Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
7,059 Commits 32 Branches 59 Tags
Commit Graph

635 Commits

Author SHA1 Message Date
Jiaming Yuan
d7d1b6e3a6
CPU evaluation for cat data. (#7393) 
* Implementation for one hot based.
* Implementation for partition based. (LightGBM)
 2021-11-06 14:41:35 +08:00
Jiaming Yuan
ccdabe4512
Support building gradient index with cat data. (#7371) 2021-11-03 22:37:37 +08:00
Jiaming Yuan
57a4b4ff64
Handle OMP_THREAD_LIMIT. (#7390) 2021-11-03 15:44:38 +08:00
Jiaming Yuan
a55d43ccfd
Add test for invalid categorical data values. (#7380) 
* Add test for invalid categorical data values.

* Add check during sketching.
 2021-11-02 18:00:52 +08:00
Jiaming Yuan
32e673d8c4
Support building with CTK11.5. (#7379) 
* Support building with CTK11.5.

* Require system cub installation for CTK11.4+.
* Check thrust version for segmented sort.
 2021-11-02 16:22:26 +08:00
Jiaming Yuan
6295dc3b67
Fix span reverse iterator. (#7387) 
* Fix span reverse iterator.

* Disable `rbegin` on device code to avoid calling host function.
* Add `trbegin` and friends.
 2021-11-02 13:35:59 +08:00
Jiaming Yuan
ac9bfaa4f2
Handle missing values in dataframe with category dtype. (#7331) 
* Replace -1 in pandas initializer.
* Unify `IsValid` functor.
* Mimic pandas data handling in cuDF glue code.
* Check invalid categories.
* Fix DDM sketching.
 2021-10-28 03:33:54 +08:00
Jiaming Yuan
d4349426d8
Re-implement PR-AUC. (#7297) 
* Support binary/multi-class classification, ranking.
* Add documents.
* Handle missing data.
 2021-10-26 13:07:50 +08:00
Jiaming Yuan
ca17f8a5fc
Dispatch thrust versions and upgrade rmm. (#7254) 
Co-authored-by: Philip Hyunsu Cho <chohyu01@cs.washington.edu>
 2021-09-25 03:43:23 +08:00
Jiaming Yuan
c311a8c1d8
Enable compiling with system cub. (#7232) 
- Tested with all CUDA 11.x.
- Workaround cub scan by using discard iterator in AUC.
- Limit the size of Argsort when compiled with CUDA cub.
 2021-09-17 14:28:18 +08:00
Jiaming Yuan
31c1e13f90
Categorical data support in CPU sketching. (#7221) 2021-09-17 04:37:09 +08:00
Jiaming Yuan
2942dc68e4
Fix mixed types in GPU sketching. (#7228) 2021-09-16 00:10:25 +08:00
Jiaming Yuan
3515931305
Initial support for external memory in gradient index. (#7183) 
* Add hessian to batch param in preparation of new approx impl.
* Extract a push method for gradient index matrix.
* Use span instead of vector ref for hessian in sketching.
* Create a binary format for gradient index.
 2021-09-13 12:40:56 +08:00
Jiaming Yuan
b12e7f7edd
Add noexcept to JSON objects. (#7205) 2021-09-07 13:56:48 +08:00
Jiaming Yuan
ba69244a94
Restore the custom double atomic add. (#7198) 2021-08-28 18:30:42 +08:00
Jiaming Yuan
7a1d67f9cb
[breaking] Use integer atomic for GPU histogram. (#7180) 
On GPU we use rouding factor to truncate the gradient for deterministic results. This PR changes the gradient representation to fixed point number with exponent aligned with rounding factor.

    [breaking] Drop non-deterministic histogram.
    Use fixed point for shared memory.

This PR is to improve the performance of GPU Hist. 

Co-authored-by: Andy Adinets <aadinets@nvidia.com>
 2021-08-28 05:17:05 +08:00
Jiaming Yuan
e7d7ab6bc3
Better error message for ncclUnhandledCudaError. (#7190) 2021-08-27 10:29:22 +08:00
Jiaming Yuan
9600ca83f3
Remove synchronization in monitor. (#7164) 
* Remove synchronization in monitor.

Calling rabit functions during destruction is flaky.

* Add xgboost prefix to nvtx marker.
 2021-08-11 16:33:53 +08:00
Jiaming Yuan
149f209af6
Extract histogram builder from CPU Hist. (#7152) 
* Extract the CPU histogram builder.
* Fix tests.
* Reduce number of histograms being built.
 2021-08-09 21:15:21 +08:00
Robert Maynard
1a75f43304
Allow compilation with nvcc 11.4 (#7131) 
* Use type aliases for discard iterators

* update to include host_vector as thrust 1.12 doesn't bring it in as a side-effect

* cub::DispatchRadixSort requires signed offset types
 2021-07-27 20:05:33 +08:00
Taewoo Kim
41e882f80b
Check input value is duplicated when quantile queue is full (#7091) 
Co-authored-by: Taewoo Kim <taewoo@layer6.com>
 2021-07-23 03:07:01 +08:00
farfarawayzyt
e64ee6592f
fix typo in src/common/hist.cc BuildHistKernel (#7116) 2021-07-21 19:53:05 +08:00
farfarawayzyt
d7c14496d2
fix typo in arguments of PartitionBuilder::Init (#7113) 
Co-authored-by: Yuntian Zhang <zhangyt@lamda.nju.edu.cn>
 2021-07-16 15:46:22 +08:00
Jiaming Yuan
bd1f3a38f0
Rewrite sparse dmatrix using callbacks. (#7092) 
- Reduce dependency on dmlc parsers and provide an interface for users to load data by themselves.
- Remove use of threaded iterator and IO queue.
- Remove `page_size`.
- Make sure the number of pages in memory is bounded.
- Make sure the cache can not be violated.
- Provide an interface for internal algorithms to process data asynchronously.
 2021-07-16 12:33:31 +08:00
Jiaming Yuan
77f6cf2d13
Support hessian in host sketch container. (#7081) 
Prepare for migrating approx onto hist's codebase.
 2021-07-08 16:33:58 +08:00
Jiaming Yuan
1cd20efe68
Move GHistIndex into DMatrix. (#7064) 2021-07-01 00:44:49 +08:00
Jiaming Yuan
1c8fdf2218
Remove use of device_idx in dh::LaunchN. (#7063) 
It's an unused parameter, removing it can make the CI log more readable.
 2021-06-29 11:37:26 +08:00
Jiaming Yuan
8fa32fdda2
Implement categorical data support for SHAP. (#7053) 
* Add CPU implementation.
* Update GPUTreeSHAP.
* Add GPU implementation by defining custom split condition.
 2021-06-25 19:02:46 +08:00
Jiaming Yuan
86715e4cd4
Support categorical data for dask functional interface and DQM. (#7043) 
* Support categorical data for dask functional interface and DQM.

* Implement categorical data support for GPU GK-merge.
* Add support for dask functional interface.
* Add support for DQM.

* Get newer cupy.
 2021-06-18 13:06:52 +08:00
ShvetsKS
2567404ab6
Simplify sparse and dense CPU hist kernels (#7029) 
* Simplify sparse and dense kernels
* Extract row partitioner.

Co-authored-by: Kirill Shvets <kirill.shvets@intel.com>
 2021-06-11 18:26:30 +08:00
TP Boudreau
bd2ca543c4
Fix BinarySearchBin() argument types (#7026) 2021-06-08 19:05:46 +08:00
ShvetsKS
5cdaac00c1
Remove feature grouping (#7018) 
Co-authored-by: Kirill Shvets <kirill.shvets@intel.com>
 2021-06-03 04:35:26 +08:00
Andrew Ziem
3e7e426b36
Fix spelling in documents (#6948) 
* Update roxygen2 doc.

Co-authored-by: fis <jm.yuan@outlook.com>
 2021-05-11 20:44:36 +08:00
Jiaming Yuan
a2ecbdaa31
Add an API guard to prevent global variables being changed. (#6891) 2021-04-23 10:27:57 +08:00
Jiaming Yuan
1b26a2a561
Copy output data for argsort. (#6866) 
Fix GPU AUC.
 2021-04-16 21:05:01 +08:00
Jiaming Yuan
f294c4e023
Use constexpr in dh::CopyIf. (#6828) 2021-04-08 07:37:47 +08:00
Jiaming Yuan
7bcc8b3e5c
Use batched copy if. (#6826) 2021-04-06 10:34:04 +08:00
Jiaming Yuan
3039dd194b
Don't estimate sketch batch size when rmm is used. (#6807) 2021-03-31 15:29:56 +08:00
ShvetsKS
8825670c9c
Memory consumption fix for row-major adapters (#6779) 
Co-authored-by: Kirill Shvets <kirill.shvets@intel.com>
Co-authored-by: fis <jm.yuan@outlook.com>
 2021-03-26 08:44:30 +08:00
Jiaming Yuan
a7083d3c13
Fix dart inplace prediction with GPU input. (#6777) 
* Fix dart inplace predict with data on GPU, which might trigger a fatal check
for device access right.
* Avoid copying data whenever possible.
 2021-03-25 12:00:32 +08:00
Jiaming Yuan
bcc0277338
Re-implement ROC-AUC. (#6747) 
* Re-implement ROC-AUC.

* Binary
* MultiClass
* LTR
* Add documents.

This PR resolves a few issues:
  - Define a value when the dataset is invalid, which can happen if there's an
  empty dataset, or when the dataset contains only positive or negative values.
  - Define ROC-AUC for multi-class classification.
  - Define weighted average value for distributed setting.
  - A correct implementation for learning to rank task.  Previous
  implementation is just binary classification with averaging across groups,
  which doesn't measure ordered learning to rank.
 2021-03-20 16:52:40 +08:00
Philip Hyunsu Cho
4230dcb614
Re-introduce double buffer in UpdatePosition, to fix perf regression in gpu_hist (#6757) 
* Revert "gpu_hist performance tweaks (#5707)"

This reverts commit f779980f7ea7f6f07e86229b8e78144e8a74e6b3.

* Address reviewer's comment

* Fix build error
 2021-03-18 13:56:10 -07:00
Jiaming Yuan
1a73a28511
Add device argsort. (#6749) 
This is part of https://github.com/dmlc/xgboost/pull/6747 .
 2021-03-16 16:05:22 +08:00
Philip Hyunsu Cho
366f3cb9d8
Add use_rmm flag to global configuration (#6656) 
* Ensure RMM is 0.18 or later

* Add use_rmm flag to global configuration

* Modify XGBCachingDeviceAllocatorImpl to skip CUB when use_rmm=True

* Update the demo

* [CI] Pin NumPy to 1.19.4, since NumPy 1.19.5 doesn't work with latest Shap
 2021-03-09 14:53:05 -08:00
Louis Desreumaux
9b530e5697
Improve OpenMP exception handling (#6680) 2021-02-25 13:56:16 +08:00
Jiaming Yuan
1b70a323a7
Improve string view to reduce string allocation. (#6644) 2021-01-27 19:08:52 +08:00
Jiaming Yuan
f2f7dd87b8
Use view for SparsePage exclusively. (#6590) 2021-01-11 18:04:55 +08:00
Philip Hyunsu Cho
bf6cfe3b99
[Breaking] Upgrade cuDF and RMM to 0.18 nightlies; require RMM 0.18+ for RMM plugin (#6510) 
* [CI] Upgrade cuDF and RMM to 0.18 nightlies

* Modify RMM plugin to be compatible with RMM 0.18

* Update src/common/device_helpers.cuh

Co-authored-by: Mark Harris <mharris@nvidia.com>

Co-authored-by: Mark Harris <mharris@nvidia.com>
 2020-12-16 10:07:52 -08:00
Jiaming Yuan
886486a519
Support categorical data in GPU weighted sketching. (#6508) 2020-12-16 14:23:28 +08:00
ShvetsKS
956beead70
Thread local memory allocation for BuildHist (#6358) 
* thread mem locality

* fix apply

* cleanup

* fix lint

* fix tests

* simple try

* fix

* fix

* apply comments

* fix comments

* fix

* apply simple comment

Co-authored-by: ShvetsKS <kirill.shvets@intel.com>
 2020-11-25 17:50:12 +03:00