def downsampling():
crs = load_and_sort(config.downsamplingconfigs)
for c, r in crs:
print("Downsampling used:", c.embedding.downsampling)
print("Error reduction", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def trainsizes():
crs = load_and_sort(config.trainsizeconfigs)
for c, r in crs:
print("Train size:", c.run.train_size)
print("Error reduction:", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def negative():
crs = load_and_sort(config.negativeconfigs)
for c, r in crs:
print("Negative sampling count:", c.embedding.negative)
print("Error reduction", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def estimator():
crs = load_and_sort(config.estimatorconfigs)
for c, r in crs:
print("Estimator for word embedding:", c.embedding.estimator)
print("Error reduction", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def filter_skips():
crs = load_and_sort(config.filter_skipsconfigs)
for c, r in crs:
print("Skips filtered:", c.gram.filter_skips)
print("Error reduction:", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def dimension():
crs = load_and_sort(config.dimensionconfigs)
for c, r in crs:
print("Dimension:", c.embedding.dimension)
print("Error reduction:", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def window():
crs = load_and_sort(config.windowconfigs)
for c, r in crs:
print("Window size:", c.embedding.window)
print("Gram size:", c.gram.gram_size)
print("Error reduction", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def skipwords():
crs = load_and_sort(config.skipwordconfigs)
skipwordss = {c.gram.skipwords for (c, _) in crs}
for skipwords in skipwordss:
(c, r) = get_single_one(((c, r) for (c, r) in crs if c.gram.skipwords == skipwords))
print("Skipwords:", *skipwords)
print("Error reduction:", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def default():
print("File contents:")
with gzip.open(resultpath(config.default_config), mode="rt") as f:
for line in f:
print(line, end="")
print_barline()
r = load_result(config.default_config).result
print("Error reduction:", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
def interesting_embeddings2():
crs = load_and_sort(config.interesting_embeddingconfigs2)
for c, r in crs:
print("Estimator:", c.embedding.estimator)
print("Downsampling:", c.embedding.downsampling)
print("Min count:", c.embedding.min_count)
print("Error reduction", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
def min_count():
crs = load_and_sort(config.min_countconfigs)
for filter_unknown in [True, False]:
print("Grams with unknown words filtered:", filter_unknown)
print_barline()
relevant_crs = [(c, r) for (c, r) in crs if c.run.filter_unknown == filter_unknown]
for c, r in relevant_crs:
print("Min count for word embedding:", c.embedding.min_count)
print("Error reduction", error_reduction(r))
print("Prior correctness:", prior_correctness(r))
print("Correctness:", correctness(r))
print_barline()
print_barline()
def key(cr):
if is_failed(cr[1]):
return -float("inf")
else:
return error_reduction(cr[1])