Skip to content
Snippets Groups Projects
Select Git revision
  • master default
  • v0.3.5
  • v0.2.3
  • v0.2.2
4 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.017Dec426Nov29Oct1715126Aug2520182Jul18Jun17228May2723Apr25Mar43Feb18Dec17109release v0.4.15mastermasterclearified doc2vec algorithm parameterrelease v0.4.14: added counter contributorfixed mypy issuesfixed mypy issuesadded counter contributor, removed cli (never worked)tiny refactorrelease v0.4.13added pqgram-based string kernelsadded possibility to use 2-dimensional input arrays, as of Scikit specifications.refactored contributors to allow entire dataframes to be passed, and also columns to be deleted.switched to CPU for torch testsadded date debugs for pipelineadded option to specify where the unified dataset should be stored for reddit to common scriptrelease v0.4.11: added tests for StanzaToNlpFieldTransformer, changed output to nested list instead of strings.release v0.4.10release v0.4.9: Enable non-string svm kernelsblacksped up string kernel implementationrelease v0.4.7release v0.4.6enable string kernels to be used in SVM classrelease v0.4.5release v0.4.4enable auto-detection of cudasimplified cnn model to be able to use cuda, i think.fixed lstm model, moved seq length to model-specific paramsreleased v0.4.2added calculation of first fc layer sizealmost there! the first value of the fc_configs is pre-determined by the last config of the conv-layers, but not sure how.added wrapper for skorch models to not need to specify classes on init, but on fitrelease v0.4.0added mypy, removed flake8, removed tensorflowrelease v0.3.8fixed bug in writing non-existing parsed filesrelease v0.3.7removed error dumping - not generating parse creates other problemsrelease v0.3.6fixed bug, added error dumpingrelease v0.3.5v0.3.5v0.3.5
Loading