Functions — github.com/BlackHC/BatchBALD

hub / github.com/BlackHC/BatchBALD / functions

Functions305 in github.com/BlackHC/BatchBALD

⨍ Functions 305◇ Types & classes 28

↓ 9 callersFunction_vgg

(arch, cfg, batch_norm, pretrained, progress, pretrained_features_only=False, **kwargs)

src/vgg_model.py:111

↓ 6 callersMethodacquire

(self, available_indices)

src/active_learning_data.py:29

↓ 6 callersFunctionget_target_bins

(dataset)

src/dataset_enum.py:392

↓ 6 callersFunctionsplit_tensors

(output, input, chunk_size)

src/torch_utils.py:221

↓ 5 callersMethodget_dataset_indices

(self, available_indices: List[int])

src/active_learning_data.py:25

↓ 5 callersFunctionget_targets

Get the targets of a dataset without any target target transforms(!).

src/dataset_enum.py:431

↓ 3 callersMethod_update_indices

(self)

src/active_learning_data.py:21

↓ 3 callersFunctionbalance_dataset_by_repeating

(dataset, num_classes, target_size, upsample=True)

src/dataset_enum.py:398

↓ 3 callersFunctionbasic_exact_joint_entropy

(logits_N_K_C)

src/joint_entropy/test_joint_entropy.py:12

↓ 3 callersFunctionbatch_multi_choices

probs_b_C: Ni... x C Returns: choices: Ni... x M

src/torch_utils.py:187

↓ 3 callersFunctiondesc

(name)

src/recover_model.py:117

↓ 3 callersFunctiongather_expand

(data, dim, index)

src/torch_utils.py:203

↓ 3 callersMethodget_data_source

(self)

src/dataset_enum.py:99

↓ 3 callersFunctionget_samples_values_I

:returns a list of (samples, values) tuples. accuracy refers to the accuracy after having added samples to the training set.

src/al_notebook/results_loader.py:289

↓ 3 callersFunctionhandle_unary_funcs

(pred_kv, pred_k, pred_v, default=None)

src/al_notebook/results_loader.py:103

↓ 3 callersFunctionignite_progress_bar

(engine: ignite.engine.Engine, desc=None, log_interval=0)

src/ignite_progress_bar.py:35

↓ 3 callersFunctionload_laaos_files

(path=None, files=None, vanilla=False, tag=None, prefix=None)

src/al_notebook/results_loader.py:200

↓ 3 callersFunctionmap_dict

(d: dict, *, kv=None, k=None, v=None)

src/al_notebook/results_loader.py:126

↓ 3 callersFunctionparse_enum_str

(enum_str: str, enum_cls)

src/recover_model.py:47

↓ 3 callersFunctionparse_enum_str

(enum_str: str, enum_cls)

src/al_notebook/results_loader.py:159

↓ 3 callersMethodtrain_model

( self, train_loader, test_loader, validation_loader,

src/dataset_enum.py:212

↓ 2 callersFunctionSubrangeDataset

(dataset, begin, end)

src/subrange_dataset.py:6

↓ 2 callersMethod_create_mask

(self, input, k)

src/mc_dropout.py:96

↓ 2 callersFunction_get_cuda_assumed_available_memory

()

src/torch_utils.py:25

↓ 2 callersFunctionbuild_metrics

()

src/train_model.py:18

↓ 2 callersFunctioncompose_transformers

(iterable)

src/dataset_enum.py:382

↓ 2 callersFunctiondiscard_eng_args

(args_dict)

src/al_notebook/results_loader.py:456

↓ 2 callersFunctionentropy

(logits, dim: int, keepdim: bool = False)

src/torch_utils.py:95

↓ 2 callersFunctionexpand_samples_I_values_I

Subsample a list of accuracies as if an accuracy were acquired after each sample. Provides optimistic subsamples. (The finaly accuracy is attribu

src/al_notebook/results_loader.py:302

↓ 2 callersMethodextract_dataset

Extract a dataset randomly from the available dataset and make those indices unavailable.

src/active_learning_data.py:47

↓ 2 callersMethodextract_dataset_from_indices

Extract a dataset from the available dataset and make those indices unavailable.

src/active_learning_data.py:51

↓ 2 callersFunctionfill_values_sample_points_T

Ensures that each list of accuracies in accuracies_T has the same length. :param values_sample_points_T: tuple of list of accuracies and sample_p

src/al_notebook/results_loader.py:326

↓ 2 callersFunctionget_balanced_sample_indices

(target_classes: typing.List, num_classes, n_per_digit=2)

src/torch_utils.py:121

↓ 2 callersFunctionget_experiment_data

( data_source, num_classes, initial_samples, reduced_dataset, samples_

src/dataset_enum.py:245

↓ 2 callersMethodget_random_available_indices

(self, size)

src/active_learning_data.py:42

↓ 2 callersFunctionis_cuda_out_of_memory

(exception)

src/torch_utils.py:54

↓ 2 callersFunctionlog_epoch_results

(engine: ignite.engine.Engine, name, trainer: ignite.engine.Engine)

src/ignite_utils.py:19

↓ 2 callersFunctionmerge_args

(stores)

src/al_notebook/results_loader.py:419

↓ 2 callersFunctionmerge_sample_points_T

Merge sample_points_T into just one list. :returns a list of sample_points.

src/al_notebook/results_loader.py:341

↓ 2 callersFunctionp

(device, shape, size_MB, name)

src/al_notebook/torch_utils.py:24

↓ 2 callersFunctionplot_aggregated_values

( grouped: Dict[object, rl.AggregateAccuracies], key2text=None, axes=None, show_num_trials=Tru

src/al_notebook/plots.py:9

↓ 2 callersFunctionprint_cuda_info

()

src/al_notebook/torch_utils.py:8

↓ 2 callersFunctionreduced_eval_consistent_bayesian_model

Performs a scoring step with k inference samples while reducing the dataset to at most min_remaining_percentage. Before computing anything at all

src/reduced_consistent_mc_sampler.py:26

↓ 2 callersMethodrestore_best

(self)

src/ignite_restoring_score_guard.py:78

↓ 2 callersMethodset_dropout_p

(self, p)

src/mc_dropout.py:46

↓ 2 callersFunctionstore_epoch_results

(engine: ignite.engine.Engine, store_object, name=None)

src/ignite_utils.py:47

↓ 2 callersMethodunflatten_tensor

(input: torch.Tensor, k: int)

src/mc_dropout.py:54

↓ 1 callersMethod__init__

(self, bayesian_net: mc_dropout.BayesianModule, k)

src/sampler_model.py:74

↓ 1 callersMethod__init__

(self)

src/mc_dropout.py:71

↓ 1 callersMethod_get_sample_mask_shape

(self, sample_shape)

src/mc_dropout.py:93

↓ 1 callersMethodacquire_batch

( self, bayesian_model: nn.Module, acquisition_function: AcquisitionFunction,

src/acquisition_method.py:15

↓ 1 callersFunctionaggregate_values

(stores, values_getter, percentiles=None, thresholds=None)

src/al_notebook/results_loader.py:389

↓ 1 callersFunctionaggregate_values_sample_points_T

(values_sample_points_T, percentiles=None, thresholds=None)

src/al_notebook/results_loader.py:349

↓ 1 callersMethodattach

(self, engine: ignite.engine.Engine)

src/ignite_progress_bar.py:12

↓ 1 callersFunctionbatch_exact_joint_entropy

This one switches between devices, too.

src/multi_bald.py:171

↓ 1 callersFunctioncamel_case_name

(snake_case_name)

src/al_notebook/results_loader.py:46

↓ 1 callersFunctionchain

(engine: ignite.engine.Engine, then_engine: ignite.engine.Engine, dataloader: data.DataLoader)

src/ignite_utils.py:13

↓ 1 callersMethodcompute_scores

(self, logits_B_K_C, available_loader, device)

src/acquisition_functions.py:56

↓ 1 callersFunctionconditional_entropy_from_logits_B_K_C

(logits_B_K_C)

src/joint_entropy/exact.py:83

↓ 1 callersMethodcreate_bayesian_model

(self, device)

src/dataset_enum.py:185

↓ 1 callersFunctioncreate_experiment_config_argparser

(parser)

src/run_experiment.py:25

↓ 1 callersMethodcreate_optimizer

(self, model)

src/dataset_enum.py:202

↓ 1 callersMethodcreate_train_model_extra_args

(self, optimizer)

src/dataset_enum.py:209

↓ 1 callersMethoddeterministic_forward_impl

(self, input: torch.Tensor)

src/mc_dropout.py:40

↓ 1 callersFunctiondiff_args

(stores)

src/al_notebook/results_loader.py:480

↓ 1 callersFunctionentropy_from_probs_b_M_C

(probs_b_M_C)

src/joint_entropy/exact.py:42

↓ 1 callersFunctionentropy_joint_probs_B_M_C

(probs_B_K_C, prev_joint_probs_M_K)

src/joint_entropy/exact.py:47

↓ 1 callersFunctionepoch_chain

(engine: ignite.engine.Engine, then_engine: ignite.engine.Engine, dataloader: data.DataLoader)

src/ignite_utils.py:7

↓ 1 callersFunctionfix_chosen_samples

(chosen_samples)

src/al_notebook/results_loader.py:237

↓ 1 callersMethodflatten_tensor

(mc_input: torch.Tensor)

src/mc_dropout.py:59

↓ 1 callersFunctiongather_accuracy

Gathers all accuracy values from the iterations in a store. :returns an array of accuracies.

src/al_notebook/results_loader.py:225

↓ 1 callersFunctiongather_samples_I

Gathers all samples in a store. :returns a list of list of samples.

src/al_notebook/results_loader.py:241

↓ 1 callersFunctiongather_values

(stores, values_getter)

src/al_notebook/results_loader.py:400

↓ 1 callersFunctiongc_cuda

()

src/torch_utils.py:16

↓ 1 callersFunctionget_CINIC10

(root="./")

src/dataset_enum.py:44

↓ 1 callersFunctionget_MNIST

()

src/dataset_enum.py:70

↓ 1 callersFunctionget_base_indices

(dataset: Dataset, indices: typing.List[int])

src/torch_utils.py:151

↓ 1 callersFunctionget_cuda_blocked_memory

()

src/torch_utils.py:34

↓ 1 callersFunctionget_cuda_total_memory

()

src/torch_utils.py:21

↓ 1 callersFunctionget_laaos_files

(laaos_dir=None)

src/al_notebook/results_loader.py:182

↓ 1 callersFunctionget_marks

(values_S, threshold)

src/al_notebook/results_loader.py:265

↓ 1 callersFunctionget_merge_args_field

(store, field_name)

src/al_notebook/results_loader.py:445

↓ 1 callersFunctionget_samples_from_laaos_store

(laaos_store, target_iteration=None)

src/recover_model.py:36

↓ 1 callersFunctionget_subset_base_indices

(dataset: Subset, indices: typing.List[int])

src/torch_utils.py:147

↓ 1 callersFunctionhandle_map_funcs

(func_kv, func_k, func_v, default=None)

src/al_notebook/results_loader.py:80

↓ 1 callersFunctionimportance_weighted_entropy_p_b_M_C

(p_b_M_C, q_1_M_1, M: int)

src/joint_entropy/sampling.py:88

↓ 1 callersFunctionindex_of_first

Gets the index of the first element in `iter` that satifies `pred`.

src/al_notebook/results_loader.py:257

↓ 1 callersFunctionis_cudnn_snafu

(exception)

src/torch_utils.py:60

↓ 1 callersFunctionjoint_probs_M_K_impl

(probs_N_K_C, prev_joint_probs_M_K)

src/joint_entropy/exact.py:7

↓ 1 callersFunctionkey2text

(name, store)

src/al_notebook/results_loader.py:587

↓ 1 callersFunctionload_logits

()

src/joint_entropy/test_joint_entropy.py:34

↓ 1 callersFunctionlogit_mean

r"""Computes $\log \left ( \frac{1}{n} \sum_i p_i \right ) = \log \left ( \frac{1}{n} \sum_i e^{\log p_i} \right )$. We pass in logits.

src/torch_utils.py:85

↓ 1 callersFunctionmain

()

src/run_experiment_no_al.py:25

↓ 1 callersFunctionmain

()

src/run_experiment.py:127

↓ 1 callersFunctionmake_layers

(cfg, batch_norm=False)

src/vgg_model.py:83

↓ 1 callersMethodmake_unavailable

(self, available_indices)

src/active_learning_data.py:36

↓ 1 callersFunctionmap_field_name

(field_name)

src/al_notebook/results_loader.py:536

↓ 1 callersFunctionmap_field_value

(field_name, value)

src/al_notebook/results_loader.py:548

↓ 1 callersMethodmc_forward_impl

(self, mc_input_BK: torch.Tensor)

src/mc_dropout.py:43

↓ 1 callersMethodmc_tensor

(input: torch.tensor, k: int)

src/mc_dropout.py:63

next →1–100 of 305, ranked by callers