def __init__(self, W, corpus, power=0.75, sample_size=5):
self.sample_size = sample_size
self.sampler = UnigramSampler(corpus, power, sample_size)
self.loss_layers = [SigmoidWithLoss() for _ in range(sample_size + 1)]
self.embed_dot_layers = [
EmbeddingDot(W) for _ in range(sample_size + 1)
]
self.params, self.grads = [], []
for layer in self.embed_dot_layers:
self.params += layer.params
self.grads += layer.grads
class NegativeSamplingLoss:
def __init__(self, W, corpus, power=0.75, sample_size=5):
self.sample_size = sample_size
self.sampler = UnigramSampler(corpus, power, sample_size)
self.loss_layers = [SigmoidWithLoss() for _ in range(sample_size + 1)]
self.embed_dot_layers = [
EmbeddingDot(W) for _ in range(sample_size + 1)
]
self.params, self.grads = [], []
for layer in self.embed_dot_layers:
self.params += layer.params
self.grads += layer.grads
def forward(self, h, target):
batch_size = target.shape[0]
negative_sample = self.sampler.get_negative_sample(target)
score = self.embed_dot_layers[0].forward(h, target)
correct_label = np.ones(batch_size, dtype=np.int32)
loss = self.loss_layers[1 + i].forward(score, negative_label)
return loss
def backward(self, dout=1):
dh = 0
for l0, l1 in zip(self.loss_layers, self.embed_dot_layers):
dscore = l0.backward(dout)
dh += l1.backward(dscore)
return dh
print(np.random.choice(words, size=5, replace=False))
p = [0.5, 0.1, 0.05, 0.2, 0.05, 0.1]
print(np.random.choice(words, p=p))
p = [0.7, 0.29, 0.01]
new_p = np.power(p, 0.75)
new_p /= np.sum(new_p)
print(new_p)
import sys
sys.path.append('..')
from ch04.negative_sampling_layer import UnigramSampler
corpus = np.array([0, 1, 2, 3, 4, 1, 2, 3])
power = 0.75
sample_size = 2
sampler = UnigramSampler(corpus, power, sample_size)
target = np.array([1, 3, 0])
negative_sample = sampler.get_negative_sample(target)
print(negative_sample)
class NegativeSamplingLoss:
def __init__(self, W, corpus, power=0.75, sample_size=5):
self.sample_size = sample_size
self.sampler = UnigramSampler(corpus, power, sample_size)
self.loss_layers = [SigmoidWithLoss() for _ in range(sample_size + 1)]
self.embed_dot_layers = [
EmbeddingDot(W) for _ in range(sample_size + 1)
]
self.params, self.grads = [], []