Commits · 7586d9a88f374b107a4540d2bf617727fcb30a3f · tianlh / LightGBM-DCU

07 Apr, 2017 1 commit
- fix bug when indices exceed int32.max for dense matrix. · 7586d9a8
  Guolin Ke authored Apr 07, 2017
  
  7586d9a8
06 Apr, 2017 1 commit

Move all prediction transform to the objective. (#383) · bfb0217a

Guolin Ke authored Apr 06, 2017

* many refactors.

* remove multi_loglossova.

* fix tests.

* avoid using lambda function.

* fix some format.

* reduce branching.

bfb0217a

05 Apr, 2017 2 commits
- improve speed of regression task. (#381) · d4c4d9ae
  Guolin Ke authored Apr 05, 2017
```
* reduce the sumup cost of constant hessians.

* fix test.

* fix bug when have weights.

* fix a comment.

* reduce branching.
```
  d4c4d9ae
- Let TreeLearner share the same code of ConstructHistograms. · 98ffbb2b
  Guolin Ke authored Apr 05, 2017
  
  98ffbb2b
31 Mar, 2017 2 commits
- skip the boosting for the empty class. · 945bc150
  Guolin Ke authored Mar 31, 2017
  
  945bc150
- fix prediction of ova multi-class. · 82c27d42
  Guolin Ke authored Mar 31, 2017
  
  82c27d42
30 Mar, 2017 4 commits
- fix warning. · 512e0c34
  Guolin Ke authored Mar 31, 2017
  
  512e0c34
- skip the training of empty class in classification. · cccd0587
  Guolin Ke authored Mar 31, 2017
  
  cccd0587
- only support boost_from_average in regression. · e404d7cf
  Guolin Ke authored Mar 30, 2017
  
  e404d7cf
- add weight decay in softmax loss. · f40e0d2e
  Guolin Ke authored Mar 30, 2017
  
  f40e0d2e
29 Mar, 2017 2 commits
- fix softmax objective. · 4398906d
  Guolin Ke authored Mar 29, 2017
  
  4398906d
- support OVA multi-classification. · 841a8987
  Guolin Ke authored Mar 29, 2017
  
  841a8987
28 Mar, 2017 2 commits
- support multi-threading exceptions. · 14195876
  Guolin Ke authored Mar 28, 2017
  
  14195876
- eliminate all VS2017 warnings (#367) · 1141ed9d
  zhangyafeikimi authored Mar 27, 2017
  
  1141ed9d
27 Mar, 2017 2 commits
- fix #362 . · d541855a
  Guolin Ke authored Mar 27, 2017
  
  d541855a
- optimize performace (#363) · 04ede377
  zhangyafeikimi authored Mar 27, 2017
  
  04ede377
26 Mar, 2017 2 commits
- fix typos (#361) · 9fe30681
  zhangyafeikimi authored Mar 26, 2017
  
  9fe30681
- avoid NaN output. · 6fac4702
  Guolin Ke authored Mar 26, 2017
  
  6fac4702
25 Mar, 2017 1 commit
- add FitByExistingTree. · 8a6bd5ec
  Guolin Ke authored Mar 25, 2017
  
  8a6bd5ec
24 Mar, 2017 6 commits
- boost_from_average seems doesn't work well in binary classification. remove it. · 349cb50d
  Guolin Ke authored Mar 24, 2017
  
  349cb50d
- don't use "ConvertToRawScore" for self-defined objective function. · 091b4139
  Guolin Ke authored Mar 24, 2017
  
  091b4139
- fix the average label for binary classification. · 76990797
  Guolin Ke authored Mar 24, 2017
  
  76990797
- convert the probabilities to raw score in boost_from_average of classification. · 7b4ead1e
  Guolin Ke authored Mar 24, 2017
  
  7b4ead1e
- robust tree model loading. · 99c6b539
  Guolin Ke authored Mar 24, 2017
  
  99c6b539
- limit the max tree output. change hessian in multi-class objective. · 6757a6aa
  Guolin Ke authored Mar 24, 2017
  
  6757a6aa
23 Mar, 2017 1 commit
- only stop training when all classes are finshed in multi-class. · 3beee91d
  Guolin Ke authored Mar 23, 2017
  
  3beee91d
22 Mar, 2017 4 commits
- fix tests. · 2e962c77
  Guolin Ke authored Mar 23, 2017
  
  2e962c77
- change to boost from average output. · e179c7c6
  Guolin Ke authored Mar 23, 2017
  
  e179c7c6
- fix bug in filter bin · 1c1749db
  Guolin Ke authored Mar 22, 2017
  
  1c1749db
- more efficient algorithm to sample k from n. · 8a0f07ae
  Guolin Ke authored Mar 22, 2017
  
  8a0f07ae
21 Mar, 2017 2 commits
- fix bug in the last commit. · 8dbc48fd
  Guolin Ke authored Mar 21, 2017
  
  8dbc48fd
- refine api of constructing from sampling data. · c060ca75
  Guolin Ke authored Mar 21, 2017
  
  c060ca75
15 Mar, 2017 2 commits
- use std::memcpy to copy array. · cc62b1c3
  Guolin Ke authored Mar 15, 2017
  
  cc62b1c3
- add array allocate api to solve the max_array_size problem in C#. · 516e66a5
  Guolin Ke authored Mar 15, 2017
  
  516e66a5
14 Mar, 2017 2 commits
- sync the global max_bin in parallel learning. · 974230fa
  Guolin Ke authored Mar 14, 2017
  
  974230fa
- add LGBM_DatasetCreateFromSampledMat api. · 9d375069
  Guolin Ke authored Mar 14, 2017
  
  9d375069
13 Mar, 2017 3 commits
- not use exception in command line due to catch-miss of multi-threading exceptions. · 4d033831
  Guolin Ke authored Mar 13, 2017
  
  4d033831
- fix max_bin of categorical feature in parallel learning. · 12ce2566
  Guolin Ke authored Mar 13, 2017
  
  12ce2566
- fix bugs for parallel learning. · ebc0de8b
  Guolin Ke authored Mar 13, 2017
  
  ebc0de8b
07 Mar, 2017 1 commit
- remove the useless training. · 586d53fb
  Guolin Ke authored Mar 07, 2017
  
  586d53fb