Functions — github.com/McGill-NLP/length-generalization

hub / github.com/McGill-NLP/length-generalization / functions

Functions942 in github.com/McGill-NLP/length-generalization

⨍ Functions 942◇ Types & classes 118

↓ 1 callersMethod_create_prompt

(self, example: Dict[str, Any])

src/data/data_instance_processor/lego_instance_processor.py:54

↓ 1 callersMethod_create_prompt

(self, example: Dict[str, Any])

src/data/data_instance_processor/data_instance_processor.py:694

↓ 1 callersMethod_create_prompt

(self, example: Dict[str, Any])

src/data/data_instance_processor/addition_instance_processor.py:49

↓ 1 callersMethod_create_prompt

(self, example: Dict[str, Any])

src/data/data_instance_processor/clutrr_instance_processor.py:52

↓ 1 callersMethod_create_prompt

(self, example: Dict[str, Any])

src/data/data_instance_processor/parity_instance_processor.py:52

↓ 1 callersMethod_create_prompt

(self, example: Dict[str, Any])

src/data/data_instance_processor/sum_instance_processor.py:126

↓ 1 callersMethod_create_sbatch_launch_script

( self, compute_script_path: Path, persistent_key: str )

scripts/launcher.py:305

↓ 1 callersMethod_create_targets

(self, answer: str)

src/data/data_instance_processor/lego_instance_processor.py:70

↓ 1 callersMethod_create_targets

(self, answer: Any)

src/data/data_instance_processor/data_instance_processor.py:702

↓ 1 callersMethod_create_targets

(self, answer: int)

src/data/data_instance_processor/addition_instance_processor.py:60

↓ 1 callersMethod_create_targets

(self, answer: str)

src/data/data_instance_processor/clutrr_instance_processor.py:61

↓ 1 callersMethod_create_targets

(self, answer: str)

src/data/data_instance_processor/parity_instance_processor.py:63

↓ 1 callersMethod_create_targets

(self, answer: Any)

src/data/data_instance_processor/sum_instance_processor.py:136

↓ 1 callersFunction_dict_generator

(input_dict, prefixes=None)

src/common/py_utils.py:120

↓ 1 callersMethod_dump_config_obj

(self, config)

src/main.py:204

↓ 1 callersMethod_embed_spans

Returns the span embeddings computed in many different ways.

src/modules/span_extractors/span_extractor_with_span_width_embedding.py:161

↓ 1 callersFunction_expand_mask

Expands attention_mask from `[batch_size, src_length]` to `[batch_size, 1, tgt_length, src_length]`.

src/models/custom_t5_decoder_only.py:578

↓ 1 callersMethod_find_optimal_batch_size

Find the optimal batch size for training. Borrowed from https://github.com/Lightning-AI/lightning/blob/master/src/lightning/pytorch/tuner/batc

src/analyzers/runtime_analyzer.py:334

↓ 1 callersFunction_flatten_low_entropy

For dict, list and set objects_ calls itself on the elements and for other types assigns the object_ to the corresponding key

src/common/nest.py:204

↓ 1 callersMethod_get_batch_size

(self, stage)

src/data/base_dl_factory.py:109

↓ 1 callersMethod_get_category

(self, example: Dict[str, Any])

src/data/data_instance_processor/data_instance_processor.py:67

↓ 1 callersMethod_get_category

(self, example: Dict[str, Any])

src/data/data_instance_processor/data_instance_processor.py:706

↓ 1 callersMethod_get_config_for_manual_sweep

(self, config: JsonDict)

src/main.py:168

↓ 1 callersMethod_get_config_for_sweep

(self, config: JsonDict)

src/main.py:85

↓ 1 callersMethod_get_group_fn

(self)

src/data/s2s_dl_factory.py:382

↓ 1 callersMethod_get_hash_key

(self)

src/data/data_instance_processor/data_instance_processor.py:205

↓ 1 callersFunction_get_suggestion

(name: str, available: List[str])

src/common/registrable.py:312

↓ 1 callersMethod_get_sweep_run_id_and_dir

(self, config: JsonDict)

src/main.py:107

↓ 1 callersMethod_get_tokenize_function

( self, add_special_tokens: bool = True, is_training: bool = True )

src/data/sequence_classification_dl_factory.py:89

↓ 1 callersMethod_get_tokenize_function

( self, add_special_tokens: bool = True, is_training: bool = True )

src/data/token_classification_dl_factory.py:77

↓ 1 callersMethod_get_tokenize_function_for_decoder_only

( self, add_special_tokens: bool = True, is_training: bool = True )

src/data/sequence_classification_dl_factory.py:103

↓ 1 callersMethod_get_tokenize_function_for_decoder_only

( self, add_special_tokens: bool = True, is_training: bool = True )

src/data/s2s_dl_factory.py:329

↓ 1 callersMethod_get_tokenize_function_for_encoder_decoder

( self, add_special_tokens: bool = True )

src/data/s2s_dl_factory.py:282

↓ 1 callersMethod_get_tokenize_function_for_encoder_only

( self, add_special_tokens: bool = True, is_training: bool = True )

src/data/sequence_classification_dl_factory.py:149

↓ 1 callersFunction_is_dict_free

Returns False if obj is a dict, or if it's a list with an element that _has_dict.

src/common/params.py:152

↓ 1 callersFunction_is_encodable

We need to filter out environment variables that can't be unicode-encoded to avoid a "surrogates not allowed" error in jsonnet.

src/common/params.py:75

↓ 1 callersMethod_load_config_obj

(self, filenames: List[str])

src/main.py:211

↓ 1 callersMethod_log_examples

(self, ds: Dataset, name: str)

src/data/s2s_dl_factory.py:404

↓ 1 callersFunction_make_causal_mask

Make causal mask used for self-attention.

src/models/custom_t5_decoder_only.py:553

↓ 1 callersMethod_measure_gpu_time

( compute_fn: Callable, repeat: int, num_warm_ups: int = 20 )

src/analyzers/runtime_analyzer.py:280

↓ 1 callersMethod_patch_config_obj_for_di

(self, config)

src/main.py:198

↓ 1 callersMethod_rel_shift

(self, x)

src/models/custom_t5_decoder_only.py:207

↓ 1 callersFunction_replace_none

(params: Any)

src/common/params.py:573

↓ 1 callersMethod_replace_target_tokens

(self, targets: str)

src/data/data_instance_processor/scan_instance_processor.py:34

↓ 1 callersMethod_save_predictions

(self, state: TrainerState, tokenizer: PreTrainedTokenizer)

src/callbacks/save_predictions_callback.py:74

↓ 1 callersMethod_save_predictions

(self, state: TrainerState, tokenizer: PreTrainedTokenizer)

src/callbacks/cartography_measure_callback.py:79

↓ 1 callersMethod_to_params

Returns a dictionary of parameters that, when turned into a `Params` object and then fed to `.from_params()`, will recreate this obje

src/common/from_params.py:838

↓ 1 callersMethod_write_predictions_to_file

(self, output_test_preds_file, preds, tokenizer)

src/callbacks/save_predictions_callback.py:103

↓ 1 callersMethod_write_predictions_to_file_for_sequence_classification

( self, output_test_preds_file, preds, tokenizer )

src/callbacks/save_predictions_callback.py:137

↓ 1 callersFunctionadd_python_paths

()

scripts/create_run_script.py:7

↓ 1 callersFunctionadd_source_code

(art)

scripts/upload_experiment.py:461

↓ 1 callersFunctionadd_sys_path

(new_path)

notebooks/utils.py:2

↓ 1 callersFunctionadd_sys_path

(new_path)

src/common/wandb_utils.py:82

↓ 1 callersFunctionaggregate_group_fn

(runs: List[Run])

notebooks/plotting/training_plots.py:225

↓ 1 callersMethodanalyze

(self, config_filenames: str, load_best: bool = True)

src/runtime/seq2seq_runtime.py:1016

↓ 1 callersMethodanalyze_all_evaluation_steps

(self)

src/analyzers/seq_cls_analyzer.py:40

↓ 1 callersMethodanalyze_all_evaluation_steps

(self)

src/analyzers/seq2seq_analyzer.py:46

↓ 1 callersMethodas_ordered_dict

Returns Ordered Dict of Params from list of partial order preferences. # Parameters preference_orders: `List[List[str]]`, o

src/common/params.py:483

↓ 1 callersFunctionbatched_index_select

The given `indices` of size `(batch_size, d_1, ..., d_n)` indexes into the sequence dimension (dimension 2) of the target, which has size `(b

src/modules/span_extractors/self_attentive_span_extractor.py:79

↓ 1 callersFunctionbatched_span_select

The given `spans` of size `(batch_size, num_spans, 2)` indexes into the sequence dimension (dimension 2) of the target, which has size `(batc

src/modules/span_extractors/self_attentive_span_extractor.py:130

↓ 1 callersFunctionbucket_values

Places the given values (designed for distances) into `num_total_buckets`semi-logscale buckets, with `num_identity_buckets` of these capturin

src/modules/span_extractors/span_extractor_with_span_width_embedding.py:12

↓ 1 callersFunctionbuild_alibi_tensor

Link to paper: https://arxiv.org/abs/2108.12409 Alibi tensor is not causal as the original paper mentions, it relies on a translation invaria

src/models/custom_t5_decoder_only.py:589

↓ 1 callersMethodbuild_dataset

( self, path: Path, stage: ExperimentStage, **kwargs )

src/data/base_dl_factory.py:134

↓ 1 callersFunctioncheck_if_numbers_are_consecutive

Returns True if numbers in the list are consecutive :param list_: list of integers :return: Boolean

src/common/nest.py:19

↓ 1 callersFunctioncompute_distance_two_matrix

(m1: torch.Tensor, m2: torch.Tensor)

src/analyzers/attention_kl_analyzer.py:34

↓ 1 callersFunctioncompute_jsd_two_matrix

(m1: torch.Tensor, m2: torch.Tensor)

src/analyzers/attention_kl_analyzer.py:47

↓ 1 callersFunctioncompute_max_attention_indices

(attn_map: torch.Tensor)

src/analyzers/attention_aggregate_analyzer.py:20

↓ 1 callersFunctioncompute_mean_kl_distance_two_score_matrix

( score_matrix1: np.ndarray, score_matrix2: np.ndarray )

src/analyzers/attention_aggregate_analyzer.py:31

↓ 1 callersMethodcompute_metrics

(eval_preds: EvalPrediction)

src/data/s2s_dl_factory.py:591

↓ 1 callersFunctionconstruct_arg

The first two parameters here are only used for logging if we encounter an error.

src/common/from_params.py:446

↓ 1 callersMethodconstructor

(self)

src/common/lazy.py:61

↓ 1 callersMethodcreate_compute_script

(self, tmp_exp_dir: Path, persistent_key: str)

scripts/launcher.py:336

↓ 1 callersMethodcreate_corpus2

Creates a corpus of Shuffle-n sequences.

scripts/dataset_builders/make_dyck_pred_dataset.py:174

↓ 1 callersMethodcreate_launch_script

(self, job_body)

scripts/launcher.py:246

↓ 1 callersFunctioncreate_md5_hash

(inp: str)

src/common/py_utils.py:230

↓ 1 callersMethodcreate_scratchpad_steps_from_data_instance

( self, data_instance: Dict[str, Any] )

src/data/data_instance_processor/lego_instance_processor.py:78

↓ 1 callersMethodcreate_scratchpad_steps_from_data_instance

( self, data_instance: Dict[str, Any] )

src/data/data_instance_processor/addition_instance_processor.py:68

↓ 1 callersMethodcreate_scratchpad_steps_from_data_instance

( self, data_instance: Dict[str, Any] )

src/data/data_instance_processor/clutrr_instance_processor.py:69

↓ 1 callersMethodcreate_scratchpad_steps_from_data_instance

( self, data_instance: Dict[str, Any] )

src/data/data_instance_processor/poly_instance_processor.py:73

↓ 1 callersMethodcreate_scratchpad_steps_from_data_instance

( self, data_instance: Dict[str, Any] )

src/data/data_instance_processor/parity_instance_processor.py:71

↓ 1 callersMethodcreate_scratchpad_steps_from_data_instance

( self, data_instance: Dict[str, Any] )

src/data/data_instance_processor/sum_instance_processor.py:144

↓ 1 callersMethodcreate_scratchpad_str

( self, scratchpad_steps: List[UnifedScratchpadStep] )

src/data/data_instance_processor/data_instance_processor.py:218

↓ 1 callersFunctioncreate_sweep_config_obj

(config_filenames: List[str])

scripts/manual_sweep.py:22

↓ 1 callersMethoddet_hash_object

(self)

src/common/from_params.py:848

↓ 1 callersFunctiondownload_all_dataset_bundles

(wandb_api: wandb.Api, download_root: Path)

scripts/wandb_download_all_bundles.py:79

↓ 1 callersFunctiondownload_and_load_results

( tags: List[str], force_download: bool = False, key: str = None, )

src/common/wandb_utils.py:405

↓ 1 callersFunctiondownload_and_save_code_bundle

( bundle_id: str, wandb_api: wandb.Api, download_root: Path )

scripts/wandb_download_all_bundles.py:42

↓ 1 callersFunctiondownload_bundle

(exp_key: str, output_dir: Path)

scripts/launch_experiment.py:617

↓ 1 callersFunctiondownload_file

Download a file from a remote host to the local machine.

notebooks/plotting/attention_analysis.py:96

↓ 1 callersFunctiondownload_remote_dir

Download a file from a remote host to the local machine.

notebooks/plotting/attention_analysis_utils.py:148

↓ 1 callersFunctiondownload_the_entire_run

(localhost: str, run: Run, root_dir: Path = None)

scripts/download_nope_runs_to_local.py:121

↓ 1 callersFunctiondownload_trained_model_on_ava

(args: argparse.Namespace)

scripts/launch_experiment.py:574

↓ 1 callersFunctiondownload_trained_model_on_cc

(args: argparse.Namespace)

scripts/launch_experiment.py:562

↓ 1 callersFunctiondownload_trained_model_on_mila

(args: argparse.Namespace)

scripts/launch_experiment.py:555

↓ 1 callersMethodevaluate

(self, split: str = "test", load_best: bool = True)

src/runtime/seq2seq_runtime.py:692

↓ 1 callersFunctionevaluate_file

(filename: str, **_kwargs)

src/common/params.py:19

↓ 1 callersMethodevaluate_parsed_scratchpad

( self, predicted_scratchpad_steps: List[UnifedScratchpadStep], failed_steps: List[str

src/data/data_instance_processor/data_instance_processor.py:491

↓ 1 callersFunctionevaluate_snippet

(_filename: str, expr: str, **_kwargs)

src/common/params.py:26

↓ 1 callersMethodexecute_job

(self, job_body)

scripts/launcher.py:67

↓ 1 callersMethodextract_answer_from_final_output

(self, final_output: str)

src/data/data_instance_processor/lego_instance_processor.py:137

← previous next →201–300 of 942, ranked by callers