jens/pyRBM
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

improved tests

Browse Source
This commit is contained in:
jens
2026-01-06 09:44:45 +01:00
parent ec5d259ff7
commit 42520e5761
3 changed files with 17 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/tests/test_deep_norbs.py
+6 -10
View File
@@ -2,15 +2,14 @@ import os
import matplotlib.pyplot as plt import matplotlib.pyplot as plt
from rbm.model import Model from rbm.model import Model
from rbm.entity import Entity, EntityParams from rbm.entity import Entity, EntityParams, TrainingParams
from rbm.matrix import Mat, np, read_armadillo from rbm.matrix import Mat, np, read_armadillo
from rbm.train import TrainingParams
class TestModel(Model): class TestModel(Model):
def __init__(self, name: str, work_dir: str = '.'): def __init__(self, name: str, work_dir: str = '.'):
super().__init__(name, work_dir) super().__init__(name, work_dir)
self.unit1 = Entity((96*96, 16), EntityParams(do_gaussian_visible=True, do_gaussian_hidden=True)) self.unit1 = Entity((96*96, 333), EntityParams(do_gaussian_visible=True, do_gaussian_hidden=False), TrainingParams(learning_rate=0.001, momentum=0.9, num_epochs=1000), enable_training=False)
self.unit2 = Entity((16, 16), EntityParams()) self.unit2 = Entity((333, 256), EntityParams(), TrainingParams(learning_rate=0.1, momentum=0.9, num_epochs=10000, do_rao_blackwell=True))
def forward(self, x: Mat): def forward(self, x: Mat):
x = self.unit1.forward(x) x = self.unit1.forward(x)
@@ -31,7 +30,7 @@ if __name__ == "__main__":
model = TestModel(prj_name, "results") model = TestModel(prj_name, "results")
# Init state # Init state
model.init(0.01) model.init(0.1)
# load state # load state
model.load() model.load()
@@ -43,10 +42,7 @@ if __name__ == "__main__":
test_batch = read_armadillo(os.path.join(prj_root, f"norb_small_16h_v2.test.dat")) test_batch = read_armadillo(os.path.join(prj_root, f"norb_small_16h_v2.test.dat"))
# Train # Train
model.train(train_batch,[ model.train(train_batch)
TrainingParams(learning_rate=0.00001, momentum=0.9, do_rao_blackwell=True, num_epochs=1000, num_gibbs_samples=3),
TrainingParams(learning_rate=0.01, momentum=0.9, do_rao_blackwell=True, num_epochs=1000, num_gibbs_samples=1)
])
# save state # save state
model.save() model.save()
@@ -56,7 +52,7 @@ if __name__ == "__main__":
out_normalized = model.backward(model.forward(inp)) out_normalized = model.backward(model.forward(inp))
img = 2*(out_normalized + 0.5) img = 2*(out_normalized + 0.5)
img = np.reshape(img, (96, 96)) img = np.reshape(img, (96, 96))
axes[index].imshow(img) axes[index].imshow(np.asnumpy(img))
axes[index].axis('off') axes[index].axis('off')
plt.show() plt.show()
src/tests/test_model.py
+10 -8
View File
@@ -1,22 +1,24 @@
from rbm.model import Model from rbm.model import Model
from rbm.entity import Entity, EntityParams from rbm.entity import Entity, EntityParams, TrainingParams
from rbm.matrix import Mat, np from rbm.matrix import Mat, np
from rbm.train import TrainingParams
class TestModel(Model): class TestModel(Model):
def __init__(self, name: str, work_dir: str = '.'): def __init__(self, name: str, work_dir: str = '.'):
super().__init__(name, work_dir) super().__init__(name, work_dir)
self.unit1 = Entity((16, 64), EntityParams()) self.unit1 = Entity((1024, 333), EntityParams(), TrainingParams(learning_rate=0.1, momentum=0.9, do_rao_blackwell=True, num_epochs=1000))
self.unit2 = Entity((64, 16), EntityParams()) self.unit2 = Entity((333, 64), EntityParams(), TrainingParams(learning_rate=0.1, momentum=0.9, do_rao_blackwell=True, num_epochs=1000))
self.unit3 = Entity((16, 64), EntityParams()) self.unit3 = Entity((64, 128), EntityParams(), TrainingParams(learning_rate=0.1, momentum=0.9, do_rao_blackwell=True, num_epochs=1000))
self.unit4 = Entity((128, 128), EntityParams(), TrainingParams(learning_rate=0.1, momentum=0.9, do_rao_blackwell=True, num_epochs=1000))
def forward(self, x: Mat): def forward(self, x: Mat):
x = self.unit1.forward(x) x = self.unit1.forward(x)
x = self.unit2.forward(x) x = self.unit2.forward(x)
x = self.unit3.forward(x) x = self.unit3.forward(x)
x = self.unit4.forward(x)
return x return x
def backward(self, x: Mat): def backward(self, x: Mat):
x = self.unit4.reconstruct(x)
x = self.unit3.reconstruct(x) x = self.unit3.reconstruct(x)
x = self.unit2.reconstruct(x) x = self.unit2.reconstruct(x)
x = self.unit1.reconstruct(x) x = self.unit1.reconstruct(x)
@@ -27,16 +29,16 @@ if __name__ == "__main__":
model = TestModel("TestModel", "results") model = TestModel("TestModel", "results")
# Init state # Init state
model.init(0.01) model.init(0.1)
# load state # load state
model.load() model.load()
# create batch # create batch
batch = (np.random.rand(64, 16) > 0.5).astype(np.float64) batch = (np.random.rand(64, 1024) > 0.5).astype(np.float64)
# Train # Train
model.train(batch, TrainingParams(learning_rate=0.01, momentum=0.9, do_rao_blackwell=True, num_epochs=1000, num_gibbs_samples=3)) model.train(batch)
# save state # save state
model.save() model.save()
src/tests/test_norbs.py
+1 -1
View File
@@ -16,7 +16,7 @@ class TestModel(Model):
else: else:
# Hidden binary # Hidden binary
self.unit1 = Entity((96 * 96, 333), EntityParams(do_gaussian_visible=True, do_gaussian_hidden=False), self.unit1 = Entity((96 * 96, 333), EntityParams(do_gaussian_visible=True, do_gaussian_hidden=False),
TrainingParams(learning_rate=0.0001, momentum=0.9, num_epochs=1000)) TrainingParams(learning_rate=0.001, momentum=0.9, num_epochs=1000))
def forward(self, x: Mat): def forward(self, x: Mat):
x = self.unit1.forward(x) x = self.unit1.forward(x)