collapse
80%
WDescription%
Build stability: 1 out of the last 5 builds failed.80
Build History
x
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Changes

#150 (Mar 5, 2017 12:08:12 PM)

  1. Update release guide — noreply / githubweb

#148 (Mar 2, 2017 1:53:11 PM)

  1. Adding scripts to run some example pipelines. — sparks / githubweb
  2. Making mnist things into env variables. — sparks / githubweb
  3. Moving VOC example to use the local filesystem. — sparks / githubweb

#147 (Feb 27, 2017 10:28:13 PM)

  1. Refactoring to prepend keystoneml to the beginning of all package — sparks / githubweb
  2. Cleaning up comments. — sparks / githubweb

#146 (Feb 27, 2017 8:34:21 AM)

  1. Adding Evaluator interface and updating pipelines to use it. — sparks / githubweb
  2. Fixing line length. — sparks / githubweb
  3. Evaluator.apply -> Evaluator.evaluate. — sparks / githubweb
  4. Adding scaladoc and cleaning up existing scaladoc. — sparks / githubweb

#145 (Feb 23, 2017 11:28:14 PM)

  1. Bumping default scala version number. — sparks / githubweb
  2. Bumping back down to scala 2.10 — sparks / githubweb
  3. Version with only epic-related unit tests broken. — sparks / githubweb
  4. Removing NER and POS Tagging. — sparks / githubweb
  5. Adding support for cross building and bumping sbt version. — sparks / githubweb

#144 (Jul 19, 2016 12:38:12 PM)

  1. Add Kernel generators, solvers and model Also include a cifar pipeline — shivaram / githubweb
  2. Add unit test — shivaram / githubweb
  3. Address code review comments — shivaram / githubweb
  4. More code review comments — shivaram / githubweb
  5. Add option to cache kernel matrix — shivaram / githubweb
  6. Changes to avoid recomputing kernel twice — shivaram / githubweb
  7. Add a CIFAR augmented kernel pipeline — shivaram / githubweb
  8. Update augmented kernel pipeline after merge — shivaram / githubweb
  9. Refactor KernelMatrix API to improve performance This changes the — shivaram / githubweb
  10. Fix a bug, update unit test — shivaram / githubweb
  11. Handle diagonal block for non-square case — shivaram / githubweb
  12. Cache DenseMatrix version to avoid some GC — shivaram / githubweb
  13. Add a shuffler before we generate the kernel — shivaram / githubweb
  14. Shuffle labels, images together — shivaram / githubweb
  15. Make checkpoint dir configurable — shivaram / githubweb
  16. Handle empty data partitions — shivaram / githubweb
  17. Describe terms used in kernel generator — shivaram / githubweb

#143 (Jun 24, 2016 11:28:14 PM)

  1. Handle empty partitions in ReWeightedLeastSquares — shivaram / githubweb
  2. Assert gradient to be small — shivaram / githubweb

#142 (Jun 15, 2016 3:53:11 PM)

  1. Exclude slf4j-simple from Scala NLP deps — shivaram / githubweb

#141 (Jun 10, 2016 1:23:15 PM)

  1. Moving old files to old/ — sparks / githubweb
  2. Renaming workflow/graph/ files to workflow/ — sparks / githubweb
  3. All of workflow compiles. — sparks / githubweb
  4. Moving Cacher back to nodes.util. — sparks / githubweb
  5. Unmarking toPipeline as private and fixing evaluators/timit pipeline. — sparks / githubweb
  6. Text pipelines now compile. — sparks / githubweb
  7. VOC compiles and imagenet almost compiles. — sparks / githubweb
  8. Fixing estimators and causing all things to compile. — sparks / githubweb
  9. Moving tests to old/ — sparks / githubweb
  10. Moving workflow/graph tests to workflow/ — sparks / githubweb
  11. Compiling test cases. Biggest refactor was pipelines.PipelineContext => — sparks / githubweb
  12. Removing unnecessary file. — sparks / githubweb
  13. Getting LeastSquaresEstimator test to pass. — sparks / githubweb
  14. Removing last traces of old backend. — sparks / githubweb
  15. Fixing style issues. — sparks / githubweb
  16. Fixing PR comments. — sparks / githubweb

#140 (Jun 5, 2016 3:38:15 PM)

  1. Fix broken build by adding ClassTag to ChainUtils — shivaram / githubweb

#139 (Jun 5, 2016 1:48:17 PM)

  1. moving over the autocacherule — tomer.kaftan / githubweb
  2. WIP autocacherule and nodeoptimizationrule — tomer.kaftan / githubweb
  3. wrote NodeOptimizationRule, autocache rule still in progress — tomer.kaftan / githubweb
  4. AutoCacheRule should match master's — tomer.kaftan / githubweb
  5. NodeOptRule PR response — tomer.kaftan / githubweb
  6. Some dedup and extra unit tests for autocacherule — tomer.kaftan / githubweb
  7. did some autocacherule dedup — tomer.kaftan / githubweb
  8. Chain utils — tomer.kaftan / githubweb

#138 (Jun 1, 2016 2:33:14 PM)

  1. added optimizers and todot method for graph data structure — tomer.kaftan / githubweb
  2. progress on execution — tomer.kaftan / githubweb
  3. implement some of the methods — tomer.kaftan / githubweb
  4. More user api wip — tomer.kaftan / githubweb
  5. more impl hacking — tomer.kaftan / githubweb
  6. sort of made progress — tomer.kaftan / githubweb
  7. first impl of new node merge rule — tomer.kaftan / githubweb
  8. yet more progress — tomer.kaftan / githubweb
  9. gradual progress on the autocache rule — tomer.kaftan / githubweb
  10. more gradual progress updating the autocache rule — tomer.kaftan / githubweb
  11. code compiles — tomer.kaftan / githubweb
  12. added a bit of extra bookkeeping to graph combining — tomer.kaftan / githubweb
  13. Gradual progress on incremental pipeline construction continues — tomer.kaftan / githubweb
  14. more progress on incremental construction — tomer.kaftan / githubweb
  15. More progress, updated equivalent node merge — tomer.kaftan / githubweb
  16. some bug fixing and unit testing — tomer.kaftan / githubweb
  17. test progress — tomer.kaftan / githubweb
  18. subtle method changes — tomer.kaftan / githubweb
  19. split some complex methods — tomer.kaftan / githubweb
  20. gradual progress — tomer.kaftan / githubweb
  21. made optimizer decision global — tomer.kaftan / githubweb
  22. initial interface unification — tomer.kaftan / githubweb
  23. commit currently commented out autocache rule & suite — tomer.kaftan / githubweb
  24. more unit test updates — tomer.kaftan / githubweb
  25. some more test suite updates — tomer.kaftan / githubweb
  26. added incremental label estimator unit test — tomer.kaftan / githubweb
  27. Finished a Pipeline.submit unit test — tomer.kaftan / githubweb
  28. Added Pipeline.gather — tomer.kaftan / githubweb
  29. More comments & cleanup — tomer.kaftan / githubweb
  30. More comments & cleanup — tomer.kaftan / githubweb
  31. More comments & cleanup — tomer.kaftan / githubweb
  32. Moved Transformer to its own file — tomer.kaftan / githubweb
  33. moved autocache stuff out of this branch — tomer.kaftan / githubweb
  34. removed untested checkpointing code from this PR — tomer.kaftan / githubweb
  35. more code cleanup — tomer.kaftan / githubweb
  36. more code cleanup — tomer.kaftan / githubweb
  37. more cleanup & documentation — tomer.kaftan / githubweb
  38. more cleanup & documentation — tomer.kaftan / githubweb
  39. Split files into separate classes — tomer.kaftan / githubweb
  40. more docs progress — tomer.kaftan / githubweb
  41. more docs progress — tomer.kaftan / githubweb
  42. more docs progress — tomer.kaftan / githubweb
  43. more documentation — tomer.kaftan / githubweb
  44. more documentation — tomer.kaftan / githubweb
  45. more documentation — tomer.kaftan / githubweb
  46. extra bit of code cleanup — tomer.kaftan / githubweb
  47. more docs — tomer.kaftan / githubweb
  48. introduced prefix — tomer.kaftan / githubweb
  49. gradual prefix replacement process — tomer.kaftan / githubweb
  50. more cruft removal and redoing things with prefixes — tomer.kaftan / githubweb
  51. got current code to compile — tomer.kaftan / githubweb
  52. more progress on prefix load — tomer.kaftan / githubweb
  53. made executor save state if non-empty prefix for a node — tomer.kaftan / githubweb
  54. more prefix progress — tomer.kaftan / githubweb
  55. got prefix stuff to work — tomer.kaftan / githubweb
  56. updated test — tomer.kaftan / githubweb
  57. updated tests that manually set an optimizer — tomer.kaftan / githubweb
  58. Changes to method naming that Evan requested — tomer.kaftan / githubweb
  59. Changes to method naming that Evan requested — tomer.kaftan / githubweb
  60. more renaming — tomer.kaftan / githubweb
  61. Added Pipeline.fit — tomer.kaftan / githubweb
  62. Added more documentation — tomer.kaftan / githubweb
  63. Added more documentation — tomer.kaftan / githubweb
  64. renaming — tomer.kaftan / githubweb
  65. renamed Prefix.get — tomer.kaftan / githubweb
  66. responses to PR comments — tomer.kaftan / githubweb
  67. Bug fix for sbt/sbt test FittedPipeline deserialization — tomer.kaftan / githubweb
  68. indent fix — tomer.kaftan / githubweb

#137 (May 29, 2016 1:08:11 PM)

  1. Samples GMM ref wasn't released in EncEval code — vaishaal / githubweb

#136 (May 17, 2016 11:38:14 AM)

  1. Revert "Strides added to convolver and unit tests for strides" — evan.sparks / githubweb

#131 (May 17, 2016 8:38:12 AM)

  1. Strides added to convolver and unit tests for strides — vaishaal / githubweb
  2. fixed failing tests cause of normalization — vaishaal / githubweb
  3. Evan PR comments — vaishaal / githubweb
  4. Perf regression fix — vaishaal / githubweb

#130 (May 6, 2016 2:53:13 PM)

  1. Named entity recognition — ngarneau / githubweb
  2. Generic type for the NER and little typo — ngarneau / githubweb

#128 (May 2, 2016 6:27:03 PM)

  1. POS Tagging — ngarneau / githubweb
  2. Testing the POSTagger with Spark — ngarneau / githubweb
  3. Rearrange tests so they reuse the model — ngarneau / githubweb
  4. Change the broadcast logic — ngarneau / githubweb
  5. Making the POS Tagger model transient — ngarneau / githubweb

#127 (May 2, 2016 6:13:13 PM)

  1. progress on utils for a new internal graph format — tomer.kaftan / githubweb
  2. more utils progress — tomer.kaftan / githubweb
  3. more utils progress — tomer.kaftan / githubweb
  4. slightly more utils progress — tomer.kaftan / githubweb
  5. Added impl for the combine util — tomer.kaftan / githubweb
  6. Added impl for the linearize util — tomer.kaftan / githubweb
  7. use a sensical replaceNodes method — tomer.kaftan / githubweb
  8. slight interface change — tomer.kaftan / githubweb
  9. internal reorg progress — tomer.kaftan / githubweb
  10. underlying DAG utils in place — tomer.kaftan / githubweb
  11. added Ops — tomer.kaftan / githubweb
  12. Added some laziness to the graph operators — tomer.kaftan / githubweb
  13. Moved internals to separate workspace — tomer.kaftan / githubweb
  14. Split analysis methods into separate utils — tomer.kaftan / githubweb
  15. Added docs to Outputs, Operators, and GraphIds — tomer.kaftan / githubweb
  16. Added operator unit testing — tomer.kaftan / githubweb
  17. renamed internals to internal — tomer.kaftan / githubweb
  18. progress on adding documentation — tomer.kaftan / githubweb
  19. Finished documentation for graph and analysis utils — tomer.kaftan / githubweb
  20. Added unit tests for the graph analysis utils — tomer.kaftan / githubweb
  21. more graph unit tests implemented — tomer.kaftan / githubweb
  22. implemented a few more unit tests — tomer.kaftan / githubweb
  23. More progress on the graph method unit tests — tomer.kaftan / githubweb
  24. replaceNodes bug fixes and finished up unit tests — tomer.kaftan / githubweb
  25. response to pr comments — tomer.kaftan / githubweb
  26. added some extra testing — tomer.kaftan / githubweb

#126 (Apr 28, 2016 9:53:10 PM)

  1. Pointing to objects located on S3. — sparks / githubweb

#125 (Apr 6, 2016 5:28:16 PM)

  1. Adding Eric an Stephen to list of contributors. — evan.sparks / githubweb

#118 (Apr 6, 2016 10:58:11 AM)

  1. Made autocaching MUCH faster for complex DAGs — tomer.kaftan / githubweb

#117 (Mar 30, 2016 2:18:12 PM)

  1. Allowed DAG branches to be taken lazily — tomer.kaftan / githubweb
  2. Added a unit test — tomer.kaftan / githubweb

#111 (Mar 24, 2016 2:18:12 PM)

  1. Bumping default spark and hadoop versions. — sparks / githubweb

#110 (Mar 23, 2016 6:23:10 PM)

  1. Fixing up EC2 launch documentation. — sparks / githubweb

#109 (Mar 22, 2016 9:30:05 AM)

  1. Added optimizable gmm fisher vector, column pca estimators — tomer.kaftan / githubweb
  2. proper cost model for pca and style nits — tomer.kaftan / githubweb
  3. Renamed CostModel trait and added pca optimizer unit tests — tomer.kaftan / githubweb
  4. made pipelines use the new optimizable pca & gmm fisher vector — tomer.kaftan / githubweb

#108 (Mar 22, 2016 9:18:13 AM)

  1. Add CIFAR augmentation pipeline and nodes for it — shivaram / githubweb
  2. Add random flipper to pipeline — shivaram / githubweb
  3. Add evaluator for augmented pipeline — shivaram / githubweb
  4. Remove sumPooler function — shivaram / githubweb
  5. Fix code review comments — shivaram / githubweb
  6. More code review comments. Also introduce a new API — shivaram / githubweb
  7. Remove unused RandomFlipper — shivaram / githubweb
  8. Add test cases for new nodes — shivaram / githubweb
  9. Style fix — shivaram / githubweb
  10. Fixes from local cluster run — shivaram / githubweb
  11. Style fixes — shivaram / githubweb
  12. Return a pipeline object — shivaram / githubweb
  13. Try to avoid recomputation — shivaram / githubweb
  14. Avoid recomputation in other CIFAR pipelines — shivaram / githubweb
  15. Code review fixes — shivaram / githubweb

#107 (Mar 21, 2016 2:58:12 PM)

  1. Updated the KeystoneML release guide — tomer.kaftan / githubweb
  2. referred to makefile in the release guide — tomer.kaftan / githubweb
  3. added some more info to the release guide — tomer.kaftan / githubweb

#106 (Mar 21, 2016 2:38:11 PM)

  1. Added the optimizable least squares estimator — tomer.kaftan / githubweb
  2. Some responses to PR — tomer.kaftan / githubweb

#105 (Mar 18, 2016 1:08:11 PM)

  1. progress on the generalization of node-level optimizations — tomer.kaftan / githubweb
  2. Node-level optimization generalization working — tomer.kaftan / githubweb
  3. Added another common sub-expression elimination stage after the — tomer.kaftan / githubweb
  4. bug fix and optimization for when to execute or to optimize things in — tomer.kaftan / githubweb
  5. made node optimization use a fixed num per partition instead of — tomer.kaftan / githubweb
  6. split some code into separate files and removed some mutability — tomer.kaftan / githubweb
  7. added a bit of documentation — tomer.kaftan / githubweb
  8. response to pr comments — tomer.kaftan / githubweb
  9. added more node optimization comments — tomer.kaftan / githubweb

#104 (Mar 16, 2016 12:38:13 PM)

  1. autocaching — tomer.kaftan / githubweb
  2. more autocaching updates — tomer.kaftan / githubweb
  3. Some bug fixes, introduced caching strategy — tomer.kaftan / githubweb
  4. Added caching strategies (pre-unit tests) — tomer.kaftan / githubweb
  5. Updates to deal correctly with the case where an instruction has — tomer.kaftan / githubweb
  6. Added autocache rule unit tests — tomer.kaftan / githubweb
  7. Fix to style nit — tomer.kaftan / githubweb
  8. added scaladoc and renamed naive to aggressive — tomer.kaftan / githubweb
  9. Switched to specifying partitionScale instead of scale in the autocache — tomer.kaftan / githubweb
  10. Added driver memory and estimator fit profiling to the auto-cacher — tomer.kaftan / githubweb
  11. Added autocache to the default optimizer and made it estimate a memory — tomer.kaftan / githubweb
  12. Added code to take into account existing Cacher nodes — tomer.kaftan / githubweb
  13. Added some extra logging for autocaching — tomer.kaftan / githubweb
  14. Made sure to treat driver correctly in memory budget calculation — tomer.kaftan / githubweb
  15. Made the autocache rule more efficient in profiling — tomer.kaftan / githubweb
  16. Made the default optimizer not auto-cache — tomer.kaftan / githubweb
  17. Fixed a bug where executing a pipeline on test data forces it to be fit — tomer.kaftan / githubweb

#103 (Mar 13, 2016 11:08:14 PM)

  1. Returning a 'Pipeline' from run method of existing examples. — sparks / githubweb

#102 (Mar 12, 2016 11:18:12 AM)

  1. Replace rowsToMatrix with rowsToMatrixIter This is safer for handling — shivaram / githubweb

#101 (Mar 12, 2016 11:03:12 AM)

  1. Add a weighted block solver for small num classes This is implemented — shivaram / githubweb
  2. Remove whitespace — shivaram / githubweb
  3. Use local JFM instead of Spark job — shivaram / githubweb
  4. Refactor weighted least squares code. — shivaram / githubweb

#100 (Mar 12, 2016 10:33:11 AM)

  1. broadcasted cosine random features — tomer.kaftan / githubweb

#99 (Mar 11, 2016 5:28:11 PM)

  1. Add release notes for snapshots — shivaram / githubweb
  2. Address code review comments — shivaram / githubweb

#97 (Mar 10, 2016 3:08:13 PM)

  1. Switching norm from singular to frobenius due to conditioning issues. — sparks / githubweb

#94 (Mar 10, 2016 10:58:14 AM)

  1. Fixing build non-determinism. Addresses #235 — sparks / githubweb
  2. Removing unnecessary implicit. — sparks / githubweb
  3. Removing one more unnecessary implicit. — sparks / githubweb

#91 (Feb 29, 2016 1:08:10 PM)

  1. Fixes to get builds to work again — tomer.kaftan / githubweb
  2. Made coord protected — tomer.kaftan / githubweb

#90 (Feb 20, 2016 12:08:10 PM)

  1. Replace LLH computation with a while loop — shivaram / githubweb
  2. Address code review comment — shivaram / githubweb
  3. Fix bug in PR comment — shivaram / githubweb

#89 (Feb 19, 2016 11:53:12 AM)

  1. Adding experimental image iteration interface and a suite of — sparks / githubweb
  2. Trying alternative ordering for makePatches. — sparks / githubweb
  3. Adding logic to summarise convolution results over 10 runs. — sparks / githubweb
  4. Removing commented out code. — sparks / githubweb
  5. Updating convolutions and adding a new convolver interface. — sparks / githubweb
  6. Adding comparison script for convolver. — sparks / githubweb
  7. Reintroducing speedups. — sparks / githubweb
  8. Fixing output order to match loop order. — sparks / githubweb
  9. Removing copy out loop. — sparks / githubweb
  10. Cleaning up unused/debug code. — sparks / githubweb
  11. Actually removing the correct file. — sparks / githubweb
  12. Updating image iterator code and adding some tests. — sparks / githubweb
  13. Fixing spacing issues. — sparks / githubweb
  14. Addressing PR comments. — sparks / githubweb
  15. Moving image flip routine to ImageUtils. — sparks / githubweb

#88 (Feb 9, 2016 11:28:10 AM)

  1. made commonsparsefeatures more scalable at large numbers of machines — tomer.kaftan / githubweb
  2. made commonsparsefeatures use priority queue per-partition — tomer.kaftan / githubweb
  3. added explicit guava dependency — tomer.kaftan / githubweb

#87 (Feb 3, 2016 7:33:12 PM)

  1. Added a node-level optimization rule — tomer.kaftan / githubweb
  2. Attempt to add more type safety to instruction interpretation — tomer.kaftan / githubweb
  3. Fixes to Evan's style nits — tomer.kaftan / githubweb

#86 (Feb 1, 2016 2:03:09 PM)

  1. renamed the optimizer — tomer.kaftan / githubweb
  2. made equivalent node merger output deterministic node ids — tomer.kaftan / githubweb
  3. changed delegatingtransformer vs transformer relationship slightly — tomer.kaftan / githubweb
  4. Added Instruction class — tomer.kaftan / githubweb
  5. Added a pipeline to instruction Sequence utility — tomer.kaftan / githubweb
  6. Added an instructions to pipeline method — tomer.kaftan / githubweb
  7. bug fix in pipeline to instructions — tomer.kaftan / githubweb
  8. minor PR style fixes — tomer.kaftan / githubweb
  9. RuleExector -> Optimizer and improved pipelineToInstructions method — tomer.kaftan / githubweb
  10. made the instructionsToPipeline method functional-style w/ no — tomer.kaftan / githubweb
  11. Added a unit test for pipeline instruction conversion — tomer.kaftan / githubweb

#85 (Jan 28, 2016 10:48:10 AM)

  1. Adding logic to set OMP_NUM_THREADS appropriately on master and — sparks / githubweb
  2. Addressing PR comments and fixing bugs after testing on Linux. — sparks / githubweb
  3. Removing spurious echo. — sparks / githubweb

#84 (Jan 27, 2016 2:55:24 PM)

  1. Adding class label indicators validation. — sparks / githubweb
  2. Fixing PR comments. — sparks / githubweb

#83 (Jan 27, 2016 2:48:13 PM)

  1. moved over shiv's least squares solvers from my keystone-vw-experiments — tomer.kaftan / githubweb
  2. Did a bunch of refactoring to merge dense and sparse LBFGS at Shiv's — tomer.kaftan / githubweb
  3. Added sparse lbfgs tests — tomer.kaftan / githubweb
  4. Added scaladocs for the gradients — tomer.kaftan / githubweb
  5. Added bias learning to the sparse lbfgs solver — tomer.kaftan / githubweb
  6. regularization bug fix and deeper lbfgs intercept unit testing — tomer.kaftan / githubweb
  7. added another test for local least squares — tomer.kaftan / githubweb
  8. removed license info and developerapi references — tomer.kaftan / githubweb

#82 (Jan 27, 2016 1:18:11 PM)

  1. Handle empty partitions in WeightedLeastSquares Fixes #196 — shivaram / githubweb

#81 (Jan 27, 2016 11:43:13 AM)

  1. Adding image cropper class. — sparks / githubweb

#80 (Jan 20, 2016 12:23:13 PM)

  1. WIP: add crop to image utils — tu.stephenl / githubweb
  2. get test working for crop — stephent / githubweb

#79 (Jan 20, 2016 12:03:13 PM)

  1. moving over the approximate & distributed pca nodes — tomer.kaftan / githubweb
  2. debugging the test — tomer.kaftan / githubweb
  3. added the approximate pca unit test — tomer.kaftan / githubweb
  4. response to pull request comments — tomer.kaftan / githubweb
  5. made pca matlab convention not a unit method — tomer.kaftan / githubweb
  6. Adding Approximate PCA tests. — sparks / githubweb
  7. Removing unnecessary line. — sparks / githubweb

#78 (Jan 20, 2016 11:43:14 AM)

  1. Fixing daisy feature ordering to match ImageNet. Closes #186. — sparks / githubweb
  2. Fixing VectorSplitter irregular block size issue and adding unit test. — sparks / githubweb
  3. Making sure recombined vectors match input and fixing test formatting — sparks / githubweb
  4. Adding test to make sure daisy and sift conform to the same expected — sparks / githubweb
  5. Removing spurious line. — sparks / githubweb

#77 (Jan 20, 2016 11:18:13 AM)

  1. Changing default regularization parameter as this can lead to bad — sparks / githubweb
  2. Adding link to algorithm description. — sparks / githubweb
  3. Adding whiteningRegularization option to cifar pipeline. — sparks / githubweb

#76 (Jan 13, 2016 8:53:10 AM)

  1. Handle non-uniform blocks in WeightedLeastSquares Fixes #195 — shivaram / githubweb

#75 (Jan 7, 2016 4:13:10 PM)

  1. Made the GMM Estimator take a seed and have configurable initialization — tomer.kaftan / githubweb
  2. Response to pull request comments — tomer.kaftan / githubweb

#74 (Jan 5, 2016 1:43:09 PM)

  1. Made the 2-center GMM test resilient to nondeterminism in the returned — tomer.kaftan / githubweb

#72 (Dec 22, 2015 12:13:12 PM)

  1. moved over scala GMM & Fisher Vectors & K-means++ nodes from other — tomer.kaftan / githubweb
  2. added the scala gmm and k-means tests, and made default gmm global — tomer.kaftan / githubweb
  3. Added comments, improved GMM tests — tomer.kaftan / githubweb
  4. Moved file names and locations around (putting c code in .external) — tomer.kaftan / githubweb

#71 (Dec 20, 2015 11:23:09 AM)

  1. Added hashing trick nodes — tomer.kaftan / githubweb
  2. added reference to the scala util hashing code — tomer.kaftan / githubweb

#68 (Nov 25, 2015 1:18:11 PM)

  1. fix for matrixutils that matters in distributed/serialized environments — tomer.kaftan / githubweb
  2. Added a comment about breeze lazy slicing — tomer.kaftan / githubweb

#67 (Nov 25, 2015 10:28:12 AM)

  1. Fix residual mean computation The previous approach of computing — shivaram / githubweb
  2. Move mean computation to a utils function. Add a unit test as well. — shivaram / githubweb

#66 (Oct 7, 2015 2:28:12 PM)

  1. Apply the optimizer automatically — tomer.kaftan / githubweb

#65 (Oct 5, 2015 5:53:13 PM)

  1. Create CNAME — evan.sparks / githubweb
  2. Whoops — evan.sparks / githubweb

#64 (Sep 29, 2015 8:55:46 AM)

  1. Moved over the Amazon pipeline — tomer.kaftan / githubweb
  2. Fixes to pull request comments — tomer.kaftan / githubweb
  3. did some funky stuff for the logistic regression to keep it from caching — tomer.kaftan / githubweb
  4. maybe a cleanup — tomer.kaftan / githubweb
  5. transitioned over the unit tests from mllib for the logistic regression — tomer.kaftan / githubweb
  6. cleaned up the unit test imports — tomer.kaftan / githubweb
  7. removed the addintercept param for now and updated the unit test — tomer.kaftan / githubweb

#63 (Sep 22, 2015 10:53:10 AM)

  1. voc bug fix in num images per column — tomer.kaftan / githubweb

#62 (Sep 18, 2015 4:18:10 PM)

  1. fix the matrixml dependency — zhangzhao0718 / githubweb
  2. mlmatrix also excludes spark — zhangzhao0718 / githubweb

#61 (Sep 17, 2015 10:33:12 PM)

  1. changed to 0.3.0 snapshot — tomer.kaftan / githubweb

#60 (Sep 17, 2015 2:23:12 PM)

  1. changed to 0.3.0 snapshot — tomer.kaftan / githubweb

#59 (Sep 17, 2015 2:08:11 PM)

  1. fixed makefile — tomer.kaftan / githubweb

#58 (Sep 17, 2015 11:33:11 AM)

  1. updated version to 0.2 — tomer.kaftan / githubweb

#57 (Sep 17, 2015 10:53:13 AM)

  1. fixed ml-matrix dep and removed repos — tomer.kaftan / githubweb

#56 (Sep 16, 2015 5:38:11 PM)

  1. SBT updates for sonatype release — tomer.kaftan / githubweb
  2. updated the keystone scripts — tomer.kaftan / githubweb

#55 (Sep 14, 2015 3:03:10 PM)

  1. Made VOC & Imagenet pipelines fully build a DAG — tomer.kaftan / githubweb
  2. builds a full DAG for the TIMIT pipeline — tomer.kaftan / githubweb
  3. Added TODO w/ regards to GMM Fisher Vector — tomer.kaftan / githubweb

#54 (Aug 12, 2015 1:33:29 PM)

  1. Make Weighted LS work when missing classes — shivaram / githubweb
  2. Remove whitespace — shivaram / githubweb

#53 (Jul 27, 2015 4:08:24 PM)

  1. Following redirects and failing on 404 with missing jars. — sparks / githubweb

#52 (Jul 16, 2015 7:03:13 PM)

  1. timit data loader bug fix — tomer.kaftan / githubweb

#51 (Jun 24, 2015 7:03:11 PM)

  1. Made the sparse feature estimators deterministic (to prevent bugs when repeated in the Pipeline DAG) — tomer.kaftan / githubweb

#50 (Jun 24, 2015 2:13:10 PM)

  1. Initial DAG construction stuff — tomer.kaftan / githubweb
  2. Moved all DAG stuff into a separate "workflow" package and marked all the unsafe methods package-private — tomer.kaftan / githubweb
  3. Moved chaining logic into Pipeline from Node, added implicit from node to pipeline, and added transformer methods so transformers can be chained w/o fit — tomer.kaftan / githubweb
  4. Moved node => pipeline implicit into a companion object for node — tomer.kaftan / githubweb
  5. Progress on Concat, made imports point to workflow, disabled direct fit on data — tomer.kaftan / githubweb
  6. Rewrote MnistRandomFFT using the concat operator — tomer.kaftan / githubweb
  7. Generalized the Scatter and added a VectorCombiner — tomer.kaftan / githubweb
  8. playing around with different names for canElevate — tomer.kaftan / githubweb
  9. Fixed some more imports & replaced .fit(data) with .withData(data) — tomer.kaftan / githubweb
  10. fixed some comments, and removed a little bit of exposed interfaces — tomer.kaftan / githubweb
  11. re-exposed some stuff to enable estimator w/data rewriting — tomer.kaftan / githubweb
  12. rewrote Pipeline as a pseudo-case class to force pipeline nodes to be rewritten upon construction — tomer.kaftan / githubweb
  13. changed some interfaces to ensure ScatterNodes & ScatterTransformers are rewritten right away — tomer.kaftan / githubweb
  14. Added some documentation — tomer.kaftan / githubweb
  15. Start on next version of Pipeline DAG — tomer.kaftan / githubweb
  16. Added some logging to the DAG execution — tomer.kaftan / githubweb
  17. Moved things around & repurposed RuleExecutor from Catalyst — tomer.kaftan / githubweb
  18. moved pipeline AndThen out of implicit — tomer.kaftan / githubweb
  19. started work on new revision of the keystone dsl — tomer.kaftan / githubweb
  20. unified Transformers & ConcretePipeline under "pipeline" to improve type inference — tomer.kaftan / githubweb
  21. removed some generics from nodes that didn't need them — tomer.kaftan / githubweb
  22. dsl progress (everything compiles, still need to implement some methods though) — tomer.kaftan / githubweb
  23. bug fixes — tomer.kaftan / githubweb
  24. progress!! — tomer.kaftan / githubweb
  25. Added first rewrite rule — tomer.kaftan / githubweb
  26. Got the gather node working — tomer.kaftan / githubweb
  27. simplified some complex typing — tomer.kaftan / githubweb
  28. Made some of the interface changes discussed with Evan & Shivaram — tomer.kaftan / githubweb
  29. changed interface some more — tomer.kaftan / githubweb
  30. shortened the estimator chaining methods — tomer.kaftan / githubweb
  31. marked DAG components as workflow-private — tomer.kaftan / githubweb
  32. removed classtag from estimators — tomer.kaftan / githubweb
  33. somewhat fleshed out the validate method — tomer.kaftan / githubweb
  34. Response to pull request comments — tomer.kaftan / githubweb
  35. Response to pull request comments — tomer.kaftan / githubweb
  36. Added unit tests — tomer.kaftan / githubweb
  37. Added memoization of data lineage (so caching transformers will work) — tomer.kaftan / githubweb
  38. Response to pull request comments — tomer.kaftan / githubweb
  39. Split long lines across several — tomer.kaftan / githubweb
  40. More response to pull request comments — tomer.kaftan / githubweb

#49 (Jun 10, 2015 7:43:10 PM)

  1. Made what matrix utils I trivially could be generic — tomer.kaftan / githubweb

#48 (Jun 9, 2015 11:43:11 AM)

  1. Fixing bug in unit test -> abs(diff(x,y)) as opposed to diff(abs(x),y) — sparks / githubweb

#47 (Jun 8, 2015 8:58:10 PM)

  1. Fixing bug in ZCA Whitener where it was using a (large) hardcoded epsilon, and fixing issue in unit test that was hiding this. — sparks / githubweb
  2. Adding more comprehensive tests. — sparks / githubweb

#46 (Jun 1, 2015 9:43:12 PM)

  1. Retain meta-inf/services in assembly — shivaram / githubweb

#45 (Jun 1, 2015 3:23:11 PM)

  1. Pick up Spark, Hadoop versions from env — shivaram / githubweb

#44 (May 29, 2015 2:53:12 PM)

  1. Rename script to keystone-ec2.sh — shivaram / githubweb

#43 (May 27, 2015 2:13:10 PM)

  1. Handle arbitrary inputs to BlockWeightedLeastSquares — shivaram / githubweb
  2. Address code review comments — shivaram / githubweb

#42 (May 22, 2015 2:53:12 PM)

  1. Removing redundant call to vl_dsift_get_descriptors. — sparks / githubweb

#41 (May 20, 2015 10:38:09 PM)

  1. Fixing the list of contributors to be better markdown formatting. — sparks / githubweb

#40 (May 19, 2015 5:23:10 PM)

  1. Adding fix for block cifar. — sparks / githubweb
  2. Fixing typo. — sparks / githubweb

#39 (May 19, 2015 11:23:11 AM)

  1. Add ImageNet SIFT+LCS+FV pipeline — shivaram / githubweb
  2. Fix shape of LCS output — shivaram / githubweb
  3. Add ability to pass on numFeatures to VectorSplitter — shivaram / githubweb
  4. VOC: Set stepScale back to zero. — shivaram / githubweb
  5. Add option to pass in numRows to ColumnSampler — shivaram / githubweb
  6. Use Top-5 error, re-use samples for PCA,GMM — shivaram / githubweb
  7. Address code review, add sqrt to sift — shivaram / githubweb
  8. Pull out fisherFeaturizer into a function — shivaram / githubweb

#38 (May 18, 2015 1:38:10 AM)

  1. First cut at NGrams iface and a simple impl. — zongheng.y / githubweb
  2. Rework interface: NGramsFeaturizer and NGramsCounts. — zongheng.y / githubweb
  3. Move NGramSuite under nlp package. — zongheng.y / githubweb
  4. WIP — zongheng.y / githubweb
  5. First cut at StupidBackoffLMPipeline in pipelines style. WIP. — zongheng.y / githubweb
  6. Add last step, StupidBackoffLM. — zongheng.y / githubweb
  7. Add indexers.scala; new test suite; minor fixes. — zongheng.y / githubweb
  8. Add NGramsCounts(mode = noAdd) test. — zongheng.y / githubweb
  9. Add NGramIndexerSuite; other fixes. — zongheng.y / githubweb
  10. Add StupidBackoffSuite (passing). — zongheng.y / githubweb
  11. API update: add StupidBackoffModel. — zongheng.y / githubweb
  12. Make a bunch of nodes case classes. — zongheng.y / githubweb
  13. Scope ngramsBuf into each line. — zongheng.y / githubweb
  14. Address some comments: add docs, rename files. — zongheng.y / githubweb
  15. Change string param for NGramsCounts to Enumeration. — zongheng.y / githubweb
  16. Use scopt. — zongheng.y / githubweb
  17. Add a comment. — zongheng.y / githubweb
  18. Make NGramsCounts as a FunctionNode (per discussion). — zongheng.y / githubweb

#37 (May 17, 2015 11:08:09 PM)

  1. Removed to method — tomer.kaftan / githubweb
  2. renamed 'then' when function inlining to 'thenFunction' — tomer.kaftan / githubweb

#36 (May 17, 2015 8:18:13 PM)

  1. Fixed bug in Multiclass Evaluator summary — tomer.kaftan / githubweb

#34 (May 17, 2015 7:30:29 PM)

  1. Adding SIFT fix and unit test. — sparks / githubweb
  2. Making VOC pipeline run with the right settings of step and scalestep. — sparks / githubweb

#33 (May 17, 2015 6:33:12 PM)

  1. Removed MatrixUtils read & write CSV — tomer.kaftan / githubweb

#32 (May 17, 2015 3:03:12 PM)

  1. Adding contributors. — sparks / githubweb

#31 (May 17, 2015 2:58:12 PM)

  1. Cleaning up scaladocs and fixing some package issues. — sparks / githubweb
  2. Refactoring and dropping nodes.misc. — sparks / githubweb
  3. Moving TermFrequency to stats. — sparks / githubweb

#30 (May 17, 2015 2:12:47 PM)

  1. First cut of EC2 wrapper script — shivaram / githubweb
  2. Update text — shivaram / githubweb
  3. Address comments — shivaram / githubweb

#29 (May 17, 2015 1:32:34 PM)

  1. Helper scripts to run Pipelines — shivaram / githubweb
  2. Code review fixes — shivaram / githubweb
  3. Address code review comments — shivaram / githubweb
  4. Add make to build instructions — shivaram / githubweb

#28 (May 17, 2015 9:58:13 AM)

  1. Newsgroups data loader no longer cares about train/test. Pipeline now uses scopt. — sparks / githubweb
  2. Response to PR, tweak to .gitignore. — sparks / githubweb

#25 (May 16, 2015 5:38:12 PM)

  1. Adding VOC Data Loader and Supporting MultiLabeledImage classes. — sparks / githubweb
  2. Adding compiling VOC pipeline and updating MAP interace/tests. — sparks / githubweb
  3. Fixing dimension mismatches, removing unnecessary copies, and resolving hadoop API issue. — shivaram / githubweb
  4. Fixing SIFT test. — sparks / githubweb
  5. Cleaning up VLFeat test and pushing fixes to VOC pipeline. — sparks / githubweb
  6. Addressing potential dimension mismatch. — sparks / githubweb
  7. Adding new splitfeatures. — sparks / githubweb
  8. Reformatting and removing unnecessary transform. — sparks / githubweb
  9. Fixing normalization bug, removing unnecessary test label logic. — sparks / githubweb
  10. Creating Im2Single node and adding config params to VOC pipeline. — sparks / githubweb
  11. We should replace the default dylib with one that's binary compatible with the current java code. — sparks / githubweb
  12. Adding javadocs. — sparks / githubweb
  13. Responses to pull requests. Adding unit test for VOC loader. Tightening pipeline. — sparks / githubweb
  14. Adding column sampler node, and making the VOC pipeline more readable. — sparks / githubweb

#24 (May 16, 2015 4:58:11 PM)

  1. Add weighted block coordinate descent — shivaram / githubweb
  2. Add test to check if norm(gradient) goes to zero — shivaram / githubweb
  3. Reduce number of passes in BCD test — shivaram / githubweb
  4. Address code review comments — shivaram / githubweb
  5. Use vertcat, horzcat correctly — shivaram / githubweb
  6. Move reduction functions out from utils — shivaram / githubweb
  7. Convert block mappers into estimators — shivaram / githubweb

#23 (May 15, 2015 6:08:12 PM)

  1. made blocksize only have to be divisible by 512 — tomer.kaftan / githubweb

#22 (May 15, 2015 5:58:12 PM)

  1. Added the randomfftpipeline — tomer.kaftan / githubweb
  2. Fixed multiclass classifier accuracy metrics & moved files around — tomer.kaftan / githubweb
  3. Response to pull request comments — tomer.kaftan / githubweb
  4. Response to pull request comments — tomer.kaftan / githubweb
  5. Made more parameters configurable in MNIST — tomer.kaftan / githubweb

#19 (May 15, 2015 12:53:10 PM)

  1. Fix BlockLinearMapper to add intercept once — shivaram / githubweb

#18 (May 14, 2015 2:33:09 PM)

  1. Added the FastFood node — tomer.kaftan / githubweb
  2. Added docs to the FastFood node — tomer.kaftan / githubweb
  3. renamed FastFood to PaddedFFT — tomer.kaftan / githubweb
  4. response to pull request comments — tomer.kaftan / githubweb
  5. changed comment to reference R not matlab — tomer.kaftan / githubweb
  6. More response to pull request comments — tomer.kaftan / githubweb
  7. misc test cleanup — tomer.kaftan / githubweb

#17 (May 13, 2015 3:28:09 PM)

  1. Exclude hadoop dependencies from Spark — shivaram / githubweb
  2. Add normalization to linear mappers. — shivaram / githubweb
  3. Make intercept an optional field — shivaram / githubweb

#16 (May 12, 2015 11:48:13 AM)

  1. Added Timit Features -> solve pipeline — tomer.kaftan / githubweb
  2. Some cleanup in response to pull request comments — tomer.kaftan / githubweb
  3. Response to pull request comments — tomer.kaftan / githubweb
  4. comment fix — tomer.kaftan / githubweb
  5. Moved tests (again -_-) — tomer.kaftan / githubweb
  6. Added cosine random features suite — tomer.kaftan / githubweb
  7. Added blocklinearmapper unit test — tomer.kaftan / githubweb
  8. Fixed cauchy cosine random features test — tomer.kaftan / githubweb
  9. Response to pull request comments — tomer.kaftan / githubweb

#15 (May 11, 2015 3:33:10 PM)

  1. Added pomOnly tag to netlib — tomer.kaftan / githubweb

#14 (May 10, 2015 2:43:12 PM)

  1. Adds a LinearDiscriminantAnalysis node — tomer.kaftan / githubweb
  2. Improved the LinearDiscriminantAnalysis tests — tomer.kaftan / githubweb
  3. fixed comment — tomer.kaftan / githubweb
  4. Response to pull request comments — tomer.kaftan / githubweb

#13 (May 9, 2015 4:48:11 PM)

  1. Fixing up random sign node, moving signed hellinger mapper to its own node, reorging some files. — sparks / githubweb
  2. Removing unnecessary newline. — sparks / githubweb

#12 (May 9, 2015 3:54:51 PM)

  1. Add nodes for HOG, Daisy, LCS — shivaram / githubweb
  2. Fix image classes after merging with master — shivaram / githubweb
  3. First cut of ImageNetLoader — shivaram / githubweb
  4. Add sample imagenet file, labels, test case — shivaram / githubweb
  5. Address code review comments — shivaram / githubweb
  6. Get rid of Option from labelsPath — shivaram / githubweb

#11 (May 9, 2015 3:13:10 PM)

  1. Remove by 1, update comment — shivaram / githubweb

#10 (May 9, 2015 2:33:10 PM)

  1. Add mean average precision evaluator — shivaram / githubweb
  2. Move evaluation tests into right package — shivaram / githubweb
  3. Update some comments — shivaram / githubweb

#9 (May 9, 2015 11:28:09 AM)

  1. Adding basic CIFAR pipelines, including Caching node, Identity node. — sparks / githubweb
  2. Adding necessary classes for linear pixel pipeline, cifar, and scopt. — sparks / githubweb
  3. Updating nodes to match tomer's changes. — sparks / githubweb
  4. Progress on AMPCamp pipelines. — sparks / githubweb
  5. Adding fixed up Random and RandomPatch pipelines for Cifar. — sparks / githubweb
  6. Cleaning up comments and tests. — sparks / githubweb
  7. Fixing spacing and require. — sparks / githubweb
  8. Adding documentation to normalizeRows. — sparks / githubweb
  9. Adding tests and making CIFAR pipelines use MultiClassEvaluator. — sparks / githubweb
  10. Removing case objects, unnecessary mixins. — sparks / githubweb
  11. Removing unused trait. — sparks / githubweb
  12. Addressing shiv's comments. — sparks / githubweb

#8 (May 8, 2015 5:03:13 PM)

  1. Adding linux binary, fixing Makefile. — sparks / githubweb
  2. Fixing up makefile. — sparks / githubweb

#7 (May 7, 2015 7:18:09 PM)

  1. Added max & top-k classifiers, & a standard scaler (formerly named featurenormalize in old pipelines) — tomer.kaftan / githubweb
  2. Made std dev optional in standard scaler. — tomer.kaftan / githubweb

#3 (May 6, 2015 8:45:12 AM)

  1. First cut of C interfaces. — sparks / githubweb
  2. Cleaning up names and Makefile and adding compiled dylib. — sparks / githubweb
  3. Adding first pass of unit tests. — sparks / githubweb
  4. Adding GMM, tests, and Scaladoc. — sparks / githubweb
  5. Adding Fisher and SIFT Transformers. Addressing PR comments. Changing namespace of the external utils. — sparks / githubweb
  6. Reorganizing imports. — sparks / githubweb
  7. Adding grayscale transformer. — sparks / githubweb
  8. Adding PCA. — sparks / githubweb
  9. Adding matrix based interface for PCA transformer. — sparks / githubweb
  10. Fixing style issues and fixing PCA estimate unit test. — sparks / githubweb
  11. Breaking out ImageUtils and ImageConversions, renaming Image.toVector -> Image.toArray, removing legacy commeted out code. — sparks / githubweb
  12. Cleaning up logging. Closes #75 — sparks / githubweb
  13. Removing misleading comment. — sparks / githubweb
  14. Removing conf/log4j.properties because now the useful thing is in src/main/resources. — sparks / githubweb
  15. First cut of C interfaces. — sparks / githubweb
  16. Cleaning up names and Makefile and adding compiled dylib. — sparks / githubweb
  17. Adding first pass of unit tests. — sparks / githubweb
  18. Adding GMM, tests, and Scaladoc. — sparks / githubweb
  19. Adding Fisher and SIFT Transformers. Addressing PR comments. Changing namespace of the external utils. — sparks / githubweb
  20. Reorganizing imports. — sparks / githubweb
  21. Adding grayscale transformer. — sparks / githubweb
  22. Breaking out ImageUtils and ImageConversions, renaming Image.toVector -> Image.toArray, removing legacy commeted out code. — sparks / githubweb
  23. Reorg of external library namespaces. Added an end-to-end GMM test. — sparks / githubweb
  24. Removing InverseIndexedArrayVectorizedImage. — sparks / githubweb
  25. Removing matrix version of PCA reduction. Saving for future PR. — sparks / githubweb
  26. Removing unnecessary classes. Style fixes. — sparks / githubweb
  27. Merged PipelineNode & Transformer classes, and added a single-item apply to transformers — tomer.kaftan / githubweb
  28. Updated the recent merged pull requests to the altered transformer interface — tomer.kaftan / githubweb
  29. response to pull request comments, following discussion w/ Shivaram and Evan about Broadcasts — tomer.kaftan / githubweb
  30. did logic de-dupe in n-grams featurizer — tomer.kaftan / githubweb