def main():
version = 2
skipSize = 0
includeSize = 400000
dimensions = [25, 50, 100, 150, 200, 300]
name = "Top{}K-Skip{}".format(round(includeSize / 1000), skipSize)
gloveOutputFolder = DatasetManager.getVisualEmbeddingsFolderPath(version) + name
'''
PerformPCA.performPCA(name, gloveOutputFolder, dimensions, includeSize, skipSize, version)
# Convert into KeyedVectors
for d in dimensions:
folderPath = gloveOutputFolder + "-{}/".format(d)
glovePath = folderPath + "VisualGlove-{}-{}.txt".format(name, d)
keyPath = folderPath + "Keyed-VisualGlove-{}".format(d)
GloveFormatter.createKeyedVectorsFromGloveFile(glovePath, keyPath)
'''
# Concat with standard Glove
sizes = [(100, 300), (50, 300)]
ConcatKeyedVectors.concatToStandardGlove(gloveOutputFolder, name, sizeCombinations=sizes, version=version)
def getTopAndSkipKVisualOnly(top, skip, version=1):
basePath = DatasetManager.getVisualEmbeddingsFolderPath(version)
return [basePath + "Top{}K-Skip{}-{}/Keyed-VisualGlove-{}".format(top, skip, i, i) for i in [50, 100, 200, 300]]
def main():
glovePath = DatasetManager.getVisualEmbeddingsFolderPath(
2) + "/VisualGlove-2.0 Full.txt"
GloveFormatter.createKeyedVectorsFromGloveFile(
glovePath, "Keyed-VisualGlove-2.0-Full")
'''
def getTop100KVisualOnly():
basePath = DatasetManager.getVisualEmbeddingsFolderPath()
return [basePath + "Top-100K-{}/Keyed-VisualGlove-{}".format(i, i) for i in [50, 100, 200, 300]]