Age | Commit message (Collapse) | Author | |
---|---|---|---|
2017-05-30 | Merge pull request #71 from SeanNaren/ctc-refactorHEADmaster | Nicholas Léonard | |
Simplify Batch First gradients for CTCCriterion | |||
2017-05-30 | Simplified batchfirst gradient modification | SeanNaren | |
2017-05-24 | Merge pull request #70 from nicholas-leonard/ZeroGrad | Nicholas Léonard | |
moved ZeroGrad to nnx | |||
2017-05-24 | moved ZeroGrad to nnx | Nicholas Leonard | |
2017-03-23 | Merge pull request #69 from psychosomaticdragon/master | Soumith Chintala | |
Added pixelsort | |||
2017-03-20 | Added pixelsort | psychosomaticdragon | |
Added pixelsort algorithm, a reversed pixelshuffle. Useful as an alternate downscaling operator - compared to pooling, which discards 3/4 of the image information, pixelsort + conv bottleneck layers discard only half of the information. Also minimizes checkerboard artefacts as compared to strided convolutions. | |||
2017-02-16 | Merge pull request #68 from twitter-forks/master | Nicholas Léonard | |
fix dependency bug | |||
2017-02-16 | fix dependency bug | Nicholas Leonard | |
2017-02-16 | Merge pull request #67 from twitter-forks/add-dpnn-stuff | Nicholas Léonard | |
SoftMaxForest + unit test fixes | |||
2017-02-16 | SoftMaxForest + unit test fixes | Nicholas Leonard | |
2016-07-27 | Merge pull request #64 from NightFury13/master | Soumith Chintala | |
update README.md for changes in CTCCriterion.lua | |||
2016-07-27 | update README.md for changes in CTCCriterion.lua | Mohit Jain | |
2016-05-20 | Merge pull request #62 from SeanNaren/master | Soumith Chintala | |
Divide loss by batch size, initialize 2D activations for CTCCriterion | |||
2016-05-18 | Divide loss by batch size, initialize 2D activations | SeanNaren | |
2016-05-17 | Merge pull request #61 from SeanNaren/master | Soumith Chintala | |
CTCCriterion variable length and batchFirst support | |||
2016-05-13 | Updated Criterion name in tests | SeanNaren | |
2016-05-13 | Updated CTCCriterion with variable lengths and batchFirst | SeanNaren | |
2016-05-12 | Merge pull request #60 from Ark-kun/fixing_cmake_prefix_path | Soumith Chintala | |
Fixing cases when $(LUAROCKS_PREFIX) != $(LUA_BINDIR)/.. | |||
2016-05-12 | Fixing cases when $(LUAROCKS_PREFIX) != $(LUA_BINDIR)/.. | Alexey Volkov | |
2016-04-28 | Merge pull request #58 from gmarceaucaron/master | Soumith Chintala | |
Adding QDRiemaNNLinear, a module for Riemannian gradient descent with the Outer Product metric | |||
2016-04-16 | Merge pull request #59 from SeanNaren/master | Soumith Chintala | |
CTCCriterion optimisation | |||
2016-04-16 | Change creation of sizes and forward pass | SeanNaren | |
2016-04-15 | Removing useless flags for OP metric | GaetanMarceauCaron | |
2016-04-15 | small modif | GaetanMarceauCaron | |
2016-04-15 | small modif | GaetanMarceauCaron | |
2016-04-15 | minor modif | GaetanMarceauCaron | |
2016-04-15 | Adding a default value for gamma depending on the minibatch size | GaetanMarceauCaron | |
2016-04-15 | Code optimization and verification | GaetanMarceauCaron | |
2016-04-14 | Changing the default value for gamma | GaetanMarceauCaron | |
2016-04-13 | reformatting | GaetanMarceauCaron | |
2016-04-13 | reformatting | GaetanMarceauCaron | |
2016-04-13 | Adding the QDRiemaNNLinear module | GaetanMarceauCaron | |
2016-04-13 | Adding the QDRiemaNNLinear module to the nnx package | GaetanMarceauCaron | |
2016-04-13 | Adding the description of the QDRiemaNNLinear module | GaetanMarceauCaron | |
2016-04-09 | Merge pull request #57 from SeanNaren/master | Nicholas Léonard | |
Fixed CPU gradInputs returning zeroes | |||
2016-04-09 | removed type change of float | SeanNaren | |
2016-04-09 | Fixed CPU gradInputs float error | SeanNaren | |
2016-04-02 | Merge pull request #55 from SeanNaren/master | Soumith Chintala | |
CTC Based Criterion | |||
2016-04-02 | Put require warp_ctc into constructor | SeanNaren | |
2016-04-02 | fixed accidental formatting of other tests | SeanNaren | |
2016-04-02 | added base tests for CTCCriterion | Sean | |
2016-04-02 | added criterion to init | Sean | |
2016-04-02 | Rephrased alignment for CTC | Sean Naren | |
2016-04-02 | added more documentation and assertion check on input dim | Sean | |
2016-04-02 | Merge remote-tracking branch 'origin/master' | Sean | |
2016-04-02 | added base CTCCriterion | Sean | |
2016-04-02 | Added CTCCriterion docs | Sean Naren | |
2016-03-06 | Merge pull request #53 from GeorgOstrovski/patch-1 | Soumith Chintala | |
Change maximal tensor size in test | |||
2016-03-06 | Change maximal tensor size in test | Georg Ostrovski | |
With the previous setting, the output size of the module could be quite large, and the Jacobian test below (which uses O(N*M) memory for N, M the module in/output size) could run out of memory (taking up to several GB). | |||
2016-02-19 | Merge pull request #52 from sergomezcol/master | Soumith Chintala | |
Extra checks and cleanup in C code |