opensource/ManimML
1
0
Fork 0
mirror of https://github.com/helblazer811/ManimML.git synced 2025-05-17 10:45:54 +08:00
Code Issues Packages Projects Releases Wiki Activity

Refactored makefile, moved around files to more appropriate places, deleted unused files,

Browse Source 
reformatted code.
This commit is contained in:
Alec Helbling
2023-02-01 22:33:42 -05:00
parent 27d235de25
commit 1406acd43e
69 changed files with 372 additions and 372 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
Makefile
View File

@ -1,18 +1,9 @@
setup:
conda activate manim
conda activate manim; \
export PROJECT_ROOT=$(pwd)
video:
manim -pqh src/variational_autoencoder.py VAEScene --media_dir media
cp media/videos/vae/720p60/VAEScene.mp4 examples
train:
cd src/autoencoder_models
python vanilla_autoencoder.py
python variational_autoencoder.py
generate_visualizations:
cd src/autoencoder_models
python generate_images.py
python generate_interpolation.py
python generate_disentanglement.py
checkstyle:
pycodestyle src
pydocstyle src
black .; \
pydocstyle .
publish_pip:
python3 -m build; \
python3 -m twine upload --repository pypi dist/*

6
examples/basic_neural_network/basic_neural_network.py
View File

@ -8,11 +8,7 @@ class NeuralNetworkScene(Scene):
def construct(self):
# Make the Layer object
layers = [
FeedForwardLayer(3),
FeedForwardLayer(5),
FeedForwardLayer(3)
]
layers = [FeedForwardLayer(3), FeedForwardLayer(5), FeedForwardLayer(3)]
nn = NeuralNetwork(layers)
nn.scale(2)
nn.move_to(ORIGIN)

4
examples/cnn/activation_functions.py
View File

@ -15,6 +15,7 @@ config.frame_height = 7.0
config.frame_width = 7.0
ROOT_DIR = Path(__file__).parents[2]
def make_code_snippet():
code_str = """
# Make the neural network
@ -42,6 +43,7 @@ def make_code_snippet():
return code
class CombinedScene(ThreeDScene):
def construct(self):
image = Image.open(ROOT_DIR / "assets/mnist/digit.jpeg")
@ -72,4 +74,4 @@ class CombinedScene(ThreeDScene):
# Play animation
forward_pass = nn.make_forward_pass_animation()
self.wait(1)
self.play(forward_pass)
self.play(forward_pass)

2
examples/cnn/cnn.py
View File

@ -15,6 +15,7 @@ config.frame_height = 7.0
config.frame_width = 7.0
ROOT_DIR = Path(__file__).parents[2]
def make_code_snippet():
code_str = """
# Make nn
@ -44,6 +45,7 @@ def make_code_snippet():
return code
class CombinedScene(ThreeDScene):
def construct(self):
image = Image.open(ROOT_DIR / "assets/mnist/digit.jpeg")

7
examples/cnn/cnn_max_pool.py
View File

@ -14,6 +14,7 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
def make_code_snippet():
code_str = """
# Make the neural network
@ -42,12 +43,14 @@ def make_code_snippet():
return code
class CombinedScene(ThreeDScene):
def construct(self):
image = Image.open("../../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(1, 8, filter_spacing=0.32),
MaxPooling2DLayer(kernel_size=2),
@ -67,4 +70,4 @@ class CombinedScene(ThreeDScene):
# Play animation
forward_pass = nn.make_forward_pass_animation()
self.wait(1)
self.play(forward_pass)
self.play(forward_pass)

1
examples/cnn/one_by_one_convolution.py
View File

@ -15,6 +15,7 @@ config.frame_height = 7.0
config.frame_width = 7.0
ROOT_DIR = Path(__file__).parents[2]
def make_code_snippet():
code_str = """
# Make nn

16
examples/cnn/padding_example.py
View File

@ -15,6 +15,7 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
def make_code_snippet():
code_str = """
# Make nn
@ -43,27 +44,28 @@ def make_code_snippet():
return code
class CombinedScene(ThreeDScene):
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
image = Image.open(ROOT_DIR / "assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(
num_feature_maps=1,
feature_map_size=6,
num_feature_maps=1,
feature_map_size=6,
padding=1,
padding_dashed=True
padding_dashed=True,
),
Convolutional2DLayer(
num_feature_maps=3,
num_feature_maps=3,
feature_map_size=6,
filter_size=3,
padding=0,
padding_dashed=False
padding_dashed=False,
),
FeedForwardLayer(3),
FeedForwardLayer(1),

11
examples/cnn/resnet_block.py
View File

@ -9,6 +9,7 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
def make_code_snippet():
code_str = """
# Make the neural network
@ -37,13 +38,14 @@ def make_code_snippet():
return code
class ConvScene(ThreeDScene):
class ConvScene(ThreeDScene):
def construct(self):
image = Image.open("../../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
nn = NeuralNetwork({
nn = NeuralNetwork(
{
"layer1": Convolutional2DLayer(1, 5, padding=1),
"layer2": Convolutional2DLayer(1, 5, 3, padding=1),
"layer3": Convolutional2DLayer(1, 5, 3, padding=1),
@ -60,7 +62,4 @@ class ConvScene(ThreeDScene):
self.add(code)
Group(code, nn).move_to(ORIGIN)
self.play(
nn.make_forward_pass_animation(),
run_time=8
)
self.play(nn.make_forward_pass_animation(), run_time=8)

4
examples/lenet/lenet.py
View File

@ -18,12 +18,14 @@ config.frame_width = 20.0
ROOT_DIR = Path(__file__).parents[2]
class CombinedScene(ThreeDScene):
def construct(self):
image = Image.open(ROOT_DIR / "assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=4.5),
Convolutional2DLayer(1, 28),
Convolutional2DLayer(6, 28, 5),

2
examples/paper_visualizations/oracle_guidance/oracle_guidance.py
View File

@ -13,7 +13,7 @@ from manim_ml.neural_network.layers import FeedForwardLayer, EmbeddingLayer
from manim_ml.neural_network.layers.util import get_connective_layer
import os
from manim_ml.probability import GaussianDistribution
from manim_ml.utils.mobjects.probability import GaussianDistribution
# Make the specific scene
config.pixel_height = 1200

13
examples/readme_example/a_simple_feed_forward_network.py
View File

@ -8,14 +8,17 @@ config.pixel_width = 1200
config.frame_height = 4.0
config.frame_width = 4.0
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
FeedForwardLayer(num_nodes=3),
FeedForwardLayer(num_nodes=5),
FeedForwardLayer(num_nodes=3)
])
nn = NeuralNetwork(
[
FeedForwardLayer(num_nodes=3),
FeedForwardLayer(num_nodes=5),
FeedForwardLayer(num_nodes=3),
]
)
self.add(nn)
# Center the nn
nn.move_to(ORIGIN)

10
examples/readme_example/activation_functions.py
View File

@ -10,12 +10,16 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(1, 7, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32, activation_function="ReLU"),
Convolutional2DLayer(
3, 5, 3, filter_spacing=0.32, activation_function="ReLU"
),
FeedForwardLayer(3, activation_function="Sigmoid"),
],
layer_spacing=0.25,
@ -26,4 +30,4 @@ class CombinedScene(ThreeDScene):
# Play animation
forward_pass = nn.make_forward_pass_animation()
self.play(ChangeSpeed(forward_pass, speedinfo={}), run_time=10)
self.wait(1)
self.wait(1)

10
examples/readme_example/animating_the_forward_pass.py
View File

@ -1,6 +1,10 @@
from manim import *
from manim_ml.neural_network import Convolutional2DLayer, FeedForwardLayer, NeuralNetwork
from manim_ml.neural_network import (
Convolutional2DLayer,
FeedForwardLayer,
NeuralNetwork,
)
# Make the specific scene
config.pixel_height = 700
@ -8,10 +12,12 @@ config.pixel_width = 1900
config.frame_height = 7.0
config.frame_width = 7.0
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(1, 7, 3, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),
Convolutional2DLayer(5, 3, 3, filter_spacing=0.18),

11
examples/readme_example/convolutional_neural_network_with_images.py
View File

@ -2,7 +2,12 @@ from manim import *
from PIL import Image
import numpy as np
from manim_ml.neural_network import Convolutional2DLayer, FeedForwardLayer, NeuralNetwork, ImageLayer
from manim_ml.neural_network import (
Convolutional2DLayer,
FeedForwardLayer,
NeuralNetwork,
ImageLayer,
)
# Make the specific scene
config.pixel_height = 700
@ -10,13 +15,15 @@ config.pixel_width = 1900
config.frame_height = 7.0
config.frame_width = 7.0
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
image = Image.open("../../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(1, 7, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),

10
examples/readme_example/convolutional_neural_networks.py
View File

@ -1,6 +1,10 @@
from manim import *
from manim_ml.neural_network import Convolutional2DLayer, FeedForwardLayer, NeuralNetwork
from manim_ml.neural_network import (
Convolutional2DLayer,
FeedForwardLayer,
NeuralNetwork,
)
# Make the specific scene
config.pixel_height = 700
@ -8,10 +12,12 @@ config.pixel_width = 1900
config.frame_height = 7.0
config.frame_width = 7.0
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(1, 7, 3, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),
Convolutional2DLayer(5, 3, 3, filter_spacing=0.18),

10
examples/readme_example/example.py
View File

@ -1,6 +1,10 @@
from manim import *
from manim_ml.neural_network import Convolutional2DLayer, FeedForwardLayer, NeuralNetwork
from manim_ml.neural_network import (
Convolutional2DLayer,
FeedForwardLayer,
NeuralNetwork,
)
# Make the specific scene
config.pixel_height = 700
@ -8,10 +12,12 @@ config.pixel_width = 1900
config.frame_height = 7.0
config.frame_width = 7.0
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(1, 7, 3, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),
Convolutional2DLayer(5, 3, 3, filter_spacing=0.18),

10
examples/readme_example/first_neural_network.py
View File

@ -1,6 +1,10 @@
from manim import *
from manim_ml.neural_network import Convolutional2DLayer, FeedForwardLayer, NeuralNetwork
from manim_ml.neural_network import (
Convolutional2DLayer,
FeedForwardLayer,
NeuralNetwork,
)
# Make the specific scene
config.pixel_height = 700
@ -8,10 +12,12 @@ config.pixel_width = 1900
config.frame_height = 7.0
config.frame_width = 7.0
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(1, 7, 3, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),
Convolutional2DLayer(5, 3, 3, filter_spacing=0.18),

7
examples/readme_example/max_pooling.py
View File

@ -12,11 +12,12 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class MaxPoolingScene(ThreeDScene):
class MaxPoolingScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(1, 8),
Convolutional2DLayer(3, 6, 3),
MaxPooling2DLayer(kernel_size=2),
@ -30,4 +31,4 @@ class MaxPoolingScene(ThreeDScene):
# Play animation
forward_pass = nn.make_forward_pass_animation()
self.play(ChangeSpeed(forward_pass, speedinfo={}), run_time=10)
self.wait(1)
self.wait(1)

7
examples/readme_example/neural_network_dropout.py
View File

@ -9,11 +9,12 @@ config.pixel_width = 1900
config.frame_height = 5.0
config.frame_width = 5.0
class DropoutNeuralNetworkScene(Scene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
FeedForwardLayer(3, rectangle_color=BLUE),
FeedForwardLayer(5, rectangle_color=BLUE),
FeedForwardLayer(3, rectangle_color=BLUE),
@ -31,4 +32,4 @@ class DropoutNeuralNetworkScene(Scene):
nn, dropout_rate=0.25, do_forward_pass=True
)
)
self.wait(1)
self.wait(1)

9
examples/readme_example/old_example.py
View File

@ -1,4 +1,4 @@
from manim import *
from manim import *
from PIL import Image
from manim_ml.neural_network.layers.convolutional_2d import Convolutional2DLayer
@ -6,11 +6,12 @@ from manim_ml.neural_network.layers.feed_forward import FeedForwardLayer
from manim_ml.neural_network.layers.image import ImageLayer
from manim_ml.neural_network.neural_network import NeuralNetwork
class ConvolutionalNetworkScene(Scene):
class ConvolutionalNetworkScene(Scene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(1, 7, 3, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),
Convolutional2DLayer(5, 3, 3, filter_spacing=0.18),
@ -22,4 +23,4 @@ class ConvolutionalNetworkScene(Scene):
# Center the nn
nn.move_to(ORIGIN)
self.add(nn)
self.play(nn.make_forward_pass_animation())
self.play(nn.make_forward_pass_animation())

5
examples/readme_example/setting_up_a_scene.py
View File

@ -1,9 +1,10 @@
from manim import *
# Import modules here
class BasicScene(ThreeDScene):
class BasicScene(ThreeDScene):
def construct(self):
# Your code goes here
text = Text("Your first scene!")
self.add(text)
self.add(text)

2
manim_ml/diffusion/mcmc.py
View File

@ -7,7 +7,7 @@ import scipy
import scipy.stats
from tqdm import tqdm
from manim_ml.probability import GaussianDistribution
from manim_ml.utils.mobjects.probability import GaussianDistribution
def gaussian_proposal(x, sigma=0.2):

0
manim_ml/flow/__init__.py
View File

22
manim_ml/flow/flow.py
View File

@ -1,22 +0,0 @@
"""
Animated flow charts.
"""
from manim import *
class FlowGraph(VGroup):
"""Graph container"""
pass
class FlowNode(VGroup):
"""Node in the FlowGraph"""
pass
class DataNode(FlowNode):
"""Node that outputs data"""
pass

13
manim_ml/lazy_animation.py
View File

@ -1,13 +0,0 @@
from manim import *
class LazyAnimation(Animation):
def __init__(self, animation_function):
self.animation_function = animation_function
super.__init__()
def begin(self):
update_func_anim = UpdateFromFunc(self.neural_network, create_new_connective)
self.add
super.begin()

3
manim_ml/manifold.py
View File

@ -1,3 +0,0 @@
"""
Visaulization of a latent Manifold
"""

38
manim_ml/neural_network/__init__.py
View File

@ -1,23 +1,41 @@
from manim_ml.neural_network.neural_network import NeuralNetwork
from manim_ml.neural_network.layers.feed_forward import FeedForwardLayer
from manim_ml.neural_network.layers.convolutional_2d_to_convolutional_2d import Convolutional2DToConvolutional2D
from manim_ml.neural_network.layers.convolutional_2d_to_feed_forward import Convolutional2DToFeedForward
from manim_ml.neural_network.layers.convolutional_2d_to_max_pooling_2d import Convolutional2DToMaxPooling2D
from manim_ml.neural_network.layers.convolutional_2d_to_convolutional_2d import (
Convolutional2DToConvolutional2D,
)
from manim_ml.neural_network.layers.convolutional_2d_to_feed_forward import (
Convolutional2DToFeedForward,
)
from manim_ml.neural_network.layers.convolutional_2d_to_max_pooling_2d import (
Convolutional2DToMaxPooling2D,
)
from manim_ml.neural_network.layers.convolutional_2d import Convolutional2DLayer
from manim_ml.neural_network.layers.embedding_to_feed_forward import EmbeddingToFeedForward
from manim_ml.neural_network.layers.embedding_to_feed_forward import (
EmbeddingToFeedForward,
)
from manim_ml.neural_network.layers.embedding import EmbeddingLayer
from manim_ml.neural_network.layers.feed_forward_to_embedding import FeedForwardToEmbedding
from manim_ml.neural_network.layers.feed_forward_to_feed_forward import FeedForwardToFeedForward
from manim_ml.neural_network.layers.feed_forward_to_embedding import (
FeedForwardToEmbedding,
)
from manim_ml.neural_network.layers.feed_forward_to_feed_forward import (
FeedForwardToFeedForward,
)
from manim_ml.neural_network.layers.feed_forward_to_image import FeedForwardToImage
from manim_ml.neural_network.layers.feed_forward_to_vector import FeedForwardToVector
from manim_ml.neural_network.layers.feed_forward import FeedForwardLayer
from manim_ml.neural_network.layers.image_to_convolutional_2d import ImageToConvolutional2DLayer
from manim_ml.neural_network.layers.image_to_convolutional_2d import (
ImageToConvolutional2DLayer,
)
from manim_ml.neural_network.layers.image_to_feed_forward import ImageToFeedForward
from manim_ml.neural_network.layers.image import ImageLayer
from manim_ml.neural_network.layers.max_pooling_2d_to_convolutional_2d import MaxPooling2DToConvolutional2D
from manim_ml.neural_network.layers.max_pooling_2d_to_convolutional_2d import (
MaxPooling2DToConvolutional2D,
)
from manim_ml.neural_network.layers.max_pooling_2d import MaxPooling2DLayer
from manim_ml.neural_network.layers.paired_query_to_feed_forward import PairedQueryToFeedForward
from manim_ml.neural_network.layers.paired_query_to_feed_forward import (
PairedQueryToFeedForward,
)
from manim_ml.neural_network.layers.paired_query import PairedQueryLayer
from manim_ml.neural_network.layers.triplet_to_feed_forward import TripletToFeedForward
from manim_ml.neural_network.layers.triplet import TripletLayer
from manim_ml.neural_network.layers.vector import VectorLayer
from manim_ml.neural_network.layers.vector import VectorLayer

11
manim_ml/neural_network/activation_functions/__init__.py
View File

@ -1,13 +1,12 @@
from manim_ml.neural_network.activation_functions.relu import ReLUFunction
from manim_ml.neural_network.activation_functions.sigmoid import SigmoidFunction
name_to_activation_function_map = {
"ReLU": ReLUFunction,
"Sigmoid": SigmoidFunction
}
name_to_activation_function_map = {"ReLU": ReLUFunction, "Sigmoid": SigmoidFunction}
def get_activation_function_by_name(name):
assert name in name_to_activation_function_map.keys(), \
f"Unrecognized activation function {name}"
assert (
name in name_to_activation_function_map.keys()
), f"Unrecognized activation function {name}"
return name_to_activation_function_map[name]

1
manim_ml/neural_network/activation_functions/activation_function.py
View File

@ -4,6 +4,7 @@ import random
import manim_ml.neural_network.activation_functions.relu as relu
class ActivationFunction(ABC, VGroup):
"""Abstract parent class for defining activation functions"""

1
manim_ml/neural_network/activation_functions/sigmoid.py
View File

@ -5,6 +5,7 @@ from manim_ml.neural_network.activation_functions.activation_function import (
ActivationFunction,
)
class SigmoidFunction(ActivationFunction):
"""Sigmoid Activation Function"""

0
manim_ml/neural_network/neural_network_transformations.py → manim_ml/neural_network/animations/neural_network_transformations.py
View File

1
manim_ml/neural_network/architectures/feed_forward.py
View File

@ -1,5 +1,6 @@
from manim_ml.neural_network.layers.feed_forward import FeedForwardLayer
class FeedForwardNeuralNetwork(NeuralNetwork):
"""NeuralNetwork with just feed forward layers"""

4
manim_ml/neural_network/layers/__init__.py
View File

@ -10,7 +10,9 @@ from manim_ml.neural_network.layers.image_to_convolutional_2d import (
from manim_ml.neural_network.layers.max_pooling_2d_to_convolutional_2d import (
MaxPooling2DToConvolutional2D,
)
from manim_ml.neural_network.layers.max_pooling_2d_to_feed_forward import MaxPooling2DToFeedForward
from manim_ml.neural_network.layers.max_pooling_2d_to_feed_forward import (
MaxPooling2DToFeedForward,
)
from .convolutional_2d_to_convolutional_2d import Convolutional2DToConvolutional2D
from .convolutional_2d import Convolutional2DLayer
from .feed_forward_to_vector import FeedForwardToVector

28
manim_ml/neural_network/layers/convolutional_2d.py
View File

@ -10,7 +10,8 @@ from manim_ml.neural_network.layers.parent_layers import (
ThreeDLayer,
VGroupNeuralNetworkLayer,
)
from manim_ml.gridded_rectangle import GriddedRectangle
from manim_ml.utils.mobjects.gridded_rectangle import GriddedRectangle
class FeatureMap(VGroup):
"""Class for making a feature map"""
@ -25,7 +26,7 @@ class FeatureMap(VGroup):
padding=(0, 0),
stroke_width=2.0,
show_grid_lines=False,
padding_dashed=False
padding_dashed=False,
):
super().__init__()
self.color = color
@ -40,8 +41,12 @@ class FeatureMap(VGroup):
# Check if we have non-zero padding
if padding[0] > 0 or padding[1] > 0:
# Make the exterior rectangle dashed
width_with_padding = (self.feature_map_size[0] + self.padding[0] * 2) * self.cell_width
height_with_padding = (self.feature_map_size[1] + self.padding[1] * 2) * self.cell_width
width_with_padding = (
self.feature_map_size[0] + self.padding[0] * 2
) * self.cell_width
height_with_padding = (
self.feature_map_size[1] + self.padding[1] * 2
) * self.cell_width
self.untransformed_width = width_with_padding
self.untransformed_height = height_with_padding
@ -58,7 +63,7 @@ class FeatureMap(VGroup):
grid_stroke_width=self.stroke_width / 2,
grid_stroke_color=self.color,
show_grid_lines=self.show_grid_lines,
dotted_lines=self.padding_dashed
dotted_lines=self.padding_dashed,
)
self.add(self.exterior_rectangle)
# Add an interior rectangle with no fill color
@ -67,13 +72,13 @@ class FeatureMap(VGroup):
fill_opacity=0.0,
width=self.feature_map_size[0] * self.cell_width,
height=self.feature_map_size[1] * self.cell_width,
stroke_width=self.stroke_width
stroke_width=self.stroke_width,
)
self.add(self.interior_rectangle)
else:
# Just make an exterior rectangle with no dashes.
self.untransformed_height = self.feature_map_size[1] * self.cell_width,
self.untransformed_width = self.feature_map_size[0] * self.cell_width,
self.untransformed_height = (self.feature_map_size[1] * self.cell_width,)
self.untransformed_width = (self.feature_map_size[0] * self.cell_width,)
# Make the exterior rectangle
self.exterior_rectangle = GriddedRectangle(
color=self.color,
@ -96,6 +101,7 @@ class FeatureMap(VGroup):
# Sort points through clockwise rotation of a vector in the xy plane
return self.exterior_rectangle.get_corners_dict()
class Convolutional2DLayer(VGroupNeuralNetworkLayer, ThreeDLayer):
"""Handles rendering a convolutional layer for a nn"""
@ -215,7 +221,7 @@ class Convolutional2DLayer(VGroupNeuralNetworkLayer, ThreeDLayer):
fill_color=self.color,
fill_opacity=self.fill_opacity,
padding=self.padding,
padding_dashed=self.padding_dashed
padding_dashed=self.padding_dashed,
)
# Move the feature map
feature_map.move_to([0, 0, filter_index * self.filter_spacing])
@ -231,9 +237,7 @@ class Convolutional2DLayer(VGroupNeuralNetworkLayer, ThreeDLayer):
ApplyMethod(self.feature_maps.set_color, self.color),
)
def make_forward_pass_animation(
self, run_time=5, layer_args={}, **kwargs
):
def make_forward_pass_animation(self, run_time=5, layer_args={}, **kwargs):
"""Convolution forward pass animation"""
# Note: most of this animation is done in the Convolution3DToConvolution3D layer
if not self.activation_function is None:

19
manim_ml/neural_network/layers/convolutional_2d_to_convolutional_2d.py
View File

@ -3,10 +3,11 @@ import numpy as np
from manim import *
from manim_ml.neural_network.layers.convolutional_2d import Convolutional2DLayer
from manim_ml.neural_network.layers.parent_layers import ConnectiveLayer, ThreeDLayer
from manim_ml.gridded_rectangle import GriddedRectangle
from manim_ml.utils.mobjects.gridded_rectangle import GriddedRectangle
from manim.utils.space_ops import rotation_matrix
def get_rotated_shift_vectors(input_layer, normalized=False):
"""Rotates the shift vectors"""
# Make base shift vectors
@ -24,6 +25,7 @@ def get_rotated_shift_vectors(input_layer, normalized=False):
return right_shift, down_shift
class Filters(VGroup):
"""Group for showing a collection of filters connecting two layers"""
@ -144,7 +146,7 @@ class Filters(VGroup):
)
# Shift based on the amount of output layer padding
rectangle.shift(
self.output_layer.padding[0] * right_shift,
self.output_layer.padding[0] * right_shift,
)
rectangle.shift(
self.output_layer.padding[1] * down_shift,
@ -446,10 +448,7 @@ class Convolutional2DToConvolutional2D(ConnectiveLayer, ThreeDLayer):
# Do last row move right
for x_move in range(num_x_moves):
# Shift right
shift_animation = ApplyMethod(
filters.shift,
self.stride * right_shift
)
shift_animation = ApplyMethod(filters.shift, self.stride * right_shift)
# shift_animation = self.animate.shift(right_shift)
animations.append(shift_animation)
# Remove the filters
@ -460,18 +459,14 @@ class Convolutional2DToConvolutional2D(ConnectiveLayer, ThreeDLayer):
# Change the output feature map colors
change_color_animations = []
change_color_animations.append(
ApplyMethod(
feature_map.set_color,
original_feature_map_color
)
ApplyMethod(feature_map.set_color, original_feature_map_color)
)
# Change the input feature map colors
input_feature_maps = self.input_layer.feature_maps
for input_feature_map in input_feature_maps:
change_color_animations.append(
ApplyMethod(
input_feature_map.set_color,
original_feature_map_color
input_feature_map.set_color, original_feature_map_color
)
)
# Combine the animations

36
manim_ml/neural_network/layers/convolutional_2d_to_max_pooling_2d.py
View File

@ -1,6 +1,6 @@
import random
from manim import *
from manim_ml.gridded_rectangle import GriddedRectangle
from manim_ml.utils.mobjects.gridded_rectangle import GriddedRectangle
from manim_ml.neural_network.layers.convolutional_2d_to_convolutional_2d import (
get_rotated_shift_vectors,
)
@ -10,6 +10,7 @@ from manim_ml.neural_network.layers.parent_layers import ConnectiveLayer, ThreeD
from manim_ml.neural_network.layers.feed_forward import FeedForwardLayer
from manim_ml.neural_network.layers.convolutional_2d import Convolutional2DLayer
class Uncreate(Create):
def __init__(
self,
@ -27,6 +28,7 @@ class Uncreate(Create):
**kwargs,
)
class Convolutional2DToMaxPooling2D(ConnectiveLayer, ThreeDLayer):
"""Feed Forward to Embedding Layer"""
@ -38,7 +40,7 @@ class Convolutional2DToMaxPooling2D(ConnectiveLayer, ThreeDLayer):
input_layer: Convolutional2DLayer,
output_layer: MaxPooling2DLayer,
active_color=ORANGE,
**kwargs
**kwargs,
):
super().__init__(input_layer, output_layer, **kwargs)
self.active_color = active_color
@ -47,7 +49,7 @@ class Convolutional2DToMaxPooling2D(ConnectiveLayer, ThreeDLayer):
self,
input_layer: "NeuralNetworkLayer",
output_layer: "NeuralNetworkLayer",
**kwargs
**kwargs,
):
return super().construct_layer(input_layer, output_layer, **kwargs)
@ -94,7 +96,7 @@ class Convolutional2DToMaxPooling2D(ConnectiveLayer, ThreeDLayer):
width=cell_width,
fill_opacity=0.7,
stroke_width=0.0,
z_index=10
z_index=10,
)
# Move to the correct location
kernel_shift_vector = [
@ -119,10 +121,7 @@ class Convolutional2DToMaxPooling2D(ConnectiveLayer, ThreeDLayer):
highlighted_cells.append(cell_rectangle)
# Rotate the gridded rectangles so they match the angle
# of the conv maps
gridded_rectangle_group = VGroup(
gridded_rectangle,
*highlighted_cells
)
gridded_rectangle_group = VGroup(gridded_rectangle, *highlighted_cells)
gridded_rectangle_group.rotate(
ThreeDLayer.rotation_angle,
about_point=gridded_rectangle.get_center(),
@ -137,26 +136,20 @@ class Convolutional2DToMaxPooling2D(ConnectiveLayer, ThreeDLayer):
create_rectangle = Create(
gridded_rectangle,
)
create_gridded_rectangle_animations.append(
create_rectangle
)
create_gridded_rectangle_animations.append(create_rectangle)
# 4. Create and fade out highlighted cells
create_group = AnimationGroup(
*[Create(highlighted_cell) for highlighted_cell in highlighted_cells],
lag_ratio=1.0
lag_ratio=1.0,
)
uncreate_group = AnimationGroup(
*[Uncreate(highlighted_cell) for highlighted_cell in highlighted_cells],
lag_ratio=0.0
lag_ratio=0.0,
)
create_and_remove_cell_animation = Succession(
create_group,
Wait(1.0),
uncreate_group
)
create_and_remove_cell_animations.append(
create_and_remove_cell_animation
create_group, Wait(1.0), uncreate_group
)
create_and_remove_cell_animations.append(create_and_remove_cell_animation)
# 5. Move and resize the gridded rectangle to the output
# feature maps.
output_gridded_rectangle = GriddedRectangle(
@ -178,9 +171,10 @@ class Convolutional2DToMaxPooling2D(ConnectiveLayer, ThreeDLayer):
self.output_layer.feature_maps[feature_map_index].copy()
)
transform_rectangle = ReplacementTransform(
gridded_rectangle, output_gridded_rectangle,
gridded_rectangle,
output_gridded_rectangle,
introducer=True,
remover=True
remover=True,
)
transform_gridded_rectangle_animations.append(
transform_rectangle,

2
manim_ml/neural_network/layers/embedding.py
View File

@ -1,5 +1,5 @@
from manim import *
from manim_ml.probability import GaussianDistribution
from manim_ml.utils.mobjects.probability import GaussianDistribution
from manim_ml.neural_network.layers.parent_layers import VGroupNeuralNetworkLayer

5
manim_ml/neural_network/layers/feed_forward.py
View File

@ -1,9 +1,12 @@
from manim import *
from manim_ml.neural_network.activation_functions import get_activation_function_by_name
from manim_ml.neural_network.activation_functions.activation_function import ActivationFunction
from manim_ml.neural_network.activation_functions.activation_function import (
ActivationFunction,
)
from manim_ml.neural_network.layers.parent_layers import VGroupNeuralNetworkLayer
class FeedForwardLayer(VGroupNeuralNetworkLayer):
"""Handles rendering a layer for a neural network"""

2
manim_ml/neural_network/layers/image.py
View File

@ -1,6 +1,6 @@
from manim import *
import numpy as np
from manim_ml.image import GrayscaleImageMobject
from manim_ml.utils.mobjects.image import GrayscaleImageMobject
from manim_ml.neural_network.layers.parent_layers import NeuralNetworkLayer
from PIL import Image

2
manim_ml/neural_network/layers/image_to_convolutional_2d.py
View File

@ -7,7 +7,7 @@ from manim_ml.neural_network.layers.parent_layers import (
ThreeDLayer,
VGroupNeuralNetworkLayer,
)
from manim_ml.gridded_rectangle import GriddedRectangle
from manim_ml.utils.mobjects.gridded_rectangle import GriddedRectangle
class ImageToConvolutional2DLayer(VGroupNeuralNetworkLayer, ThreeDLayer):

2
manim_ml/neural_network/layers/max_pooling_2d.py
View File

@ -1,5 +1,5 @@
from manim import *
from manim_ml.gridded_rectangle import GriddedRectangle
from manim_ml.utils.mobjects.gridded_rectangle import GriddedRectangle
from manim_ml.neural_network.layers.parent_layers import (
ThreeDLayer,

5
manim_ml/neural_network/layers/max_pooling_2d_to_feed_forward.py
View File

@ -1,8 +1,11 @@
from manim import *
from manim_ml.neural_network.layers.convolutional_2d_to_feed_forward import Convolutional2DToFeedForward
from manim_ml.neural_network.layers.convolutional_2d_to_feed_forward import (
Convolutional2DToFeedForward,
)
from manim_ml.neural_network.layers.feed_forward import FeedForwardLayer
from manim_ml.neural_network.layers.max_pooling_2d import MaxPooling2DLayer
class MaxPooling2DToFeedForward(Convolutional2DToFeedForward):
"""Feed Forward to Embedding Layer"""

2
manim_ml/neural_network/layers/paired_query.py
View File

@ -1,6 +1,6 @@
from manim import *
from manim_ml.neural_network.layers.parent_layers import NeuralNetworkLayer
from manim_ml.image import GrayscaleImageMobject, LabeledColorImage
from manim_ml.utils.mobjects.image import GrayscaleImageMobject, LabeledColorImage
import numpy as np

2
manim_ml/neural_network/layers/triplet.py
View File

@ -1,6 +1,6 @@
from manim import *
from manim_ml.neural_network.layers import NeuralNetworkLayer
from manim_ml.image import GrayscaleImageMobject, LabeledColorImage
from manim_ml.utils.mobjects.image import GrayscaleImageMobject, LabeledColorImage
import numpy as np

6
manim_ml/neural_network/layers/util.py
View File

@ -4,6 +4,7 @@ from manim import *
from manim_ml.neural_network.layers.parent_layers import BlankConnective, ThreeDLayer
from manim_ml.neural_network.layers import connective_layers_list
def get_connective_layer(input_layer, output_layer):
"""
Deduces the relevant connective layer
@ -12,8 +13,9 @@ def get_connective_layer(input_layer, output_layer):
for candidate_class in connective_layers_list:
input_class = candidate_class.input_class
output_class = candidate_class.output_class
if isinstance(input_layer, input_class) and \
isinstance(output_layer, output_class):
if isinstance(input_layer, input_class) and isinstance(
output_layer, output_class
):
connective_layer_class = candidate_class
break

66
manim_ml/neural_network/neural_network.py
View File

@ -17,12 +17,13 @@ from manim import *
from manim_ml.neural_network.layers.parent_layers import ConnectiveLayer, ThreeDLayer
from manim_ml.neural_network.layers.util import get_connective_layer
from manim_ml.list_group import ListGroup
from manim_ml.neural_network.neural_network_transformations import (
from manim_ml.utils.mobjects.list_group import ListGroup
from manim_ml.neural_network.animations.neural_network_transformations import (
InsertLayer,
RemoveLayer,
)
class NeuralNetwork(Group):
"""Neural Network Visualization Container Class"""
@ -59,10 +60,7 @@ class NeuralNetwork(Group):
# Make the connective layers
self.connective_layers, self.all_layers = self._construct_connective_layers()
# Make overhead title
self.title = Text(
self.title_text,
font_size=DEFAULT_FONT_SIZE / 2
)
self.title = Text(self.title_text, font_size=DEFAULT_FONT_SIZE / 2)
self.title.next_to(self, UP, 1.0)
self.add(self.title)
# Place layers at correct z index
@ -92,11 +90,11 @@ class NeuralNetwork(Group):
raise Exception(f"Uncrecognized input layers type: {type(input_layers)}")
def add_connection(
self,
start_layer_name,
end_layer_name,
self,
start_layer_name,
end_layer_name,
connection_style="default",
connection_position="bottom"
connection_position="bottom",
):
"""Add connection from start layer to end layer"""
assert connection_style in ["default"]
@ -106,7 +104,7 @@ class NeuralNetwork(Group):
connection = NetworkConnection(
self.input_layers_dict[start_layer_name],
self.input_layers_dict[end_layer_name],
arc_direction="down" # TODO generalize this more
arc_direction="down", # TODO generalize this more
)
self.connections.append(connection)
self.add(connection)
@ -138,20 +136,20 @@ class NeuralNetwork(Group):
current_layer.move_to(previous_layer.get_center())
if layout_direction == "left_to_right":
x_shift = previous_layer.get_width() / 2 \
+ current_layer.get_width() / 2 \
+ self.layer_spacing
x_shift = (
previous_layer.get_width() / 2
+ current_layer.get_width() / 2
+ self.layer_spacing
)
shift_vector = np.array([x_shift, 0, 0])
elif layout_direction == "top_to_bottom":
y_shift = -((
previous_layer.get_width() / 2 \
+ current_layer.get_width() / 2
) + self.layer_spacing)
y_shift = -(
(previous_layer.get_width() / 2 + current_layer.get_width() / 2)
+ self.layer_spacing
)
shift_vector = np.array([0, y_shift, 0])
else:
raise Exception(
f"Unrecognized layout direction: {layout_direction}"
)
raise Exception(f"Unrecognized layout direction: {layout_direction}")
current_layer.shift(shift_vector)
# After all layers have been placed place their activation functions
@ -159,13 +157,15 @@ class NeuralNetwork(Group):
# Place activation function
if hasattr(current_layer, "activation_function"):
if not current_layer.activation_function is None:
up_movement = np.array([
0,
current_layer.get_height() / 2
+ current_layer.activation_function.get_height() / 2
+ 0.5 * self.layer_spacing,
0,
])
up_movement = np.array(
[
0,
current_layer.get_height() / 2
+ current_layer.activation_function.get_height() / 2
+ 0.5 * self.layer_spacing,
0,
]
)
current_layer.activation_function.move_to(
current_layer,
)
@ -259,9 +259,7 @@ class NeuralNetwork(Group):
current_layer_args = layer_args[layer]
# Perform the forward pass of the current layer
layer_forward_pass = layer.make_forward_pass_animation(
layer_args=current_layer_args,
run_time=per_layer_runtime,
**kwargs
layer_args=current_layer_args, run_time=per_layer_runtime, **kwargs
)
# Animate a forward pass for incoming connections
connection_input_pass = AnimationGroup()
@ -272,14 +270,12 @@ class NeuralNetwork(Group):
connection_input_pass = ShowPassingFlash(
connection,
run_time=layer_forward_pass.run_time,
time_width=0.2
time_width=0.2,
)
break
layer_forward_pass = AnimationGroup(
layer_forward_pass,
connection_input_pass,
lag_ratio=0.0
layer_forward_pass, connection_input_pass, lag_ratio=0.0
)
all_animations.append(layer_forward_pass)
# Make the animation group

11
manim_ml/one_to_one_sync.py
View File

@ -1,11 +0,0 @@
"""
Module for handling syncing two animations one to one.
The goal here is to zip up two classes and their respective animations,
and create a joint class with the same animations that runs the animations
for both classes at the same time. This way we can connect two isomorphic
views of the same concept and visualize them at the same time.
"""
class OneToOneSync:
pass

61
manim_ml/utils/mobjects/connections.py
View File

@ -1,30 +1,27 @@
import numpy as np
from manim import *
class NetworkConnection(VGroup):
"""
This class allows for creating connections
between locations in a network
This class allows for creating connections
between locations in a network
"""
direction_vector_map = {
"up": UP,
"down": DOWN,
"left": LEFT,
"right": RIGHT
}
direction_vector_map = {"up": UP, "down": DOWN, "left": LEFT, "right": RIGHT}
def __init__(
self,
start_mobject,
end_mobject,
self,
start_mobject,
end_mobject,
arc_direction="straight",
buffer=0.05,
arc_distance=0.3,
stroke_width=2.0,
color=WHITE,
active_color=ORANGE
active_color=ORANGE,
):
"""Creates an arrow with right angles in it connecting
"""Creates an arrow with right angles in it connecting
two mobjects.
Parameters
@ -72,34 +69,28 @@ class NetworkConnection(VGroup):
# Make an arrow
arrow_line = Line(
left_mobject.get_right() + np.array([self.buffer, 0.0, 0.0]),
right_mobject.get_left() + np.array([-1 * self.buffer, 0.0, 0.0])
)
arrow = Arrow(
arrow_line,
color=self.color,
stroke_width=self.stroke_width
right_mobject.get_left() + np.array([-1 * self.buffer, 0.0, 0.0]),
)
arrow = Arrow(arrow_line, color=self.color, stroke_width=self.stroke_width)
self.straight_arrow = arrow
self.add(arrow)
else:
# Figure out the direction of the arc
direction_vector = NetworkConnection.direction_vector_map[self.arc_direction]
direction_vector = NetworkConnection.direction_vector_map[
self.arc_direction
]
# Make the start arc piece
start_line_start = left_mobject.get_critical_point(
direction_vector
)
start_line_start = left_mobject.get_critical_point(direction_vector)
start_line_start += direction_vector * self.buffer
start_line_end = start_line_start + direction_vector * self.arc_distance
self.start_line = Line(
start_line_start,
start_line_end,
color=self.color,
stroke_width=self.stroke_width
stroke_width=self.stroke_width,
)
# Make the end arc piece with an arrow
end_line_end = right_mobject.get_critical_point(
direction_vector
)
end_line_end = right_mobject.get_critical_point(direction_vector)
end_line_end += direction_vector * self.buffer
end_line_start = end_line_end + direction_vector * self.arc_distance
self.end_arrow = Arrow(
@ -108,14 +99,14 @@ class NetworkConnection(VGroup):
color=WHITE,
fill_color=WHITE,
stroke_opacity=1.0,
buff=0.0
buff=0.0,
)
# Make the middle arc piece
self.middle_line = Line(
start_line_end,
end_line_start,
color=self.color,
stroke_width=self.stroke_width
stroke_width=self.stroke_width,
)
# Add the mobjects
self.add(
@ -130,23 +121,23 @@ class NetworkConnection(VGroup):
if self.arc_direction == "straight":
return ShowPassingFlash(
self.straight_arrow.copy().set_color(self.active_color),
time_width=time_width
time_width=time_width,
)
else:
# Animate the start line
start_line_animation = ShowPassingFlash(
self.start_line.copy().set_color(self.active_color),
time_width=time_width
time_width=time_width,
)
# Animate the middle line
middle_line_animation = ShowPassingFlash(
self.middle_line.copy().set_color(self.active_color),
time_width=time_width
time_width=time_width,
)
# Animate the end line
end_line_animation = ShowPassingFlash(
self.end_arrow.copy().set_color(self.active_color),
time_width=time_width
time_width=time_width,
)
return AnimationGroup(
@ -154,5 +145,5 @@ class NetworkConnection(VGroup):
middle_line_animation,
end_line_animation,
lag_ratio=1.0,
run_time=run_time
)
run_time=run_time,
)

10
manim_ml/gridded_rectangle.py → manim_ml/utils/mobjects/gridded_rectangle.py
View File

@ -14,7 +14,7 @@ class GriddedRectangle(VGroup):
close_new_points=True,
grid_xstep=None,
grid_ystep=None,
grid_stroke_width=0.0, # DEFAULT_STROKE_WIDTH/2,
grid_stroke_width=0.0, # DEFAULT_STROKE_WIDTH/2,
grid_stroke_color=ORANGE,
grid_stroke_opacity=1.0,
stroke_width=2.0,
@ -48,7 +48,7 @@ class GriddedRectangle(VGroup):
fill_color=color,
stroke_opacity=0.0,
fill_opacity=fill_opacity,
shade_in_3d=True
shade_in_3d=True,
)
self.rectangle = no_border_rectangle
border_rectangle = Rectangle(
@ -58,7 +58,7 @@ class GriddedRectangle(VGroup):
fill_color=color,
fill_opacity=fill_opacity,
shade_in_3d=True,
stroke_width=stroke_width
stroke_width=stroke_width,
)
self.dotted_lines = DashedVMobject(
border_rectangle,
@ -73,7 +73,7 @@ class GriddedRectangle(VGroup):
stroke_width=stroke_width,
fill_color=color,
fill_opacity=fill_opacity,
shade_in_3d=True
shade_in_3d=True,
)
self.add(self.rectangle)
# Make grid lines
@ -123,7 +123,7 @@ class GriddedRectangle(VGroup):
stroke_color=self.grid_stroke_color,
stroke_width=self.grid_stroke_width,
stroke_opacity=self.grid_stroke_opacity,
shade_in_3d=True
shade_in_3d=True,
)
for i in range(1, count)
)

0
manim_ml/image.py → manim_ml/utils/mobjects/image.py
View File

0
manim_ml/list_group.py → manim_ml/utils/mobjects/list_group.py
View File

0
manim_ml/probability.py → manim_ml/utils/mobjects/probability.py
View File

3
manim_ml/utils/testing/frames_comparison.py
View File

@ -27,6 +27,7 @@ _tests_root_dir_path = Path(__file__).absolute().parents[2]
print(f"Tests root path: {_tests_root_dir_path}")
PATH_CONTROL_DATA = _tests_root_dir_path / Path("control_data", "graphical_units_data")
def frames_comparison(
func=None,
*,
@ -254,4 +255,4 @@ def _config_test(last_frame: bool) -> ManimConfig:
else "config_graphical_tests_multiframes.cfg"
),
),
)
)

2
setup.py
View File

@ -2,7 +2,7 @@ from setuptools import setup, find_packages
setup(
name="manim_ml",
version="0.0.15",
version="0.0.16",
description=(" Machine Learning Animations in python using Manim."),
packages=find_packages(),
)

3
tests/conftest.py
View File

@ -1,5 +1,3 @@
from __future__ import annotations
import os
@ -9,6 +7,7 @@ import pytest
from manim import config, tempconfig
def pytest_addoption(parser):
parser.addoption(
"--skip_slow",

10
tests/test_activation_function.py
View File

@ -13,15 +13,19 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class CombinedScene(ThreeDScene):
def construct(self):
image = Image.open("../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(1, 7, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32, activation_function="ReLU"),
Convolutional2DLayer(
3, 5, 3, filter_spacing=0.32, activation_function="ReLU"
),
FeedForwardLayer(3, activation_function="Sigmoid"),
],
layer_spacing=0.25,
@ -32,4 +36,4 @@ class CombinedScene(ThreeDScene):
# Play animation
forward_pass = nn.make_forward_pass_animation()
self.wait(1)
self.play(forward_pass, run_time=30)
self.play(forward_pass, run_time=30)

14
tests/test_camera_move.py
View File

@ -12,14 +12,11 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class NeuralNetworkScene(ThreeDScene):
"""Test Scene for the Neural Network"""
def play_camera_follow_forward_pass(
self,
neural_network,
buffer=0.1
):
def play_camera_follow_forward_pass(self, neural_network, buffer=0.1):
per_layer_animations = neural_network.make_forward_pass_animation(
return_per_layer_animations=True
)
@ -32,7 +29,7 @@ class NeuralNetworkScene(ThreeDScene):
current_layer = all_layers[layer_index]
next_layer = all_layers[layer_index + 1]
group = Group(prev_layer, current_layer, next_layer)
max_width = max(max_width, group.width)
max_height = max(max_height, group.height)
@ -46,7 +43,8 @@ class NeuralNetworkScene(ThreeDScene):
# Make the Layer object
image = Image.open("../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(1, 7, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),
@ -60,4 +58,4 @@ class NeuralNetworkScene(ThreeDScene):
# Make Animation
self.add(nn)
# self.play(Create(nn))
self.play_camera_follow_forward_pass(nn)
self.play_camera_follow_forward_pass(nn)

34
tests/test_conv_padding.py
View File

@ -14,23 +14,24 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class CombinedScene(ThreeDScene):
class CombinedScene(ThreeDScene):
def construct(self):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(
num_feature_maps=1,
feature_map_size=7,
num_feature_maps=1,
feature_map_size=7,
padding=1,
padding_dashed=True
padding_dashed=True,
),
Convolutional2DLayer(
num_feature_maps=3,
feature_map_size=7,
num_feature_maps=3,
feature_map_size=7,
filter_size=3,
padding=0,
padding_dashed=False
padding_dashed=False,
),
FeedForwardLayer(3),
],
@ -44,23 +45,22 @@ class CombinedScene(ThreeDScene):
self.wait(1)
self.play(forward_pass, run_time=30)
@frames_comparison
def test_ConvPadding(scene):
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
Convolutional2DLayer(
num_feature_maps=1,
feature_map_size=7,
padding=1,
padding_dashed=True
num_feature_maps=1, feature_map_size=7, padding=1, padding_dashed=True
),
Convolutional2DLayer(
num_feature_maps=3,
feature_map_size=7,
num_feature_maps=3,
feature_map_size=7,
filter_size=3,
padding=1,
filter_spacing=0.35,
padding_dashed=False
padding_dashed=False,
),
FeedForwardLayer(3),
],
@ -71,4 +71,4 @@ def test_ConvPadding(scene):
scene.add(nn)
# Play animation
forward_pass = nn.make_forward_pass_animation()
scene.play(forward_pass, run_time=30)
scene.play(forward_pass, run_time=30)

21
tests/test_convolutional_2d_layer.py
View File

@ -6,6 +6,7 @@ from manim_ml.neural_network.layers.feed_forward import FeedForwardLayer
from manim_ml.neural_network.layers.image import ImageLayer
from manim_ml.neural_network.neural_network import NeuralNetwork
class SingleConvolutionalLayerScene(ThreeDScene):
def construct(self):
# Make nn
@ -21,6 +22,7 @@ class SingleConvolutionalLayerScene(ThreeDScene):
)
# self.play(nn.make_forward_pass_animation(run_time=5))
class Simple3DConvScene(ThreeDScene):
def construct(self):
"""
@ -33,16 +35,8 @@ class Simple3DConvScene(ThreeDScene):
"""
# Make nn
layers = [
Convolutional2DLayer(
num_feature_maps=1,
feature_map_size=3,
filter_size=3
),
Convolutional2DLayer(
num_feature_maps=1,
feature_map_size=3,
filter_size=3
),
Convolutional2DLayer(num_feature_maps=1, feature_map_size=3, filter_size=3),
Convolutional2DLayer(num_feature_maps=1, feature_map_size=3, filter_size=3),
]
nn = NeuralNetwork(layers)
# Center the nn
@ -52,18 +46,21 @@ class Simple3DConvScene(ThreeDScene):
# self.set_camera_orientation(phi=280*DEGREES, theta=-10*DEGREES, gamma=90*DEGREES)
self.play(nn.make_forward_pass_animation(run_time=30))
# Make the specific scene
config.pixel_height = 1200
config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class CombinedScene(ThreeDScene):
def construct(self):
image = Image.open("../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(1, 7, filter_spacing=0.32),
Convolutional2DLayer(3, 5, 3, filter_spacing=0.32),
@ -79,4 +76,4 @@ class CombinedScene(ThreeDScene):
# Play animation
forward_pass = nn.make_forward_pass_animation()
self.wait(1)
self.play(forward_pass)
self.play(forward_pass)

4
tests/test_decision_tree.py
View File

@ -9,6 +9,7 @@ from sklearn import datasets
import sklearn
import matplotlib.pyplot as plt
def learn_iris_decision_tree(iris):
decision_tree = DecisionTreeClassifier(
random_state=1, max_depth=3, max_leaf_nodes=6
@ -17,11 +18,13 @@ def learn_iris_decision_tree(iris):
# output the decisioin tree in some format
return decision_tree
def make_sklearn_tree(dataset, max_tree_depth=3):
tree = learn_iris_decision_tree(dataset)
feature_names = dataset.feature_names[0:2]
return tree, tree.tree_
class DecisionTreeScene(Scene):
def construct(self):
"""Makes a decision tree object"""
@ -44,6 +47,7 @@ class DecisionTreeScene(Scene):
self.play(create_decision_tree)
# self.play(create_decision_tree)
class SurfacePlot(Scene):
def construct(self):
iris_dataset = datasets.load_iris()

19
tests/test_feed_forward.py
View File

@ -5,26 +5,21 @@ from manim_ml.neural_network import NeuralNetwork, FeedForwardLayer
__module_test__ = "feed_forward"
@frames_comparison
def test_FeedForwardScene(scene):
"""Tests the appearance of a feed forward network"""
nn = NeuralNetwork([
FeedForwardLayer(3),
FeedForwardLayer(5),
FeedForwardLayer(3)
])
nn = NeuralNetwork([FeedForwardLayer(3), FeedForwardLayer(5), FeedForwardLayer(3)])
scene.add(nn)
class FeedForwardScene(Scene):
class FeedForwardScene(Scene):
def construct(self):
nn = NeuralNetwork([
FeedForwardLayer(3),
FeedForwardLayer(5),
FeedForwardLayer(3)
])
nn = NeuralNetwork(
[FeedForwardLayer(3), FeedForwardLayer(5), FeedForwardLayer(3)]
)
self.add(nn)
self.play(nn.make_forward_pass_animation())
self.play(nn.make_forward_pass_animation())

12
tests/test_gridded_rectangle.py
View File

@ -1,12 +1,8 @@
from manim import *
from manim_ml.gridded_rectangle import GriddedRectangle
from manim_ml.utils.mobjects.gridded_rectangle import GriddedRectangle
class TestGriddedRectangleScene(ThreeDScene):
def construct(self):
rect = GriddedRectangle(
color=ORANGE,
width=3,
height=3
)
self.add(rect)
rect = GriddedRectangle(color=ORANGE, width=3, height=3)
self.add(rect)

2
tests/test_image_homotopy.py
View File

@ -1,7 +1,7 @@
from PIL import Image
from manim import *
from manim_ml.image import GrayscaleImageMobject
from manim_ml.utils.mobjects.image import GrayscaleImageMobject
from manim_ml.neural_network.layers.parent_layers import ThreeDLayer

10
tests/test_max_pool.py
View File

@ -14,12 +14,14 @@ config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class CombinedScene(ThreeDScene):
def construct(self):
image = Image.open("../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(1, 8, filter_spacing=0.32),
Convolutional2DLayer(3, 6, 3, filter_spacing=0.32),
@ -37,12 +39,14 @@ class CombinedScene(ThreeDScene):
self.wait(1)
self.play(forward_pass)
class SmallNetwork(ThreeDScene):
def construct(self):
image = Image.open("../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
# Make nn
nn = NeuralNetwork([
nn = NeuralNetwork(
[
ImageLayer(numpy_image, height=1.5),
Convolutional2DLayer(1, 8, filter_spacing=0.32),
MaxPooling2DLayer(kernel_size=2),
@ -55,4 +59,4 @@ class SmallNetwork(ThreeDScene):
# Play animation
forward_pass = nn.make_forward_pass_animation()
self.wait(1)
self.play(forward_pass)
self.play(forward_pass)

20
tests/test_mcmc.py
View File

@ -1,31 +1,33 @@
from manim import *
from manim_ml.diffusion.mcmc import MCMCAxes, MultidimensionalGaussianPosterior, metropolis_hastings_sampler
from manim_ml.diffusion.mcmc import (
MCMCAxes,
MultidimensionalGaussianPosterior,
metropolis_hastings_sampler,
)
# Make the specific scene
config.pixel_height = 1200
config.pixel_width = 1200
config.frame_height = 12.0
config.frame_width = 12.0
def test_metropolis_hastings_sampler(iterations=100):
samples, _, candidates = metropolis_hastings_sampler(iterations=iterations)
assert samples.shape == (iterations, 2)
class MCMCTest(Scene):
class MCMCTest(Scene):
def construct(self):
axes = MCMCAxes()
self.play(Create(axes))
gaussian_posterior = MultidimensionalGaussianPosterior(
mu=np.array([0.0, 0.0]),
var=np.array([4.0, 2.0])
)
show_gaussian_animation = axes.show_ground_truth_gaussian(
gaussian_posterior
mu=np.array([0.0, 0.0]), var=np.array([4.0, 2.0])
)
show_gaussian_animation = axes.show_ground_truth_gaussian(gaussian_posterior)
self.play(show_gaussian_animation)
chain_sampling_animation = axes.visualize_metropolis_hastings_chain_sampling(
log_prob_fn=gaussian_posterior,
sampling_kwargs={"iterations": 1000}
log_prob_fn=gaussian_posterior, sampling_kwargs={"iterations": 1000}
)
self.play(chain_sampling_animation)

2
tests/test_nested_neural_networks.py
View File

@ -1,4 +1,4 @@
"""
The purpose of this test is to ensure that it is possible
to have nested neural network layers.
"""
"""

45
tests/test_residual_connection.py
View File

@ -9,49 +9,53 @@ import numpy as np
__module_test__ = "residual"
@frames_comparison
def test_ResidualConnectionScene(scene):
"""Tests the appearance of a residual connection"""
nn = NeuralNetwork({
"layer1": FeedForwardLayer(3),
"layer2": FeedForwardLayer(5),
"layer3": FeedForwardLayer(3)
})
nn = NeuralNetwork(
{
"layer1": FeedForwardLayer(3),
"layer2": FeedForwardLayer(5),
"layer3": FeedForwardLayer(3),
}
)
scene.add(nn)
# Make the specific scene
config.pixel_height = 1200
config.pixel_width = 1900
config.frame_height = 6.0
config.frame_width = 6.0
class FeedForwardScene(Scene):
class FeedForwardScene(Scene):
def construct(self):
nn = NeuralNetwork({
"layer1": FeedForwardLayer(4),
"layer2": FeedForwardLayer(4),
"layer3": FeedForwardLayer(4)
},
layer_spacing=0.45)
nn = NeuralNetwork(
{
"layer1": FeedForwardLayer(4),
"layer2": FeedForwardLayer(4),
"layer3": FeedForwardLayer(4),
},
layer_spacing=0.45,
)
nn.add_connection("layer1", "layer3")
self.add(nn)
self.play(
nn.make_forward_pass_animation(),
run_time=8
)
self.play(nn.make_forward_pass_animation(), run_time=8)
class ConvScene(ThreeDScene):
def construct(self):
image = Image.open("../assets/mnist/digit.jpeg")
numpy_image = np.asarray(image)
nn = NeuralNetwork({
nn = NeuralNetwork(
{
"layer1": Convolutional2DLayer(1, 5, padding=1),
"layer2": Convolutional2DLayer(1, 5, 3, padding=1),
"layer3": Convolutional2DLayer(1, 5, 3, padding=1),
@ -63,7 +67,4 @@ class ConvScene(ThreeDScene):
self.add(nn)
self.play(
nn.make_forward_pass_animation(),
run_time=8
)
self.play(nn.make_forward_pass_animation(), run_time=8)

8
tests/test_show_gaussian.py
View File

@ -1,14 +1,12 @@
from manim import *
from manim_ml.probability import GaussianDistribution
from manim_ml.utils.mobjects.probability import GaussianDistribution
class TestShowGaussian(Scene):
def construct(self):
axes = Axes()
self.add(axes)
gaussian = GaussianDistribution(
axes,
mean=np.array([0.0, 0.0]),
cov=np.array([[2.0, 0.0], [0.0, 1.0]])
axes, mean=np.array([0.0, 0.0]), cov=np.array([[2.0, 0.0], [0.0, 1.0]])
)
self.add(gaussian)

4
tests/test_succession.py
View File

@ -1,7 +1,7 @@
from manim import *
class TestSuccession(Scene):
class TestSuccession(Scene):
def construct(self):
white_dot = Dot(color=WHITE)
white_dot.shift(UP)
@ -16,4 +16,4 @@ class TestSuccession(Scene):
Wait(1),
Uncreate(red_dot),
)
)
)