benchmark and fastest | |||
2016-04-11 | add notes on benchmark and fastest | Hugh Perkins | |
2016-04-08 | use nn.utils.clear | Sergey Zagoruyko | |
2016-03-31 | Merge pull request #158 from szagoruyko/error-handling | Soumith Chintala | |
Improve error handling | |||
2016-03-30 | print function called on error | Sergey Zagoruyko | |
2016-03-30 | Merge pull request #152 from gheinrich/fix-temporal-convolution | Soumith Chintala | |
Fix TemporalConvolution output size | |||
2016-03-30 | Merge pull request #157 from szagoruyko/gradinput-init-fix | Soumith Chintala | |
Move gradInput resize to updateGradInput (SpatialConvolution) | |||
2016-03-30 | move gradInput resize to updateGradInput | Sergey Zagoruyko | |
2016-03-29 | Merge pull request #147 from lospooky/avgpooling-count_include_pad | Soumith Chintala | |
Fixed cudnn -> nn avg-pooling conversion | |||
2016-03-29 | Fixed cudnn -> nn avg-pooling conversion | Simone Cirillo | |
cudnn.convert now properly initializes nn.SpatialAveragePooling.count_include_pad | |||
2016-03-25 | Fix TemporalConvolution output size | Greg Heinrich | |
Issue seen in updateOutput() when batch size is decreasing (e.g. 256->128): ``` .../torch/install/share/lua/5.1/torch/Tensor.lua:462: Wrong size for view. Input size: 66060288. Output size: 128x256x1008x1 stack traceback: [C]: in function 'error' .../torch/install/share/lua/5.1/torch/Tensor.lua:462: in function 'view' ...orch/install/share/lua/5.1/cudnn/TemporalConvolution.lua:62: in function <...orch/install/share/lua/5.1/cudnn/TemporalConvolution.lua:54> ``` See new test cudnntest.TemporalConvolution_reduceBatchSize() for a repro. Calling a tensor's set() method without extra parameters but the requested storage causes the tensor to see a 1-D view of the full storage (not the associated tensor). For example: ``` th> x=torch.Tensor(10) [0.0001s] th> y=x:resize(5) [0.0001s] th> torch.Tensor():set(y:storage()):size() 10 [torch.LongStorage of size 1] [0.0002s] ``` Proposed fix is to specify the desired view through the optional parameters of the set() method. | |||
2016-03-24 | Merge pull request #150 from szagoruyko/convert-fixes | Soumith Chintala | |
Fix saving cudnn->nn converted models | |||
2016-03-24 | fix conversion for saved cudnn->nn | Sergey Zagoruyko | |
2016-03-24 | Merge pull request #149 from borisfom/fullconv | Soumith Chintala | |
fix failing SpatialFullConvolution test | |||
2016-03-24 | fix failing SpatialFullConvolution test | Natalia Gimelshein | |
2016-03-22 | Merge pull request #148 from szagoruyko/conv-pad-check | Soumith Chintala | |
Check if padW and padH exist in SpatialConvolution | |||
2016-03-22 | check if padW and padH exist | Sergey Zagoruyko | |