MCPcopy Create free account

hub / github.com/HA-Transformer/MAT / functions

Functions2,038 in github.com/HA-Transformer/MAT

⨍ Functions 2,038◇ Types & classes 299
↓ 2 callersMethodcheck_index
(self, i)
fairseq/data/indexed_dataset.py:140
↓ 2 callersMethodcheck_index
(self, i)
fairseq/data/indexed_dataset.py:255
↓ 2 callersFunctioncheckpoint_paths
Retrieves all checkpoints found in `path` directory. Checkpoints are identified by matching filename to the specified pattern. If the pattern
fairseq/checkpoint_utils.py:179
↓ 2 callersFunctionclean
(text)
fairseq/models/roberta/alignment_utils.py:26
↓ 2 callersMethodcls
Helper to get index of cls symbol
fairseq/data/legacy/masked_lm_dictionary.py:52
↓ 2 callersFunctioncode
(dtype)
fairseq/data/indexed_dataset.py:95
↓ 2 callersFunctioncollate
( samples, pad_idx, eos_idx, left_pad_source=True, left_pad_target=False, input_feeding=True, )
fairseq/data/language_pair_dataset.py:12
↓ 2 callersMethodcollate
utility function to collate samples into batch for speech recognition.
examples/speech_recognition/data/collaters.py:60
↓ 2 callersMethodcompute_checkpoint
(self, word_index)
fairseq/modules/sparse_multihead_attention.py:37
↓ 2 callersFunctioncompute_cross_entropy_loss
Function to compute the cross entropy loss. The default value of ignore_index is the same as the default value for F.cross_entropy in pyt
fairseq/criterions/legacy_masked_lm.py:15
↓ 2 callersMethodcompute_loss
(self, model, net_output, sample, reduce=True)
fairseq/criterions/cross_entropy.py:39
↓ 2 callersMethodcompute_subset_summaries
(self, absolute_max)
fairseq/modules/sparse_multihead_attention.py:48
↓ 2 callersFunctionconsumer
(tensor)
fairseq_cli/preprocess.py:238
↓ 2 callersFunctiondataset_dest_prefix
(args, output_prefix, lang)
preprocess.py:247
↓ 2 callersFunctiondataset_dest_prefix
(args, output_prefix, lang)
fairseq_cli/preprocess.py:247
↓ 2 callersMethoddecode
(self, tokens: torch.LongTensor)
fairseq/hub_utils.py:150
↓ 2 callersMethoddecode
(self, tokens: torch.LongTensor)
fairseq/models/roberta/hub_interface.py:64
↓ 2 callersMethoddecode
(self, tokens)
examples/roberta/multiprocessing_bpe_encoder.py:103
↓ 2 callersFunctiondeprecation_warning
(message, stacklevel=3)
fairseq/utils.py:313
↓ 2 callersFunctiondest_path
(prefix, lang)
preprocess.py:42
↓ 2 callersFunctiondest_path
(prefix, lang)
fairseq_cli/preprocess.py:42
↓ 2 callersFunctioneval_lm_main
(data_dir)
tests/test_binaries.py:580
↓ 2 callersFunctionextend_conv_spec
Extends convolutional spec that is a list of tuples of 2 or 3 parameters (kernel size, dim size and optionally how many layers behind to look
fairseq/models/fconv.py:544
↓ 2 callersMethodextract_features
Similar to *forward* but only return features.
fairseq/models/fairseq_model.py:53
↓ 2 callersMethodextract_features
(self, tokens: torch.LongTensor, return_all_hiddens: bool = False)
fairseq/models/roberta/hub_interface.py:77
↓ 2 callersFunctionfile_name
(prefix, lang)
preprocess.py:36
↓ 2 callersFunctionfile_name
(prefix, lang)
fairseq_cli/preprocess.py:36
↓ 2 callersFunctionfind_span
(sentence, search_text, start=0)
examples/roberta/wsc/wsc_utils.py:61
↓ 2 callersFunctionformat_stat
(stat)
fairseq/progress_bar.py:55
↓ 2 callersFunctionget_available_dataset_impl
()
fairseq/data/indexed_dataset.py:24
↓ 2 callersMethodget_decoder_langtok
(self, tgt_lang)
fairseq/tasks/multilingual_translation.py:166
↓ 2 callersFunctionget_dummy_dictionary
(vocab_size=DEFAULT_TEST_VOCAB_SIZE)
tests/speech_recognition/asr_test_base.py:32
↓ 2 callersFunctionget_dummy_encoder_output
This only provides an example to generate dummy encoder output
tests/speech_recognition/asr_test_base.py:104
↓ 2 callersFunctionget_dummy_task_and_parser
to build a fariseq model, we need some dummy parse and task. This function is used to create dummy task and parser to faciliate model/criteri
tests/speech_recognition/asr_test_base.py:53
↓ 2 callersMethodget_embedding
Build sinusoidal embeddings. This matches the implementation in tensor2tensor, but differs slightly from the description in Section 3
fairseq/modules/sinusoidal_positional_embedding.py:37
↓ 2 callersMethodget_normalized_probs
(self, net_output, log_probs, sample=None)
fairseq/criterions/composite_loss.py:49
↓ 2 callersMethodget_normalized_probs
Get normalized probabilities (or log probs) from a net's output.
fairseq/models/fairseq_decoder.py:54
↓ 2 callersMethodget_output_path
(self, fname=None)
scripts/wav2vec_featurize.py:166
↓ 2 callersFunctionget_pairs
Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length strings).
fairseq/data/encoders/gpt2_bpe_utils.py:34
↓ 2 callersMethodget_test_sample
(self, correct, soft_target, aggregate)
tests/speech_recognition/asr_test_base.py:544
↓ 2 callersFunctionget_training_stats
(trainer)
train.py:169
↓ 2 callersFunctionget_training_stats
(trainer)
fairseq_cli/train.py:169
↓ 2 callersMethodgrad_denom
(self, sample_sizes, criterion)
fairseq/tasks/fairseq_task.py:251
↓ 2 callersMethodhas_overflow
(grad_norm)
fairseq/optim/fp16_optimizer.py:50
↓ 2 callersFunctioninfer_conv_output_dim
(conv_op, input_dim, sample_inchannel)
fairseq/modules/vggblock.py:22
↓ 2 callersFunctionkeys
(ns)
scripts/compare_namespaces.py:12
↓ 2 callersFunctionlengths_to_encoder_padding_mask
convert lengths (a 1-D Long/Int tensor) to 2-D binary tensor Args: lengths: a (B, )-shaped tensor Return: max_length: m
examples/speech_recognition/data/data_utils.py:27
↓ 2 callersFunctionlightconv_wmt_en_de_big
(args)
fairseq/models/lightconv.py:734
↓ 2 callersFunctionload_checkpoint_to_cpu
Loads a checkpoint to CPU (with upgrading for backward compatibility).
fairseq/checkpoint_utils.py:133
↓ 2 callersMethodload_dictionary
Load the dictionary from the filename Args: filename (str): the filename
fairseq/tasks/sentence_prediction.py:61
↓ 2 callersFunctionload_langpair_dataset
( data_path, split, src, src_dict, tgt, tgt_dict, combine, dataset_impl, upsample_primary,
fairseq/tasks/translation.py:20
↓ 2 callersMethodload_state_dict
Copies parameters and buffers from *state_dict* into this module and its descendants. Overrides the method in :class:`nn.Module`. Com
fairseq/models/fairseq_model.py:61
↓ 2 callersMethodlr_step_update
Update the learning rate after each update.
fairseq/trainer.py:468
↓ 2 callersFunctionmain
(args, init_distributed=False)
train.py:22
↓ 2 callersFunctionmain
(args, init_distributed=False)
fairseq_cli/train.py:22
↓ 2 callersFunctionmake_all
(lang, vocab)
preprocess.py:172
↓ 2 callersFunctionmake_all
(lang, vocab)
fairseq_cli/preprocess.py:172
↓ 2 callersMethodmake_dataset
(type, dictionary)
fairseq/tasks/sentence_ranking.py:89
↓ 2 callersMethodmax_positions
Maximum output length supported by the encoder.
fairseq/models/roberta/model.py:299
↓ 2 callersMethodmerge_file_
(self, another_file)
fairseq/data/indexed_dataset.py:312
↓ 2 callersMethodmultiply_grads
Multiplies grads by a constant ``c``.
fairseq/optim/fp16_optimizer.py:165
↓ 2 callersFunctionnorm_block
(is_layer_norm, dim, affine=True)
fairseq/models/wav2vec.py:244
↓ 2 callersFunctionoutput_doc
(i)
scripts/shard_docs.py:34
↓ 2 callersFunctionpairwise
(sents)
examples/translation_moe/score.py:123
↓ 2 callersFunctionparse_checkpoints
(files)
scripts/rm_pt.py:19
↓ 2 callersMethodprecision
(self)
fairseq/bleu.py:324
↓ 2 callersMethodprepare
(cls, args, **kargs)
fairseq/tasks/multilingual_translation.py:123
↓ 2 callersFunctionprepare_transformer_decoder_params
( input_dim, num_heads, ffn_dim, normalize_before, dropout, attention_dropout, rel
examples/speech_recognition/models/vggtransformer.py:190
↓ 2 callersFunctionprepare_transformer_encoder_params
( input_dim, num_heads, ffn_dim, normalize_before, dropout, attention_dropout, rel
examples/speech_recognition/models/vggtransformer.py:170
↓ 2 callersMethodread_data
(self, path)
fairseq/data/indexed_dataset.py:137
↓ 2 callersMethodreorder_encoder_out
(self, encoder_outs, new_order)
fairseq/sequence_generator.py:595
↓ 2 callersMethodrequire_output_path
(self, fname=None)
scripts/wav2vec_featurize.py:149
↓ 2 callersMethodreset
(self)
fairseq/data/token_block_dataset.py:205
↓ 2 callersMethodresult_string
(self, order=4)
fairseq/bleu.py:276
↓ 2 callersFunctionscalar_bias
(input, dim, bias_init=0)
fairseq/modules/scalar_bias.py:30
↓ 2 callersFunctionscore
(fdsys)
score.py:57
↓ 2 callersFunctionscore
(fdsys)
fairseq_cli/score.py:57
↓ 2 callersMethodsep
Helper to get index of sep symbol
fairseq/data/legacy/masked_lm_dictionary.py:56
↓ 2 callersMethodsetUp
(self)
tests/speech_recognition/asr_test_base.py:409
↓ 2 callersMethodset_num_updates
Set the number of parameters updates.
fairseq/trainer.py:490
↓ 2 callersMethodsize
(self, index)
fairseq/data/indexed_dataset.py:168
↓ 2 callersMethodskip
Fast-forward the iterator by skipping *num_to_skip* elements.
fairseq/data/iterators.py:46
↓ 2 callersFunctionsplit_s3_path
Split a full s3 path into the bucket name and path.
fairseq/file_utils.py:164
↓ 2 callersMethodstep_update
Update the learning rate after each update.
fairseq/optim/lr_scheduler/fixed_schedule.py:54
↓ 2 callersMethodstring
(self, tokens: torch.LongTensor)
fairseq/hub_utils.py:178
↓ 2 callersMethodtokenize
(self, sentence: str)
fairseq/hub_utils.py:155
↓ 2 callersFunctiontorch_persistent_save
(*args, **kwargs)
fairseq/checkpoint_utils.py:198
↓ 2 callersFunctiontrain_language_model
(data_dir, arch, extra_flags=None, run_validation=False)
tests/test_binaries.py:540
↓ 2 callersFunctiontrain_legacy_masked_language_model
(data_dir, arch, extra_args=())
tests/test_binaries.py:338
↓ 2 callersFunctiontransformer_lm_big
(args)
fairseq/models/transformer_lm.py:187
↓ 2 callersFunctionunfold1d
unfold T x B x C to T x B x C x K
fairseq/modules/unfold.py:9
↓ 2 callersMethodunk_string
Return unknown string, optionally escaped as: <<unk>>
fairseq/data/dictionary.py:80
↓ 2 callersFunctionupdate_sample
(doc)
scripts/split_train_valid_docs.py:30
↓ 2 callersMethodupdate_scale
(self, overflow)
fairseq/optim/fp16_optimizer.py:29
↓ 2 callersMethodupgrade_state_dict
Upgrade old state dicts to work with newer code.
fairseq/models/fairseq_model.py:71
↓ 2 callersMethodupgrade_state_dict_named
Upgrade old state dicts to work with newer code. Args: state_dict (dict): state dictionary to upgrade, in place name
fairseq/models/fairseq_model.py:75
↓ 2 callersFunctionupgrade_state_dict_with_xlm_weights
Load XLM weights into a Transformer encoder or decoder model. Args: state_dict: state dict for either TransformerEncoder or
fairseq/models/transformer_from_pretrained_xlm.py:74
↓ 2 callersFunctionvalidate
Evaluate the model on the validation set(s) and return the losses.
train.py:194
↓ 2 callersFunctionvalidate
Evaluate the model on the validation set(s) and return the losses.
fairseq_cli/train.py:194
↓ 2 callersMethodweights_for_band
(self, band: int)
fairseq/modules/adaptive_input.py:58
← previousnext →301–400 of 2,038, ranked by callers