Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Add GPU support to NVFlare demo (#9552)

Browse Source
This commit is contained in:
Rong Ou 2023-09-06 02:03:59 -07:00 committed by GitHub
parent 3b9e5909fb
commit 0f35493b65
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 14 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
demo/nvflare/horizontal/README.md
View File

@ -85,8 +85,8 @@ shutdown server
## Training with GPUs ## Training with GPUs
To demo with Federated Learning using GPUs, make sure your machine has at least 2 GPUs. To demo with Federated Learning using GPUs, make sure your machine has at least 2 GPUs.
Build XGBoost with the federated learning plugin enabled along with CUDA, but with NCCL Build XGBoost with the federated learning plugin enabled along with CUDA
turned off (see the [README](../../plugin/federated/README.md)). (see the [README](../../plugin/federated/README.md)).
Modify `config/config_fed_client.json` and set `use_gpus` to `true`, then repeat the steps Modify `../config/config_fed_client.json` and set `use_gpus` to `true`, then repeat the steps
above. above.

2
demo/nvflare/horizontal/custom/trainer.py
View File

@ -67,7 +67,7 @@ class XGBoostTrainer(Executor):
dtest = xgb.DMatrix('agaricus.txt.test?format=libsvm') dtest = xgb.DMatrix('agaricus.txt.test?format=libsvm')
# Specify parameters via map, definition are same as c++ version # Specify parameters via map, definition are same as c++ version
param = {'max_depth': 2, 'eta': 1, 'objective': 'binary:logistic'} param = {'tree_method': 'hist', 'max_depth': 2, 'eta': 1, 'objective': 'binary:logistic'}
if self._use_gpus: if self._use_gpus:
self.log_info(fl_ctx, f'Training with GPU {rank}') self.log_info(fl_ctx, f'Training with GPU {rank}')
param['device'] = f"cuda:{rank}" param['device'] = f"cuda:{rank}"

7
demo/nvflare/vertical/README.md
View File

@ -56,4 +56,9 @@ shutdown server
## Training with GPUs ## Training with GPUs
Currently GPUs are not yet supported by vertical federated XGBoost. To demo with Vertical Federated Learning using GPUs, make sure your machine has at least 2 GPUs.
Build XGBoost with the federated learning plugin enabled along with CUDA
(see the [README](../../plugin/federated/README.md)).
Modify `../config/config_fed_client.json` and set `use_gpus` to `true`, then repeat the steps
above.

6
demo/nvflare/vertical/custom/trainer.py
View File

@ -77,13 +77,15 @@ class XGBoostTrainer(Executor):
'gamma': 1.0, 'gamma': 1.0,
'max_depth': 8, 'max_depth': 8,
'min_child_weight': 100, 'min_child_weight': 100,
'tree_method': 'approx', 'tree_method': 'hist',
'grow_policy': 'depthwise', 'grow_policy': 'depthwise',
'objective': 'binary:logistic', 'objective': 'binary:logistic',
'eval_metric': 'auc', 'eval_metric': 'auc',
} }
if self._use_gpus: if self._use_gpus:
self.log_info(fl_ctx, 'GPUs are not currently supported by vertical federated XGBoost') if self._use_gpus:
self.log_info(fl_ctx, f'Training with GPU {rank}')
param['device'] = f"cuda:{rank}"
# specify validations set to watch performance # specify validations set to watch performance
watchlist = [(dtest, "eval"), (dtrain, "train")] watchlist = [(dtest, "eval"), (dtrain, "train")]