Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
6,713 Commits 32 Branches 59 Tags
Commit Graph

1652 Commits

Author SHA1 Message Date
Rong Ou
8dbe0510de
More collective aggregators (#9060) 2023-04-22 03:32:05 +08:00
Jiaming Yuan
7032981350
Fix timer annotation. (#9057) 2023-04-21 22:53:58 +08:00
Jiaming Yuan
b908680bec
Fix race condition in cpp metric tests. (#9058) 2023-04-21 05:24:10 +08:00
Jiaming Yuan
a7b3dd3176
Fix compiler warnings. (#9055) 2023-04-21 02:26:47 +08:00
amdsc21
65d83e288f fix device query 2023-04-19 19:53:26 +02:00
Rong Ou
42d100de18
Make sure metrics work with federated learning (#9037) 2023-04-19 15:39:11 +08:00
Jiaming Yuan
ef13dd31b1
Rework the NDCG objective. (#9015) 2023-04-18 21:16:06 +08:00
Rong Ou
ba9d24ff7b
Make sure metrics work with column-wise distributed training (#9020) 2023-04-18 03:48:23 +08:00
amdsc21
db8420225b fix RCCL 2023-04-12 01:09:14 +02:00
amdsc21
843fdde61b sync Apr 11 2023 2023-04-11 20:03:25 +02:00
amdsc21
08bc4b0c0f Merge branch 'master' into sync-condition-2023Apr11 2023-04-11 19:38:38 +02:00
Rong Ou
15e073ca9d
Make objectives work with vertical distributed and federated learning (#9002) 2023-04-03 17:07:42 +08:00
Jiaming Yuan
d062a9e009
Define pair generation strategies for LTR. (#8984) 2023-03-30 12:00:35 +08:00
amdsc21
acad01afc9 sync Mar 29 2023-03-30 00:46:50 +02:00
Rong Ou
ff26cd3212
More tests for column split and vertical federated learning (#8985) 
Added some more tests for the learner and fit_stump, for both column-wise distributed learning and vertical federated learning.

Also moved the `IsRowSplit` and `IsColumnSplit` methods from the `DMatrix` to the `MetaInfo` since in some places we only have access to the `MetaInfo`. Added a new convenience method `IsVerticalFederatedLearning`.

Some refactoring of the testing fixtures.
 2023-03-28 16:40:26 +08:00
amdsc21
06d9b998ce fix CAPI BuildInfo 2023-03-28 00:14:18 +02:00
amdsc21
c50cc424bc sync Mar 27 2023 2023-03-27 18:54:41 +02:00
Jiaming Yuan
acc110c251
[MT-TREE] Support prediction cache and model slicing. (#8968) 
- Fix prediction range.
- Support prediction cache in mt-hist.
- Support model slicing.
- Make the booster a Python iterable by defining `__iter__`.
- Cleanup removed/deprecated parameters.
- A new field in the output model `iteration_indptr` for pointing to the ranges of trees for each iteration.
 2023-03-27 23:10:54 +08:00
amdsc21
8c77e936d1 tune grid size 2023-03-26 17:45:19 +02:00
amdsc21
7ee4734d3a rm device_helpers.hip.h from cu 2023-03-26 00:24:11 +01:00
amdsc21
ee582f03c3 rm device_helpers.hip.h from cuh 2023-03-25 23:35:57 +01:00
amdsc21
f3286bac04 rm warp header 2023-03-25 23:01:44 +01:00
amdsc21
3ee3bea683 fix warp header 2023-03-25 22:37:37 +01:00
amdsc21
e74b3bbf3c fix macro 2023-03-25 05:17:39 +01:00
amdsc21
22525c002a fix macro 2023-03-25 05:08:30 +01:00
amdsc21
80961039d7 fix macro 2023-03-25 05:00:55 +01:00
amdsc21
1474789787 add new file 2023-03-25 04:54:02 +01:00
amdsc21
1dc138404a initial merge, fix linalg.h 2023-03-25 04:48:47 +01:00
amdsc21
7fbc561e17 initial merge 2023-03-25 04:31:55 +01:00
amdsc21
e0716afabf fix objective/objective.cc, CMakeFile and setup.py 2023-03-23 20:22:34 +01:00
Jiaming Yuan
15a2724ff7
Removed outdated configuration serialization logic. (#8942) 
- `saved_params` is empty.
- `saved_configs_` contains `num_round`, which is not used anywhere inside xgboost.
 2023-03-23 01:31:46 +08:00
Jiaming Yuan
151882dd26
Initial support for multi-target tree. (#8616) 
* Implement multi-target for hist.

- Add new hist tree builder.
- Move data fetchers for tests.
- Dispatch function calls in gbm base on the tree type.
 2023-03-22 23:49:56 +08:00
Jiaming Yuan
ea04d4c46c
[doc] [dask] Troubleshooting NCCL errors. (#8943) 2023-03-22 22:17:26 +08:00
Jiaming Yuan
a551bed803
Remove duplicated learning rate parameter. (#8941) 2023-03-22 20:51:14 +08:00
Jiaming Yuan
a05799ed39
Specify char type in JSON. (#8949) 
char is defined as signed on x86 but unsigned on arm64

- Use `std::int8_t` instead of char.
- Fix include when clang is pretending to be gcc.
 2023-03-22 19:13:44 +08:00
Jiaming Yuan
5891f752c8
Rework the MAP metric. (#8931) 
- The new implementation is more strict as only binary labels are accepted. The previous implementation converts values greater than 1 to 1.
- Deterministic GPU. (no atomic add).
- Fix top-k handling.
- Precise definition of MAP. (There are other variants on how to handle top-k).
- Refactor GPU ranking tests.
 2023-03-22 17:45:20 +08:00
Rong Ou
b240f055d3
Support vertical federated learning (#8932) 2023-03-22 14:25:26 +08:00
amdsc21
595cd81251 add max shared mem workaround 2023-03-19 20:08:42 +01:00
Jiaming Yuan
9b6cc0ed07
Refactor hist to prepare for multi-target builder. (#8928) 
- Extract the builder from the updater class. We need a new builder for multi-target.
- Extract `UpdateTree`, it can be reused for different builders. Eventually, other tree
  updaters can use it as well.
 2023-03-17 17:21:04 +08:00
Jiaming Yuan
a093770f36
Partitioner for multi-target tree. (#8922) 2023-03-16 18:49:34 +08:00
amdsc21
a79a35c22c add warp size 2023-03-15 22:00:26 +01:00
Jiaming Yuan
26209a42a5
Define git attributes for renormalization. (#8921) 2023-03-16 02:43:11 +08:00
Jiaming Yuan
f186c87cf9
Check inf in data for all types of DMatrix. (#8911) 2023-03-15 11:24:35 +08:00
amdsc21
4484c7f073 disable Optin Shared Mem 2023-03-15 02:10:16 +01:00
Jiaming Yuan
72e8331eab
Reimplement the NDCG metric. (#8906) 
- Add support for non-exp gain.
- Cache the DMatrix object to avoid re-calculating the IDCG.
- Make GPU implementation deterministic. (no atomic add)
 2023-03-15 03:26:17 +08:00
Jiaming Yuan
8685556af2
Implement hist evaluator for multi-target tree. (#8908) 2023-03-15 01:42:51 +08:00
Jiaming Yuan
910ce580c8
Clear all cache after model load. (#8904) 2023-03-14 22:09:36 +08:00
Jiaming Yuan
c400fa1e8d
Predictor for vector leaf. (#8898) 2023-03-14 19:07:10 +08:00
amdsc21
364df7db0f fix ../tree/gpu_hist/evaluate_splits.hip bugs, size 64 2023-03-14 06:17:21 +01:00
amdsc21
a2bab03205 fix aft_obj.hip 2023-03-13 23:19:59 +01:00