gbure

Graph-based approaches on unsupervised relation extraction evaluated as a fewshot problem
git clone https://esimon.eu/repos/gbure.git
Log | Files | Refs | README | LICENSE

soares_semeval.py (472B)

      1 from gbure.model.supervised import Model
      2 from torch.optim import Adam as Optimizer
      3 
      4 
      5 dataset_name = "SemEval 2010 Task 8"
      6 linguistic_similarity = "dot"
      7 
      8 # From Table 1
      9 transformer_model = "bert-base-cased"
     10 post_transformer_layer = "layer_norm"
     11 max_epoch = 10
     12 learning_rate = 3e-5
     13 accumulated_batch_size = 64
     14 
     15 # Guessed
     16 validation_metric = "half_directed_macro_f1"
     17 early_stopping_patience = 2
     18 clip_gradient = 1
     19 
     20 # Implementation details
     21 seed = 0
     22 batch_size = 8
     23 workers = 8