parser.add_argument("--multimodal_fusion_hidden_size",
type=int,
default=512)
parser.add_argument("--classification_hidden_size", type=int, default=512)
parser.add_argument("--batch_size", type=int, default=256)
parser.add_argument("--num_epochs", type=int, default=100)
parser.add_argument("--learning_rate", type=float, default=0.001)
parser.add_argument("--l2_reg", type=float, default=0.000005)
parser.add_argument("--patience", type=int, default=3)
args = parser.parse_args()
start_logger(args.model_save_filename + ".train_log")
atexit.register(stop_logger)
print("-- Building vocabulary")
embeddings, token2id, id2token = load_glove(args.vectors_filename,
args.max_vocab,
args.embeddings_size)
label2id = {"neutral": 0, "entailment": 1, "contradiction": 2}
id2label = {v: k for k, v in label2id.items()}
num_tokens = len(token2id)
num_labels = len(label2id)
print("Number of tokens: {}".format(num_tokens))
print("Number of labels: {}".format(num_labels))
with open(args.model_save_filename + ".params", mode="w") as out_file:
json.dump(vars(args), out_file)
print("Params saved to: {}".format(args.model_save_filename +
".params"))
with open(args.model_save_filename + ".index", mode="wb") as out_file:
pickle.dump(
def load_e_vsnli_dataset_and_glove(nli_dataset_filename,
label2id,
vectors_filename,
max_vocab,
embeddings_size,
buffer_size=None,
padding_length=None,
min_threshold=0,
keep_neutrals=True):
labels = []
padded_explanations = []
padded_premises = []
padded_hypotheses = []
image_names = []
pairIDs = []
original_explanations = []
original_premises = []
original_hypotheses = []
all_premise_tokens = []
all_hypothesis_tokens = []
all_explanation_tokens = []
with open(nli_dataset_filename) as in_file:
reader = csv.reader(in_file, delimiter="\t")
next(reader, None) #skip header
for i, row in enumerate(reader):
if buffer_size and i >= buffer_size:
break
label = row[0].strip()
if keep_neutrals == False and label == 'neutral':
continue
premise_tokens = row[1].strip().split()
hypothesis_tokens = row[2].strip().split()
image = row[3].strip().split("#")[0]
premise = row[4].strip()
hypothesis = row[5].strip()
pairID = row[6]
explanation = row[7].strip()
explanation_tokens = row[8].strip().split()
labels.append(label2id[label])
#TODO: add <start> and </end>
explanation_tokens = ['<start>'] + explanation_tokens + ['<end>']
hypothesis_tokens = ['<start>'] + hypothesis_tokens + ['<end>']
all_premise_tokens.append(premise_tokens)
all_hypothesis_tokens.append(hypothesis_tokens)
all_explanation_tokens.append(explanation_tokens)
image_names.append(image)
pairIDs.append(pairID)
original_premises.append(premise)
original_hypotheses.append(hypothesis)
original_explanations.append(explanation)
labels = np.array(labels)
pairIDs = np.array(pairIDs)
if min_threshold:
word_freq = Counter(x for xs in all_explanation_tokens
for x in set(xs))
word_freq = {
x: word_freq[x]
for x in word_freq if word_freq[x] >= min_threshold
}
else:
word_freq = None
embeddings, token2id, id2token = load_glove(vectors_filename,
max_vocab, embeddings_size,
word_freq)
padded_premises = [[
token2id.get(token, token2id["#unk#"]) for token in premise_tokens
] for premise_tokens in all_premise_tokens]
padded_hypotheses = [[
token2id.get(token, token2id["#unk#"])
for token in hypothesis_tokens
] for hypothesis_tokens in all_hypothesis_tokens]
padded_explanations = [[
token2id.get(token, token2id["#unk#"])
for token in explanation_tokens
] for explanation_tokens in all_explanation_tokens]
max_length = max(len(pad_expl) for pad_expl in padded_explanations)
if padding_length is None:
padding_length = max_length
padded_premises = pad_sequences(padded_premises,
maxlen=padding_length,
padding="post",
value=token2id["#pad#"],
dtype=np.long)
padded_hypotheses = pad_sequences(padded_hypotheses,
maxlen=padding_length,
padding="post",
value=token2id["#pad#"],
dtype=np.long)
padded_explanations = pad_sequences(padded_explanations,
maxlen=padding_length,
padding="post",
value=token2id["#pad#"],
dtype=np.long)
return labels, padded_explanations, padded_premises, padded_hypotheses, image_names, original_explanations, original_premises, original_hypotheses, max_length, embeddings, token2id, id2token, pairIDs
