experiment

Classes

Experiment class for training and evaluation

Module Contents

class experiment.Experiment(*, dataset, model, parameters, ith_logger, store_emb_dataframe=False, storage_path='')

Experiment class for training and evaluation

dataset

model

store_emb_dataframe

embedding_dim

num_of_epochs

learning_rate

batch_size

decay_rate

label_smoothing

optim

cuda

num_of_workers

optimizer = None

negative_label = 0.0

positive_label = 1.0

kwargs

storage_path

get_data_idxs(data)

static get_er_vocab(data)

static get_re_vocab(data)

get_batch_1_to_N(er_vocab, er_vocab_pairs, idx)

describe()

evaluate_one_to_n(model, data, log_info='Evaluate one to N.'): Evaluate model

evaluate_standard(model, data, log_info='Evaluate one to N.')

eval(model): trained model

val(model): Validation

train(model): Training.

train_and_eval(): Train and evaluate phases.

k_vs_all_training_schema(model)

negative_sampling_training_schema(model)