记录一下课程学习的心得和分享一下作业,推荐还是自己先写一遍
膜拜一下大佬Tianqi Chen
附上课程主页:https://dlsyscourse.org/lectures/
def add(x, y):
""" A trivial 'add' function you should implement to get used to the
autograder and submission system. The solution to this problem is in the
the homework notebook.
Args:
x (Python number or numpy array)
y (Python number or numpy array)
Return:
Sum of x + y
"""
### BEGIN YOUR CODE
return x + y
### END YOUR CODE
def parse_mnist(image_filename, label_filename):
""" Read an images and labels file in MNIST format. See this page:
http://yann.lecun.com/exdb/mnist/ for a description of the file format.
Args:
image_filename (str): name of gzipped images file in MNIST format
label_filename (str): name of gzipped labels file in MNIST format
Returns:
Tuple (X,y):
X (numpy.ndarray[np.float32]): 2D numpy array containing the loaded
data. The dimensionality of the data should be
(num_examples x input_dim) where 'input_dim' is the full
dimension of the data, e.g., since MNIST images are 28x28, it
will be 784. Values should be of type np.float32, and the data
should be normalized to have a minimum value of 0.0 and a
maximum value of 1.0. The normalization should be applied uniformly
across the whole dataset, _not_ individual images.
y (numpy.ndarray[dtype=np.uint8]): 1D numpy array containing the
labels of the examples. Values should be of type np.uint8 and
for MNIST will contain the values 0-9.
"""
### BEGIN YOUR CODE
with gzip.open(label_filename, 'rb') as f_in:
with open(label_filename.split('.gz')[0], 'wb') as f_out:
shutil.copyfileobj(f_in, f_out)
with open(label_filename.split('.gz')[0], 'rb') as lbpath:
magic, n = struct.unpack('>II',lbpath.read(8))
labels = np.fromfile(lbpath,dtype=np.uint8)
with gzip.open(image_filename, 'rb') as f_in:
with open(image_filename.split('.gz')[0], 'wb') as f_out:
shutil.copyfileobj(f_in, f_out)
with open(image_filename.split('.gz')[0], 'rb') as imgpath:
magic, num, rows, cols = struct.unpack('>IIII',imgpath.read(16))
images = np.fromfile(imgpath,dtype=np.uint8).reshape(len(labels), 784)
images = images.astype(np.float32)
min_val, max_val =np.min(images), np.max(images)
images = (images - min_val) / (max_val -min_val)
return images,labels
### END YOUR CODE
def softmax_loss(Z, y):
""" Return softmax loss. Note that for the purposes of this assignment,
you don't need to worry about "nicely" scaling the numerical properties
of the log-sum-exp computation, but can just compute this directly.
Args:
Z (np.ndarray[np.float32]): 2D numpy array of shape
(batch_size, num_classes), containing the logit predictions for
each class.
y (np.ndarray[np.int8]): 1D numpy array of shape (batch_size, )
containing the true label of each example.
Returns:
Average softmax loss over the sample.
"""
### BEGIN YOUR CODE
b = np.zeros((Z.shape[0], Z.shape[1]))
b[np.arange(Z.shape[0]), y] = 1
ans = np.exp(Z) * b / (np.exp(Z).sum(axis=1)[:,None])
return -np.log(ans.sum(axis=1)).mean()
### END YOUR CODE
def softmax_regression_epoch(X, y, theta, lr = 0.1, batch=100):
""" Run a single epoch of SGD for softmax regression on the data, using
the step size lr and specified batch size. This function should modify the
theta matrix in place, and you should iterate through batches in X _without_
randomizing the order.
Args:
X (np.ndarray[np.float32]): 2D input array of size
(num_examples x input_dim).
y (np.ndarray[np.uint8]): 1D class label array of size (num_examples,)
theta (np.ndarrray[np.float32]): 2D array of softmax regression
parameters, of shape (input_dim, num_classes)
lr (float): step size (learning rate) for SGD
batch (int): size of SGD minibatch
Returns:
None
"""
### BEGIN YOUR CODE
input_dim, num_classes = X.shape[1], theta.shape[1]
#theta = np.zeros((input_dim,num_classes), dtype=np.float32)
for i in range(0, y.shape[0], batch):
X_batch, y_batch = X[i: i+batch],y[i:i+batch]
pred = np.matmul(X_batch, theta)
norm_z =np.exp(pred) / (np.exp(pred).sum(axis=1)[:,None])
one_hot = np.zeros((y_batch.shape[0], num_classes),dtype=np.float32)
one_hot[np.arange(y_batch.shape[0]), y_batch] = 1.
dpred = (-one_hot + norm_z)/batch
dtheta = np.matmul(X_batch.T, dpred)
theta -= lr * dtheta
### END YOUR CODE
def nn_epoch(X, y, W1, W2, lr = 0.1, batch=100):
""" Run a single epoch of SGD for a two-layer neural network defined by the
weights W1 and W2 (with no bias terms):
logits = ReLU(X * W1) * W2
The function should use the step size lr, and the specified batch size (and
again, without randomizing the order of X). It should modify the
W1 and W2 matrices in place.
Args:
X (np.ndarray[np.float32]): 2D input array of size
(num_examples x input_dim).
y (np.ndarray[np.uint8]): 1D class label array of size (num_examples,)
W1 (np.ndarray[np.float32]): 2D array of first layer weights, of shape
(input_dim, hidden_dim)
W2 (np.ndarray[np.float32]): 2D array of second layer weights, of shape
(hidden_dim, num_classes)
lr (float): step size (learning rate) for SGD
batch (int): size of SGD minibatch
Returns:
None
"""
### BEGIN YOUR CODE
input_dim, num_classes = X.shape[1], W2.shape[1]
for i in range(0, y.shape[0], batch):
X_batch, y_batch = X[i: i+batch],y[i:i+batch]
Z1 = np.matmul(X_batch, W1)
layer1_mask = Z1 > 0
Z1[~layer1_mask]=0
pred = np.matmul(Z1, W2)
norm_z =np.exp(pred) / (np.exp(pred).sum(axis=1)[:,None])
one_hot = np.zeros((y_batch.shape[0], num_classes),dtype=np.float32)
one_hot[np.arange(y_batch.shape[0]), y_batch] = 1.
G2 = (-one_hot + norm_z)
G1 = layer1_mask * (np.matmul(G2, W2.T))
W2 -= lr * np.matmul(Z1.T,G2)/batch
W1 -= lr * np.matmul(X_batch.T,G1)/batch
### END YOUR CODE
Q6
void softmax_regression_epoch_cpp(const float *X, const unsigned char *y,
float *theta, size_t m, size_t n, size_t k,
float lr, size_t batch)
{
/**
* A C++ version of the softmax regression epoch code. This should run a
* single epoch over the data defined by X and y (and sizes m,n,k), and
* modify theta in place. Your function will probably want to allocate
* (and then delete) some helper arrays to store the logits and gradients.
*
* Args:
* X (const float *): pointer to X data, of size m*n, stored in row
* major (C) format
* y (const unsigned char *): pointer to y data, of size m
* theta (float *): pointer to theta data, of size n*k, stored in row
* major (C) format
* m (size_t): number of examples
* n (size_t): input dimension
* k (size_t): number of classes
* lr (float): learning rate / SGD step size
* batch (int): SGD minibatch size
*
* Returns:
* (None)
*/
/// BEGIN YOUR CODE
float pred[batch][k], norm_z[batch][k], dpred[batch][k], dtheta[n][k];
for(int b_start = 0; b_start < m; b_start += batch)
{
//std::cout<<b_start<<std::endl;
//cal pred
float total = 0;
for(int row = 0; row < batch; row++)
{
for(int col=0; col < k; col++)
{
pred[row][col] = 0;
for(int step=0; step < n; step++)
pred[row][col] += X[b_start*n + row*n +step] * theta[step * k + col];
total += pred[row][col];
}
}
for(int row=0; row < batch;row++)
{
float sum = 0;
for(int col=0; col <k;col++){
norm_z[row][col] = exp(pred[row][col]);
sum+=norm_z[row][col];
}
for(int col=0; col <k;col++){
norm_z[row][col] /= sum;
}
}
for(int row=0; row < batch; row++)
{
for(int col=0; col < k; col++)
dpred[row][col] = norm_z[row][col] / batch;
}
for(int row=0; row < batch; row++)
{
char label = y[b_start + row];
dpred[row][label] = -1.0/(batch * 1.0) + dpred[row][label];
}
//cal dtheta
for(int row=0; row < n; row ++){
for(int col = 0; col < k; col++){
dtheta[row][col] = 0;
for(int step = 0; step < batch; step++){
dtheta[row][col] += X[b_start * n + row+ step * n] * dpred[step][col];
}
}
}
for(int row=0; row < n; row++)
{
for(int col=0; col <k;col++)
theta[row*k+col]-= lr * dtheta[row][col];
}
}
/// END YOUR CODE
}