1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
|
"""Test model IO with pickle."""
import os
import pickle
import subprocess
import numpy as np
import pytest
import xgboost as xgb
from xgboost import XGBClassifier
from xgboost import testing as tm
model_path = "./model.pkl"
pytestmark = tm.timeout(30)
def build_dataset():
N = 10
x = np.linspace(0, N * N, N * N)
x = x.reshape((N, N))
y = np.linspace(0, N, N)
return x, y
def save_pickle(bst, path):
with open(path, "wb") as fd:
pickle.dump(bst, fd)
def load_pickle(path):
with open(path, "rb") as fd:
bst = pickle.load(fd)
return bst
class TestPickling:
args_template = ["pytest", "--verbose", "-s", "--fulltrace"]
def run_pickling(self, bst) -> None:
save_pickle(bst, model_path)
args = [
"pytest",
"--verbose",
"-s",
"--fulltrace",
"./tests/python-gpu/load_pickle.py::TestLoadPickle::test_load_pkl",
]
command = ""
for arg in args:
command += arg
command += " "
cuda_environment = {"CUDA_VISIBLE_DEVICES": "-1"}
env = os.environ.copy()
# Passing new_environment directly to `env' argument results
# in failure on Windows:
# Fatal Python error: _Py_HashRandomization_Init: failed to
# get random numbers to initialize Python
env.update(cuda_environment)
# Load model in a CPU only environment.
status = subprocess.call(command, env=env, shell=True)
assert status == 0
os.remove(model_path)
# TODO: This test is too slow
@pytest.mark.skipif(**tm.no_sklearn())
def test_pickling(self):
x, y = build_dataset()
train_x = xgb.DMatrix(x, label=y)
param = {"tree_method": "gpu_hist", "gpu_id": 0}
bst = xgb.train(param, train_x)
self.run_pickling(bst)
bst = xgb.XGBRegressor(**param).fit(x, y)
self.run_pickling(bst)
param = {"booster": "gblinear", "updater": "gpu_coord_descent", "gpu_id": 0}
bst = xgb.train(param, train_x)
self.run_pickling(bst)
bst = xgb.XGBRegressor(**param).fit(x, y)
self.run_pickling(bst)
@pytest.mark.mgpu
def test_wrap_gpu_id(self):
X, y = build_dataset()
dtrain = xgb.DMatrix(X, y)
bst = xgb.train(
{"tree_method": "gpu_hist", "gpu_id": 1}, dtrain, num_boost_round=6
)
model_path = "model.pkl"
save_pickle(bst, model_path)
cuda_environment = {"CUDA_VISIBLE_DEVICES": "0"}
env = os.environ.copy()
env.update(cuda_environment)
args = self.args_template.copy()
args.append(
"./tests/python-gpu/" "load_pickle.py::TestLoadPickle::test_wrap_gpu_id"
)
status = subprocess.call(args, env=env)
assert status == 0
os.remove(model_path)
def test_pickled_context(self):
x, y = tm.make_sparse_regression(10, 10, sparsity=0.8, as_dense=True)
train_x = xgb.DMatrix(x, label=y)
param = {"tree_method": "gpu_hist", "verbosity": 1}
bst = xgb.train(param, train_x)
save_pickle(bst, model_path)
args = self.args_template.copy()
root = tm.project_root(__file__)
path = os.path.join(root, "tests", "python-gpu", "load_pickle.py")
args.append(path + "::TestLoadPickle::test_context_is_removed")
cuda_environment = {"CUDA_VISIBLE_DEVICES": "-1"}
env = os.environ.copy()
env.update(cuda_environment)
# Load model in a CPU only environment.
status = subprocess.call(args, env=env)
assert status == 0
args = self.args_template.copy()
args.append(
"./tests/python-gpu/"
"load_pickle.py::TestLoadPickle::test_context_is_preserved"
)
# Load in environment that has GPU.
env = os.environ.copy()
assert "CUDA_VISIBLE_DEVICES" not in env.keys()
status = subprocess.call(args, env=env)
assert status == 0
os.remove(model_path)
@pytest.mark.skipif(**tm.no_sklearn())
def test_predict_sklearn_pickle(self) -> None:
from sklearn.datasets import load_digits
x, y = load_digits(return_X_y=True)
kwargs = {
"tree_method": "gpu_hist",
"objective": "binary:logistic",
"gpu_id": 0,
"n_estimators": 10,
}
model = XGBClassifier(**kwargs)
model.fit(x, y)
save_pickle(model, "model.pkl")
del model
# load model
model = load_pickle("model.pkl")
os.remove("model.pkl")
gpu_pred = model.predict(x, output_margin=True)
# Switch to CPU predictor
bst = model.get_booster()
bst.set_param({"device": "cpu"})
cpu_pred = model.predict(x, output_margin=True)
np.testing.assert_allclose(cpu_pred, gpu_pred, rtol=1e-5)
def test_training_on_cpu_only_env(self):
cuda_environment = {"CUDA_VISIBLE_DEVICES": "-1"}
env = os.environ.copy()
env.update(cuda_environment)
args = self.args_template.copy()
args.append(
"./tests/python-gpu/"
"load_pickle.py::TestLoadPickle::test_training_on_cpu_only_env"
)
status = subprocess.call(args, env=env)
assert status == 0
|
