Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
4,178 Commits 32 Branches 59 Tags
e726dd990281d9b0b1f8403fc837a6e76a0baecc
Commit Graph
100 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Jiaming Yuan e726dd9902 Set device in device dmatrix. (#5596) 2020-04-25 13:42:53 +08:00
Jiaming Yuan f27b6f9ba6 Update document. (#5572) 2020-04-22 02:37:37 +08:00
Jiaming Yuan c355ab65ed Enable parameter validation for R. (#5569) 
* Enable parameter validation for R.

* Add test.
 2020-04-21 11:19:09 -07:00
Jiaming Yuan 564b22cee5 Restore attributes in complete. (#5573) 2020-04-21 11:06:55 -07:00
Jiaming Yuan 9c1103e06c [Breaking] Set output margin to True for custom objective. (#5564) 
* Set output margin to True for custom objective in Python and R.

* Add a demo for writing multi-class custom objective function.

* Run tests on selected demos.
 2020-04-20 20:44:12 +08:00
Jiaming Yuan fcbedcedf8 Fix configuration I load model. (#5562) 2020-04-20 17:25:11 +08:00
Jiaming Yuan 29a4cfe400 Group aware GPU sketching. (#5551) 
* Group aware GPU weighted sketching.

* Distribute group weights to each data point.
* Relax the test.
* Validate input meta info.
* Fix metainfo copy ctor.
 2020-04-20 17:18:52 +08:00
Jiaming Yuan b809f5d8b8 Don't set seed on CLI interface. (#5563) 2020-04-20 12:17:03 +08:00
Jiaming Yuan ccd30e4491 Fix non-openmp build. (#5566) 
* Add test to Jenkins.
* Fix threading utils tests.
* Require thread library.
 2020-04-20 12:16:38 +08:00
Jiaming Yuan e1f22baf8c Fix slice and get info. (#5552) 2020-04-18 18:00:13 +08:00
Jiaming Yuan c245eb8755 Fix r interaction constraints (#5543) 
* Unify the parsing code.

* Cleanup.
 2020-04-18 06:53:51 +08:00
Jiaming Yuan 93df871c8c Assert matching length of evaluation inputs. (#5540) 2020-04-18 06:52:55 +08:00
Jiaming Yuan c69a19e2b1 Fix skl nan tag. (#5538) 2020-04-18 06:52:17 +08:00
Jiaming Yuan cfee9fae91 Don't use uint for threads. (#5542) 2020-04-17 09:45:42 +08:00
Jiaming Yuan bb29ce2818 Add missing aft parameters. [skip ci] (#5553) 2020-04-16 12:08:55 -07:00
Jiaming Yuan 468b1594d3 Fix CLI model IO. (#5535) 
* Add test for comparing Python and CLI training result.
 2020-04-16 07:48:47 +08:00
Jiaming YuanandHyunsu Cho 8b04736b81 [dask] dask cudf inplace prediction. (#5512) 
* Add inplace prediction for dask-cudf.

* Remove Dockerfile.release, since it's not used anywhere

* Use Conda exclusively in CUDF and GPU containers

* Improve cupy memory copying.

* Add skip marks to tests.

* Add mgpu-cudf category on the CI to run all distributed tests.

Co-authored-by: Hyunsu Cho <chohyu01@cs.washington.edu>
 2020-04-15 18:15:51 +08:00
Jiaming Yuan a2f54963b6 Write binary header. (#5532) 2020-04-15 17:47:57 +08:00
Jiaming Yuan c90119eb67 Update Python doc. [skip ci] (#5517) 
* Update doc for copying booster. [skip ci]

The issue is resolved in  #5312 .

* Add version for new APIs. [skip ci]
 2020-04-14 16:25:20 +08:00
Jiaming Yuan b56c902841 [R] R raw serialization. (#5123) 
* Add bindings for serialization.
* Change `xgb.save.raw' into full serialization instead of simple model.
* Add `xgb.load.raw' for unserialization.
* Run devtools.
 2020-04-11 17:16:54 +08:00
Jiaming Yuan a3db79df22 Remove makefiles. (#5513) 2020-04-11 13:25:53 +08:00
Jiaming Yuan 4a0c8ef237 Update doc for parameter validation. (#5508) 
* Update doc for parameter validation.

* Fix github rebase.
 2020-04-11 00:43:46 +08:00
Jiaming Yuan 1334aca437 Fix github merge. (#5509) 2020-04-10 22:17:38 +08:00
Jiaming Yuan 866a477319 Unify max nodes. (#5497) 2020-04-10 19:26:35 +08:00
Jiaming Yuan bd653fad4c Remove distcol updater. (#5507) 
Closes #5498.
 2020-04-10 12:52:56 +08:00
Jiaming Yuan 7d52c0b8c2 Requires setting leaf stat when expanding tree. (#5501) 
* Fix GPU Hist feature importance.
 2020-04-10 12:27:03 +08:00
Jiaming Yuan dc2950fd90 Fix checking booster. (#5505) 
* Use `get_params()` instead of `getattr` intrinsic.
 2020-04-10 12:21:21 +08:00
Jiaming Yuan 6671b42dd4 Use ellpack for prediction only when sparsepage doesn't exist. (#5504) 2020-04-10 12:15:46 +08:00
Jiaming YuanandHyunsu Cho 0012f2ef93 Upgrade clang-tidy on CI. (#5469) 
* Correct all clang-tidy errors.
* Upgrade clang-tidy to 10 on CI.

Co-authored-by: Hyunsu Cho <chohyu01@cs.washington.edu>
 2020-04-05 04:42:29 +08:00
Jiaming Yuan a9313802ea Fix dump model. (#5485) 2020-04-05 03:52:54 +08:00
Jiaming Yuan 939973630d Accept other gradient types for split entry. (#5467) 2020-04-03 10:38:44 +08:00
Jiaming Yuan 86beb68ce8 Implement host span. (#5459) 2020-04-03 10:37:51 +08:00
Jiaming Yuan 459b175dc6 Split up test helpers header. (#5455) 2020-04-03 10:36:53 +08:00
Jiaming Yuan c218d8ffbf Enable parameter validation for skl. (#5477) 2020-04-03 10:23:58 +08:00
Jiaming Yuan d0b86c75d9 Remove silent parameter. (#5476) 2020-04-03 08:03:26 +08:00
Jiaming Yuan 29c6ad943a Prevent copying SimpleDMatrix. (#5453) 
* Set default dtor for SimpleDMatrix to initialize default copy ctor, which is
deleted due to unique ptr.

* Remove commented code.
* Remove warning for calling host function (std::max).
* Remove warning for initialization order.
* Remove warning for unused variables.
 2020-04-02 07:01:49 +08:00
Jiaming Yuan e86030c360 Update dmlc-core. (#5466) 
* Copy dmlc travis script to XGBoost.
 2020-04-02 04:16:39 +08:00
Jiaming Yuan babcb996e7 Reduce span check overhead. (#5464) 2020-04-01 22:07:24 +08:00
Jiaming Yuan 6601a641d7 Thread safe, inplace prediction. (#5389) 
Normal prediction with DMatrix is now thread safe with locks.  Added inplace prediction is lock free thread safe.

When data is on device (cupy, cudf), the returned data is also on device.

* Implementation for numpy, csr, cudf and cupy.

* Implementation for dask.

* Remove sync in simple dmatrix.
 2020-03-30 15:35:28 +08:00
Jiaming Yuan 780de49ddb Resolve travis failure. (#5445) 
* Install dependencies by pip.
 2020-03-27 19:37:58 +08:00
Jiaming Yuan 4942da64ae Refactor tests with data generator. (#5439) 2020-03-27 06:44:44 +08:00
Jiaming Yuan 7146b91d5a Force compressed buffer to be 4 bytes aligned. (#5441) 2020-03-27 06:43:52 +08:00
Jiaming Yuan cd7d6f7d59 [dask] Fix missing value for scikit-learn interface. (#5435) 2020-03-20 10:56:01 -04:00
Jiaming Yuan abca9908ba Support pandas SparseArray. (#5431) 2020-03-20 21:40:22 +08:00
Jiaming Yuan 760d5d0c3c [dask] Accept other inputs for prediction. (#5428) 
* Returns a series when input is dataframe.

* Merge assert client.
 2020-03-19 17:05:55 +08:00
Jiaming Yuan 8ca06ab329 [dask] Check non-equal when setting threads. (#5421) 
* Check non-equal.

`nthread` can be restored from internal parameter, which is mis-interpreted as
user defined parameter.

* Check None.
 2020-03-17 13:07:20 +08:00
Jiaming Yuan b51124c158 [dask] Enable gridsearching with skl. (#5417) 2020-03-16 04:51:51 +08:00
Jiaming Yuan 761a5dbdfc [dask] Honor nthreads from dask worker. (#5414) 2020-03-16 04:51:24 +08:00
Jiaming Yuan 21b671aa06 [dask] Order the prediction result. (#5416) 2020-03-15 19:34:04 +08:00
Jiaming Yuan 668e432e2d [dask] Use DMLC_TASK_ID. (#5415) 2020-03-15 16:47:03 +08:00
Jiaming Yuan fc88105620 Better error message for updating. (#5418) 2020-03-15 16:46:21 +08:00
Jiaming Yuan ab7a46a1a4 Check whether current updater can modify a tree. (#5406) 
* Check whether current updater can modify a tree.

* Fix tree model JSON IO for pruned trees.
 2020-03-14 09:24:08 +08:00
Jiaming Yuan 45a97ddf32 Split up LearnerImpl. (#5350) 2020-03-12 16:30:23 +08:00
Jiaming Yuan 0dd97c206b Move thread local entry into Learner. (#5396) 
* Move thread local entry into Learner.

This is an attempt to workaround CUDA context issue in static variable, where
the CUDA context can be released before device vector.

* Add PredictionEntry to thread local entry.

This eliminates one copy of prediction vector.

* Don't define CUDA C API in a namespace.
 2020-03-07 15:37:39 +08:00
Jiaming Yuan 8d06878bf9 Deterministic GPU histogram. (#5361) 
* Use pre-rounding based method to obtain reproducible floating point
  summation.
* GPU Hist for regression and classification are bit-by-bit reproducible.
* Add doc.
* Switch to thrust reduce for `node_sum_gradient`.
 2020-03-04 15:13:28 +08:00
Jiaming Yuan 2ba8c13b69 Revert "Enable rabit test (#5358)" (#5377) 
This reverts commit 9a5efffebe.
 2020-02-29 04:25:03 +08:00
Jiaming Yuan a461a9a90a Define lazy isinstance for Python compat. (#5364) 
* Avoid importing datatable.
* Fix #5363.
 2020-02-26 14:23:33 +08:00
Jiaming Yuan 0fd455e162 Restore loading model from buffer. (#5360) 2020-02-26 11:30:13 +08:00
Jiaming Yuan f2b8cd2922 Add number of columns to native data iterator. (#5202) 
* Change native data iter into an adapter.
 2020-02-25 23:42:01 +08:00
Jiaming Yuan e0509b3307 Fix pruner. (#5335) 
* Honor the tree depth.
* Prevent pruning pruned node.
 2020-02-25 08:32:46 +08:00
Jiaming Yuan 655cf17b60 Predict on Ellpack. (#5327) 
* Unify GPU prediction node.
* Add `PageExists`.
* Dispatch prediction on input data for GPU Predictor.
 2020-02-23 06:27:03 +08:00
Jiaming Yuan 9f77c18b0d Add JVM_CHECK_CALL. (#5199) 
* Added a check call macro in jvm package, prevents executing other functions
from jvm when error occurred in XGBoost. For example, when prediction fails jvm
should not try to allocate memory based on the output prediction size.
 2020-02-18 11:10:55 +08:00
Jiaming Yuan 0110754a76 Remove update prediction cache from predictors. (#5312) 
Move this function into gbtree, and uses only updater for doing so. As now the predictor knows exactly how many trees to predict, there's no need for it to update the prediction cache.
 2020-02-17 11:35:47 +08:00
Jiaming Yuan e433a379e4 Fix changing locale. (#5314) 
* Fix changing locale.

* Don't use locale guard.

As number parsing is implemented in house, we don't need locale.

* Update doc.
 2020-02-17 11:31:13 +08:00
Jiaming Yuan ed2465cce4 Add configuration to R interface. (#5217) 
* Save and load internal parameter configuration as JSON.
 2020-02-16 03:01:58 +08:00
Jiaming Yuan 8ca9744b07 Use scikit-learn in extra dependencies. (#5310) 2020-02-15 07:12:51 +08:00
Jiaming Yuan c35cdecddd Move prediction cache to Learner. (#5220) 
* Move prediction cache into Learner.

* Clean-ups

- Remove duplicated cache in Learner and GBM.
- Remove ad-hoc fix of invalid cache.
- Remove `PredictFromCache` in predictors.
- Remove prediction cache for linear altogether, as it's only moving the
  prediction into training process but doesn't provide any actual overall speed
  gain.
- The cache is now unique to Learner, which means the ownership is no longer
  shared by any other components.

* Changes

- Add version to prediction cache.
- Use weak ptr to check expired DMatrix.
- Pass shared pointer instead of raw pointer.
 2020-02-14 13:04:23 +08:00
Jiaming Yuan 911a902835 Merge model compatibility fixes from 1.0rc branch. (#5305) 
* Port test model compatibility.
* Port logit model fix.

https://github.com/dmlc/xgboost/pull/5248
https://github.com/dmlc/xgboost/pull/5281
 2020-02-13 20:41:58 +08:00
Jiaming Yuan 29eeea709a Pass shared pointer instead of raw pointer to Learner. (#5302) 
Extracted from https://github.com/dmlc/xgboost/pull/5220 .
 2020-02-11 14:16:38 +08:00
Jiaming Yuan 84e395d91e Fix CMake build on Windows with setuptools. (#5280) 2020-02-05 10:47:39 +08:00
Jiaming Yuan 595a00466d Rewrite setup.py. (#5271) 
The setup.py is rewritten.  This new script uses only Python code and provide customized
implementation of setuptools commands.  This way users can run most of setuptools commands
just like any other Python libraries.

* Remove setup_pip.py
* Remove soft links.
* Define customized commands.
* Remove shell script.
* Remove makefile script.
* Update the doc for building from source.
 2020-02-04 13:35:42 +08:00
Jiaming Yuan a5cc112eea Export JSON config in get_params. (#5256) 2020-02-03 12:46:51 +08:00
Jiaming Yuan ed0216642f Avoid dask test fixtures. (#5270) 
* Fix Travis OSX timeout.

* Fix classifier.
 2020-02-03 12:39:20 +08:00
Jiaming Yuan 856b81c727 Ignore gdb_history. [skip ci] (#5257) 2020-02-02 20:40:09 +08:00
Jiaming Yuan fe8d72b50b Cleanup warnings. (#5247) 
From clang-tidy-9 and gcc-7: Invalid case style, narrowing definition, wrong
initialization order, unused variables.
 2020-01-31 14:52:15 +08:00
Jiaming Yuan 472ded549d Save Scikit-Learn attributes into learner attributes. (#5245) 
* Remove the recommendation for pickle.

* Save skl attributes in booster.attr

* Test loading scikit-learn model with native booster.
 2020-01-30 16:00:18 +08:00
Jiaming Yuan 43974939f4 Better error message about loading pickled model. (#5236) 
Print the link to new tutorial.
 2020-01-28 15:29:14 +08:00
Jiaming Yuan ef19480eda Add dart to JSON schema. (#5218) 
* Add dart to JSON schema.

* Use spaces instead of tab.
 2020-01-28 13:29:09 +08:00
Jiaming Yuan 3eb1279bbf Config for linear updaters. (#5222) 2020-01-25 11:26:46 +08:00
Jiaming Yuan 40680368cf Add constraint parameters to Scikit-Learn interface. (#5227) 
* Add document for constraints.

* Fix a format error in doc for objective function.
 2020-01-25 11:12:02 +08:00
Jiaming Yuan 1891cc766d Fix metainfo from DataFrame. (#5216) 
* Fix metainfo from DataFrame.

* Unify helper functions for data and meta.
 2020-01-22 16:29:44 +08:00
Jiaming Yuan 5199b86126 Fix R dart prediction. (#5204) 
* Fix R dart prediction and add test.
 2020-01-16 12:11:04 +08:00
Jiaming Yuan 808f61081b Update R doc by roxygen2. (#5201) 2020-01-15 10:49:17 +08:00
Jiaming Yuan 7b65698187 Enforce correct data shape. (#5191) 
* Fix syncing DMatrix columns.
* notes for tree method.
* Enable feature validation for all interfaces except for jvm.
* Better tests for boosting from predictions.
* Disable validation on JVM.
 2020-01-13 15:48:17 +08:00
Jiaming Yuan ee287808fb Lazy initialization of device vector. (#5173) 
* Lazy initialization of device vector.

* Fix #5162.

* Disable copy constructor of HostDeviceVector.  Prevents implicit copying.

* Fix CPU build.

* Bring back move assignment operator.
 2020-01-07 11:23:05 +08:00
Jiaming Yuan 77cfbff5a7 Fix span constructor. (#5166) 
* Make sure copy constructor is used.
 2020-01-07 11:19:38 +08:00
Jiaming Yuan ebc86a3afa Disable parameter validation for Scikit-Learn interface. (#5167) 
* Disable parameter validation for now.

Scikit-Learn passes all parameters down to XGBoost, whether they are used or
not.

* Add option `validate_parameters`.
 2020-01-07 11:17:31 +08:00
Jiaming Yuan 2b9a62a806 Throw error when not compiled with NCCL. (#5170) 2020-01-07 11:15:19 +08:00
Jiaming Yuan 04db125699 Quick fix for memory leak in CPU Hist. (#5153) 
Closes https://github.com/dmlc/xgboost/issues/3579 .

* Don't use map.
 2019-12-31 14:05:53 +08:00
Jiaming Yuan 139ccc9902 Fix num_roots to be 1. (#5165) 2019-12-30 02:18:45 +08:00
Jiaming Yuan d55489af14 Don't use modernize-use-trailing-return-type. (#5169) 2019-12-29 20:18:23 +08:00
Jiaming Yuan 6848d0426f Clean up Python 2 compatibility code. (#5161) 2019-12-27 18:34:53 +08:00
Jiaming Yuan 61286c6e8f Fix wrapping GPU ID and prevent data copying. (#5160) 
* Removed some data copying.

* Make sure gpu_id is valid before any configuration is carried out.
 2019-12-27 16:51:08 +08:00
Jiaming Yuan f3d7877802 Parameter validation (#5157) 
* Unused code.

* Split up old colmaker parameters from train param.

* Fix dart.

* Better name.
 2019-12-26 11:59:05 +08:00
Jiaming Yuan ced3660f60 Tests for empty dmatrix. (#5159) 2019-12-26 11:51:54 +08:00
Jiaming Yuan 298ebe68ac [Breaking] Remove learning_rates in Python. (#5155) 
* Remove `learning_rates`.

It's been deprecated since we have callback.

* Set `before_iteration` of `reset_learning_rate` to False to preserve 
  the initial learning rate, and comply to the term "reset".

Closes #4709.

* Tests for various `tree_method`.
 2019-12-24 14:25:48 +08:00
Jiaming Yuan 73b1bd2789 Update demo for ranking. (#5154) 2019-12-24 13:39:07 +08:00
Jiaming Yuan 0202e04a8e Add base margin to sklearn interface. (#5151) 2019-12-24 09:43:41 +08:00
Jiaming Yuan 1d0ca49761 Example JSON model parser and Schema. (#5137) 2019-12-23 19:47:35 +08:00
Jiaming Yuan a4b929385e Note for DaskDMatrix. (#5144) 
* Brief introduction to `DaskDMatrix`.

* Add xgboost.dask.train to API doc
 2019-12-23 18:55:32 +08:00