Functions — github.com/ExplorerFreda/VGNSL

CodeHub MCPcopy Create free account

hub / github.com/ExplorerFreda/VGNSL / functions

Functions81 in github.com/ExplorerFreda/VGNSL

⨍ Functions 81 ◇ Types & classes 12

↓ 9 callersFunctionindex_one_hot_ellipsis

(tensor, dim, index)

src/utils.py:139

↓ 9 callersMethodupdate

(self, val, n=0)

src/evaluation.py:28

↓ 8 callersFunctionl2norm

src/utils.py:109

↓ 5 callersFunctionsequence_mask

(sequence_length, max_length=None)

src/utils.py:126

↓ 4 callersMethod__init__

src/model.py:247

↓ 4 callersMethodfromstring

analysis/constituent_recall.py:9

↓ 4 callersMethodupdate_with_mask

(lv, rv, cv, lm, rm, cm)

src/model.py:171

↓ 3 callersFunctionconcat_shape

src/utils.py:82

↓ 3 callersMethodforward_emb

Compute the image and caption embeddings

src/model.py:294

↓ 3 callersFunctionget_precomp_loader

(data_path, data_split, vocab, batch_size=128, shuffle=True, num_workers=2, load_img=Tr

↓ 2 callersFunctionadd_dim

(tensor, dim, size)

src/utils.py:100

↓ 2 callersFunctionadd_dim_as_except

(tensor, target, *excepts)

src/utils.py:192

↓ 2 callersFunctionextract_spans

↓ 2 callersFunctiongenerate_tree

(captions, tree_indices, pos, vocab, pad_word='<pad>')

src/utils.py:113

↓ 2 callersMethodload_state_dict

(self, state_dict)

src/model.py:278

↓ 2 callersFunctionprod

(values, default=1)

src/utils.py:210

↓ 2 callersFunctionspans

analysis/constituent_recall.py:34

↓ 2 callersMethodstate_dict

src/model.py:274

↓ 2 callersMethodtrain_start

switch to train mode

src/model.py:284

↓ 2 callersFunctionvalidate

(opt, val_loader, model, vocab)

src/train.py:59

↓ 1 callersFunctionadjust_learning_rate

Sets the learning rate to the initial LR decayed by 10 every 30 epochs

src/train.py:85

↓ 1 callersFunctionbroadcast

(tensor, dim, size)

src/utils.py:92

↓ 1 callersFunctionclean_tree

(sentence, remove_tag_set={'<start>', '<end>', '<pad>'})

src/utils.py:216

↓ 1 callersFunctionconstituent_recall

(gold_captions, pred_trees, gold_trees=None)

analysis/constituent_recall.py:47

↓ 1 callersFunctionencode_data

Encode all images and captions loadable by `data_loader`

src/evaluation.py:68

↓ 1 callersFunctionextract_statistics

(gold_tree_spans, produced_tree_spans)

↓ 1 callersFunctionf1_score

(produced_trees, gold_trees)

↓ 1 callersMethodforward_reward

Compute the loss given pairs of image and caption embeddings

src/model.py:306

↓ 1 callersFunctionget_eval_loader

(data_path, split_name, vocab, batch_size, workers, load_img=False, img_dim=2048)

↓ 1 callersFunctionget_simplified_tree

Simplify a tree by removing all non-terminal labels and punctuation marks. Args: tree: An NLTK tree. Returns: A string

preprocess/parse_simplifier.py:6

↓ 1 callersFunctioni2t

Images->Text (Image Annotation) Images: (5N, K) matrix of images Captions: (5N, K) matrix of captions

src/evaluation.py:129

↓ 1 callersFunctionindex_mask

(indices, max_length)

src/utils.py:149

↓ 1 callersMethodinit_weights

Xavier initialization for the fully connected layer

src/model.py:32

↓ 1 callersFunctionlength2mask

(lengths, max_length)

src/utils.py:202

↓ 1 callersFunctionmake_embeddings

(opt, vocab_size, dim)

src/utils.py:52

↓ 1 callersFunctionremove_label

data/add_c_parse.py:23

↓ 1 callersMethodreset

src/evaluation.py:22

↓ 1 callersFunctionsave_checkpoint

(state, is_best, curr_epoch, filename='checkpoint.pth.tar', prefix='')

src/train.py:78

↓ 1 callersFunctiont2i

Text->Images (Image Search) Images: (5N, K) matrix of images Captions: (5N, K) matrix of captions

src/evaluation.py:173

↓ 1 callersFunctiontest_trees

use the trained model to generate parse trees for text

src/evaluation.py:211

↓ 1 callersFunctiontrain

(opt, train_loader, model, epoch, val_loader, vocab)

src/train.py:16

↓ 1 callersMethodtrain_emb

one training step given images and captions

src/model.py:352

↓ 1 callersMethodupdate

analysis/constituent_recall.py:19

↓ 1 callersMethodval_start

switch to evaluate mode

src/model.py:289

src/vocab.py:20

Method__getitem__

analysis/constituent_recall.py:15

src/evaluation.py:19

src/evaluation.py:47

(self, *embeddings)

(self, data_path, data_split, vocab, load_img=True, img_dim=2048)

(self, img_dim, embed_size, no_imgnorm=False)

src/model.py:23

(self, opt, vocab_size, semantics_dim)

src/model.py:64

(self, margin=0)

src/model.py:183

(self, margin=0)

src/model.py:219

src/vocab.py:25

analysis/constituent_recall.py:30

analysis/constituent_recall.py:27

String representation for logging

src/evaluation.py:34

Concatenate the meters in one log line

src/evaluation.py:57

Functionaccuracy

Computes the precision@k for the specified values of k

src/train.py:93

src/vocab.py:14

Functioncollate_fn

build mini-batch tensors from a list of (image, caption) tuples

Functioncosine_sim

Cosine similarity between all the image and sentence pairs

src/utils.py:104

src/utils.py:13

extract image feature vectors

src/model.py:39

sample a tree for each sentence

src/model.py:82

return the reward

src/model.py:188

src/model.py:224

Functionget_train_loaders

(data_path, vocab, batch_size, workers)

Functionindex_range_ellipsis

(x, a, b, dim=1, padding_zero=True)

src/utils.py:160

Methodload_state_dict

copies parameters, overwritting the default one to accept state_dict from Full model

src/model.py:50

analysis/constituent_recall.py:24

Functionreplace_leaves

data/add_c_parse.py:10

Methodreset_weights

src/model.py:79

Functiontree2list

src/utils.py:17

Functiontree2str

src/utils.py:45

Functiontreelist2dict

src/utils.py:37

(self, k, v, n=0)

src/evaluation.py:51