julius/prototorch_models
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Unclutter the examples folder

Browse Source
This commit is contained in:
Jensun Ravichandran 2021-05-07 15:21:35 +02:00
parent e87663d10c
commit f2541acde9
7 changed files with 0 additions and 760 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

112
examples/cbc_circle.py
View File

@ -1,112 +0,0 @@
"""CBC example using the Iris dataset."""
import numpy as np
import pytorch_lightning as pl
import torch
from matplotlib import pyplot as plt
from prototorch.components import initializers as cinit
from prototorch.datasets.abstract import NumpyDataset
from sklearn.datasets import make_circles
from torch.utils.data import DataLoader
from prototorch.models.cbc import CBC, euclidean_similarity
class VisualizationCallback(pl.Callback):
def __init__(
self,
x_train,
y_train,
prototype_model=True,
title="Prototype Visualization",
cmap="viridis",
):
super().__init__()
self.x_train = x_train
self.y_train = y_train
self.title = title
self.fig = plt.figure(self.title)
self.cmap = cmap
self.prototype_model = prototype_model
def on_epoch_end(self, trainer, pl_module):
if self.prototype_model:
protos = pl_module.components
color = pl_module.prototype_labels
else:
protos = pl_module.components
color = "k"
ax = self.fig.gca()
ax.cla()
ax.set_title(self.title)
ax.set_xlabel("Data dimension 1")
ax.set_ylabel("Data dimension 2")
ax.scatter(x_train[:, 0], x_train[:, 1], c=y_train, edgecolor="k")
ax.scatter(
protos[:, 0],
protos[:, 1],
c=color,
cmap=self.cmap,
edgecolor="k",
marker="D",
s=50,
)
x = np.vstack((x_train, protos))
x_min, x_max = x[:, 0].min() - 1, x[:, 0].max() + 1
y_min, y_max = x[:, 1].min() - 1, x[:, 1].max() + 1
xx, yy = np.meshgrid(np.arange(x_min, x_max, 1 / 50),
np.arange(y_min, y_max, 1 / 50))
mesh_input = np.c_[xx.ravel(), yy.ravel()]
y_pred = pl_module.predict(torch.Tensor(mesh_input))
y_pred = y_pred.reshape(xx.shape)
ax.contourf(xx, yy, y_pred, cmap=self.cmap, alpha=0.35)
ax.set_xlim(left=x_min + 0, right=x_max - 0)
ax.set_ylim(bottom=y_min + 0, top=y_max - 0)
plt.pause(0.1)
if __name__ == "__main__":
# Dataset
x_train, y_train = make_circles(n_samples=300,
shuffle=True,
noise=0.05,
random_state=None,
factor=0.5)
train_ds = NumpyDataset(x_train, y_train)
# Dataloaders
train_loader = DataLoader(train_ds, num_workers=0, batch_size=150)
# Hyperparameters
hparams = dict(
input_dim=x_train.shape[1],
nclasses=len(np.unique(y_train)),
num_components=5,
component_initializer=cinit.RandomInitializer(x_train.shape[1]),
lr=0.01,
)
# Initialize the model
model = CBC(
hparams,
data=[x_train, y_train],
similarity=euclidean_similarity,
)
# Callbacks
dvis = VisualizationCallback(x_train,
y_train,
prototype_model=False,
title="CBC Circle Example")
# Setup trainer
trainer = pl.Trainer(
max_epochs=50,
callbacks=[
dvis,
],
)
# Training loop
trainer.fit(model, train_loader)

128
examples/cbc_mnist.py
View File

@ -1,128 +0,0 @@
"""CBC example using the MNIST dataset.
This script also shows how to use Tensorboard for visualizing the prototypes.
"""
import argparse
import pytorch_lightning as pl
import torchvision
from torch.utils.data import DataLoader
from torchvision import transforms
from torchvision.datasets import MNIST
from prototorch.models.cbc import CBC, ImageCBC, euclidean_similarity
class VisualizationCallback(pl.Callback):
def __init__(self, to_shape=(-1, 1, 28, 28), nrow=2):
super().__init__()
self.to_shape = to_shape
self.nrow = nrow
def on_epoch_end(self, trainer, pl_module: ImageCBC):
tb = pl_module.logger.experiment
# components
components = pl_module.components
components_img = components.reshape(self.to_shape)
grid = torchvision.utils.make_grid(components_img, nrow=self.nrow)
tb.add_image(
tag="MNIST Components",
img_tensor=grid,
global_step=trainer.current_epoch,
dataformats="CHW",
)
# Reasonings
reasonings = pl_module.reasonings
tb.add_images(
tag="MNIST Reasoning",
img_tensor=reasonings,
global_step=trainer.current_epoch,
dataformats="NCHW",
)
if __name__ == "__main__":
# Arguments
parser = argparse.ArgumentParser()
parser.add_argument("--epochs",
type=int,
default=10,
help="Epochs to train.")
parser.add_argument("--lr",
type=float,
default=0.001,
help="Learning rate.")
parser.add_argument("--batch_size",
type=int,
default=256,
help="Batch size.")
parser.add_argument("--gpus",
type=int,
default=0,
help="Number of GPUs to use.")
parser.add_argument("--ppc",
type=int,
default=1,
help="Prototypes-Per-Class.")
args = parser.parse_args()
# Dataset
mnist_train = MNIST(
"./datasets",
train=True,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
]),
)
mnist_test = MNIST(
"./datasets",
train=False,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
]),
)
# Dataloaders
train_loader = DataLoader(mnist_train, batch_size=32)
test_loader = DataLoader(mnist_test, batch_size=32)
# Grab the full dataset to warm-start prototypes
x, y = next(iter(DataLoader(mnist_train, batch_size=len(mnist_train))))
x = x.view(len(mnist_train), -1)
# Hyperparameters
hparams = dict(
input_dim=28 * 28,
nclasses=10,
prototypes_per_class=args.ppc,
prototype_initializer="randn",
lr=0.01,
similarity=euclidean_similarity,
)
# Initialize the model
model = CBC(hparams, data=[x, y])
# Model summary
print(model)
# Callbacks
vis = VisualizationCallback(to_shape=(-1, 1, 28, 28), nrow=args.ppc)
# Setup trainer
trainer = pl.Trainer(
gpus=args.gpus, # change to use GPUs for training
max_epochs=args.epochs,
callbacks=[vis],
track_grad_norm=2,
# accelerator="ddp_cpu", # DEBUG-ONLY
# num_processes=2, # DEBUG-ONLY
)
# Training loop
trainer.fit(model, train_loader, test_loader)

135
examples/cbc_spiral.py
View File

@ -1,135 +0,0 @@
"""CBC example using the Iris dataset."""
import numpy as np
import pytorch_lightning as pl
import torch
from matplotlib import pyplot as plt
from prototorch.datasets.abstract import NumpyDataset
from torch.utils.data import DataLoader
from prototorch.models.cbc import CBC
class VisualizationCallback(pl.Callback):
def __init__(
self,
x_train,
y_train,
prototype_model=True,
title="Prototype Visualization",
cmap="viridis",
):
super().__init__()
self.x_train = x_train
self.y_train = y_train
self.title = title
self.fig = plt.figure(self.title)
self.cmap = cmap
self.prototype_model = prototype_model
def on_epoch_end(self, trainer, pl_module):
if self.prototype_model:
protos = pl_module.prototypes
color = pl_module.prototype_labels
else:
protos = pl_module.components
color = "k"
ax = self.fig.gca()
ax.cla()
ax.set_title(self.title)
ax.set_xlabel("Data dimension 1")
ax.set_ylabel("Data dimension 2")
ax.scatter(x_train[:, 0], x_train[:, 1], c=y_train, edgecolor="k")
ax.scatter(
protos[:, 0],
protos[:, 1],
c=color,
cmap=self.cmap,
edgecolor="k",
marker="D",
s=50,
)
x = np.vstack((x_train, protos))
x_min, x_max = x[:, 0].min() - 1, x[:, 0].max() + 1
y_min, y_max = x[:, 1].min() - 1, x[:, 1].max() + 1
xx, yy = np.meshgrid(np.arange(x_min, x_max, 1 / 50),
np.arange(y_min, y_max, 1 / 50))
mesh_input = np.c_[xx.ravel(), yy.ravel()]
y_pred = pl_module.predict(torch.Tensor(mesh_input))
y_pred = y_pred.reshape(xx.shape)
ax.contourf(xx, yy, y_pred, cmap=self.cmap, alpha=0.35)
ax.set_xlim(left=x_min + 0, right=x_max - 0)
ax.set_ylim(bottom=y_min + 0, top=y_max - 0)
plt.pause(0.1)
def make_spirals(n_samples=500, noise=0.3):
def get_samples(n, delta_t):
points = []
for i in range(n):
r = i / n_samples * 5
t = 1.75 * i / n * 2 * np.pi + delta_t
x = r * np.sin(t) + np.random.rand(1) * noise
y = r * np.cos(t) + np.random.rand(1) * noise
points.append([x, y])
return points
n = n_samples // 2
positive = get_samples(n=n, delta_t=0)
negative = get_samples(n=n, delta_t=np.pi)
x = np.concatenate(
[np.array(positive).reshape(n, -1),
np.array(negative).reshape(n, -1)],
axis=0)
y = np.concatenate([np.zeros(n), np.ones(n)])
return x, y
if __name__ == "__main__":
# Dataset
x_train, y_train = make_spirals(n_samples=1000, noise=0.3)
train_ds = NumpyDataset(x_train, y_train)
# Dataloaders
train_loader = DataLoader(train_ds, num_workers=0, batch_size=150)
# Hyperparameters
hparams = dict(
input_dim=x_train.shape[1],
nclasses=2,
prototypes_per_class=40,
prototype_initializer="stratified_random",
lr=0.05,
)
# Initialize the model
model_class = CBC
model = model_class(hparams, data=[x_train, y_train])
# Pure-positive reasonings
new_reasoning = torch.zeros_like(
model.reasoning_layer.reasoning_probabilities)
for i, label in enumerate(model.component_layer.prototype_labels):
new_reasoning[0][0][i][int(label)] = 1.0
model.reasoning_layer.reasoning_probabilities.data = new_reasoning
# Model summary
print(model)
# Callbacks
vis = VisualizationCallback(x_train,
y_train,
prototype_model=hasattr(model, "prototypes"))
# Setup trainer
trainer = pl.Trainer(
max_epochs=500,
callbacks=[
vis,
],
)
# Training loop
trainer.fit(model, train_loader)

147
examples/cbc_spiral_with_glvq_start.py
View File

@ -1,147 +0,0 @@
"""CBC example using the spirals dataset.
This example shows how to jump start a model by transferring weights from
another more stable model.
"""
import numpy as np
import pytorch_lightning as pl
import torch
from matplotlib import pyplot as plt
from prototorch.datasets.abstract import NumpyDataset
from torch.utils.data import DataLoader
from prototorch.models.cbc import CBC
from prototorch.models.glvq import GLVQ
class VisualizationCallback(pl.Callback):
def __init__(
self,
x_train,
y_train,
prototype_model=True,
title="Prototype Visualization",
cmap="viridis",
):
super().__init__()
self.x_train = x_train
self.y_train = y_train
self.title = title
self.fig = plt.figure(self.title)
self.cmap = cmap
self.prototype_model = prototype_model
def on_epoch_end(self, trainer, pl_module):
if self.prototype_model:
protos = pl_module.prototypes
color = pl_module.prototype_labels
else:
protos = pl_module.components
color = "k"
ax = self.fig.gca()
ax.cla()
ax.set_title(self.title)
ax.set_xlabel("Data dimension 1")
ax.set_ylabel("Data dimension 2")
ax.scatter(x_train[:, 0], x_train[:, 1], c=y_train, edgecolor="k")
ax.scatter(
protos[:, 0],
protos[:, 1],
c=color,
cmap=self.cmap,
edgecolor="k",
marker="D",
s=50,
)
x = np.vstack((x_train, protos))
x_min, x_max = x[:, 0].min() - 1, x[:, 0].max() + 1
y_min, y_max = x[:, 1].min() - 1, x[:, 1].max() + 1
xx, yy = np.meshgrid(np.arange(x_min, x_max, 1 / 50),
np.arange(y_min, y_max, 1 / 50))
mesh_input = np.c_[xx.ravel(), yy.ravel()]
y_pred = pl_module.predict(torch.Tensor(mesh_input))
y_pred = y_pred.reshape(xx.shape)
ax.contourf(xx, yy, y_pred, cmap=self.cmap, alpha=0.35)
ax.set_xlim(left=x_min + 0, right=x_max - 0)
ax.set_ylim(bottom=y_min + 0, top=y_max - 0)
plt.pause(0.1)
def make_spirals(n_samples=500, noise=0.3):
def get_samples(n, delta_t):
points = []
for i in range(n):
r = i / n_samples * 5
t = 1.75 * i / n * 2 * np.pi + delta_t
x = r * np.sin(t) + np.random.rand(1) * noise
y = r * np.cos(t) + np.random.rand(1) * noise
points.append([x, y])
return points
n = n_samples // 2
positive = get_samples(n=n, delta_t=0)
negative = get_samples(n=n, delta_t=np.pi)
x = np.concatenate(
[np.array(positive).reshape(n, -1),
np.array(negative).reshape(n, -1)],
axis=0)
y = np.concatenate([np.zeros(n), np.ones(n)])
return x, y
def train(model, x_train, y_train, train_loader, epochs=100):
# Callbacks
vis = VisualizationCallback(x_train,
y_train,
prototype_model=hasattr(model, "prototypes"))
# Setup trainer
trainer = pl.Trainer(
max_epochs=epochs,
callbacks=[
vis,
],
)
# Training loop
trainer.fit(model, train_loader)
if __name__ == "__main__":
# Dataset
x_train, y_train = make_spirals(n_samples=1000, noise=0.3)
train_ds = NumpyDataset(x_train, y_train)
# Dataloaders
train_loader = DataLoader(train_ds, num_workers=0, batch_size=150)
# Hyperparameters
hparams = dict(
input_dim=x_train.shape[1],
nclasses=2,
prototypes_per_class=40,
prototype_initializer="stratified_random",
lr=0.05,
)
# Initialize the model
glvq_model = GLVQ(hparams, data=[x_train, y_train])
cbc_model = CBC(hparams, data=[x_train, y_train])
# Train GLVQ
train(glvq_model, x_train, y_train, train_loader, epochs=10)
# Transfer Prototypes
cbc_model.component_layer.load_state_dict(
glvq_model.proto_layer.state_dict())
# Pure-positive reasonings
new_reasoning = torch.zeros_like(
cbc_model.reasoning_layer.reasoning_probabilities)
for i, label in enumerate(cbc_model.component_layer.prototype_labels):
new_reasoning[0][0][i][int(label)] = 1.0
new_reasoning[1][0][i][1 - int(label)] = 1.0
cbc_model.reasoning_layer.reasoning_probabilities.data = new_reasoning
# Train CBC
train(cbc_model, x_train, y_train, train_loader, epochs=50)

119
examples/glvq_mnist.py
View File

@ -1,119 +0,0 @@
"""GLVQ example using the MNIST dataset.
This script also shows how to use Tensorboard for visualizing the prototypes.
"""
import argparse
import pytorch_lightning as pl
import torchvision
from prototorch.components import initializers as cinit
from torch.utils.data import DataLoader
from torchvision import transforms
from torchvision.datasets import MNIST
from prototorch.models.glvq import ImageGLVQ
class VisualizationCallback(pl.Callback):
def __init__(self, to_shape=(-1, 1, 28, 28), nrow=2):
super().__init__()
self.to_shape = to_shape
self.nrow = nrow
def on_epoch_end(self, trainer, pl_module):
protos = pl_module.proto_layer.prototypes.detach().cpu()
protos_img = protos.reshape(self.to_shape)
grid = torchvision.utils.make_grid(protos_img, nrow=self.nrow)
# grid = grid.permute((1, 2, 0))
tb = pl_module.logger.experiment
tb.add_image(
tag="MNIST Prototypes",
img_tensor=grid,
global_step=trainer.current_epoch,
dataformats="CHW",
)
if __name__ == "__main__":
# Arguments
parser = argparse.ArgumentParser()
parser.add_argument("--epochs",
type=int,
default=10,
help="Epochs to train.")
parser.add_argument("--lr",
type=float,
default=0.001,
help="Learning rate.")
parser.add_argument("--batch_size",
type=int,
default=256,
help="Batch size.")
parser.add_argument("--gpus",
type=int,
default=0,
help="Number of GPUs to use.")
parser.add_argument("--ppc",
type=int,
default=1,
help="Prototypes-Per-Class.")
args = parser.parse_args()
# Dataset
mnist_train = MNIST(
"./datasets",
train=True,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
]),
)
mnist_test = MNIST(
"./datasets",
train=False,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
]),
)
# Dataloaders
train_loader = DataLoader(mnist_train, batch_size=1024)
test_loader = DataLoader(mnist_test, batch_size=1024)
# Grab the full dataset to warm-start prototypes
x, y = next(iter(DataLoader(mnist_train, batch_size=len(mnist_train))))
x = x.view(len(mnist_train), -1)
# Hyperparameters
hparams = dict(
input_dim=28 * 28,
nclasses=10,
prototypes_per_class=1,
prototype_initializer=cinit.StratifiedMeanInitializer(x, y),
lr=args.lr,
)
# Initialize the model
model = ImageGLVQ(hparams)
# Model summary
print(model)
# Callbacks
vis = VisualizationCallback(to_shape=(-1, 1, 28, 28), nrow=args.ppc)
# Setup trainer
trainer = pl.Trainer(
gpus=args.gpus, # change to use GPUs for training
max_epochs=args.epochs,
callbacks=[vis],
# accelerator="ddp_cpu", # DEBUG-ONLY
# num_processes=2, # DEBUG-ONLY
)
# Training loop
trainer.fit(model, train_loader, test_loader)

62
examples/grlvq_iris.py
View File

@ -1,62 +0,0 @@
"""GMLVQ example using all four dimensions of the Iris dataset."""
import pytorch_lightning as pl
import torch
from prototorch.components import initializers as cinit
from prototorch.datasets.abstract import NumpyDataset
from sklearn.datasets import load_iris
from torch.utils.data import DataLoader
from prototorch.models.callbacks.visualization import VisSiameseGLVQ2D
from prototorch.models.glvq import GRLVQ
from sklearn.preprocessing import StandardScaler
class PrintRelevanceCallback(pl.Callback):
def on_epoch_end(self, trainer, pl_module: GRLVQ):
print(pl_module.relevance_profile)
if __name__ == "__main__":
# Dataset
x_train, y_train = load_iris(return_X_y=True)
x_train = x_train[:, [0, 2]]
scaler = StandardScaler()
scaler.fit(x_train)
x_train = scaler.transform(x_train)
train_ds = NumpyDataset(x_train, y_train)
# Dataloaders
train_loader = DataLoader(train_ds,
num_workers=0,
batch_size=50,
shuffle=True)
# Hyperparameters
hparams = dict(
nclasses=3,
prototypes_per_class=1,
#prototype_initializer=cinit.SMI(torch.Tensor(x_train),
# torch.Tensor(y_train)),
prototype_initializer=cinit.UniformInitializer(2),
input_dim=x_train.shape[1],
lr=0.1,
#transfer_function="sigmoid_beta",
)
# Initialize the model
model = GRLVQ(hparams)
# Model summary
print(model)
# Callbacks
vis = VisSiameseGLVQ2D(x_train, y_train)
debug = PrintRelevanceCallback()
# Setup trainer
trainer = pl.Trainer(max_epochs=200, callbacks=[vis, debug])
# Training loop
trainer.fit(model, train_loader)

57
examples/grlvq_spiral.py
View File

@ -1,57 +0,0 @@
"""GMLVQ example using all four dimensions of the Iris dataset."""
import pytorch_lightning as pl
import torch
from prototorch.components import initializers as cinit
from prototorch.datasets.abstract import NumpyDataset
from sklearn.datasets import load_iris
from torch.utils.data import DataLoader
from prototorch.models.callbacks.visualization import VisSiameseGLVQ2D
from prototorch.models.glvq import GRLVQ
from sklearn.preprocessing import StandardScaler
from prototorch.datasets.spiral import make_spiral
class PrintRelevanceCallback(pl.Callback):
def on_epoch_end(self, trainer, pl_module: GRLVQ):
print(pl_module.relevance_profile)
if __name__ == "__main__":
# Dataset
x_train, y_train = make_spiral(n_samples=1000, noise=0.3)
train_ds = NumpyDataset(x_train, y_train)
# Dataloaders
train_loader = DataLoader(train_ds, num_workers=0, batch_size=150)
# Hyperparameters
hparams = dict(
nclasses=2,
prototypes_per_class=20,
prototype_initializer=cinit.SSI(torch.Tensor(x_train),
torch.Tensor(y_train)),
#prototype_initializer=cinit.UniformInitializer(2),
input_dim=x_train.shape[1],
lr=0.1,
#transfer_function="sigmoid_beta",
)
# Initialize the model
model = GRLVQ(hparams)
# Model summary
print(model)
# Callbacks
vis = VisSiameseGLVQ2D(x_train, y_train)
debug = PrintRelevanceCallback()
# Setup trainer
trainer = pl.Trainer(max_epochs=200, callbacks=[vis, debug])
# Training loop
trainer.fit(model, train_loader)