class Inference:
def __init__(self, model_path):
self.device = torch.device(
'cuda' if torch.cuda.is_available() else 'cpu')
checkpoint = torch.load(model_path, map_location=self.device)
self.model = Controller(**checkpoint['model_args']).to(self.device)
self.model.load_state_dict(checkpoint['model'], strict=False)
self.model.eval() # Eval mode
self.tokenizer = GAPDataset.load_bert_tokenizer()
def batchify_input_text(self, doc):
"""Given an input sentence, batchify it for input.
doc: Document string
"""
doc = "[CLS] " + doc + " [SEP]"
text = self.tokenizer.tokenize(doc)
text_length = len(text)
return text, text_length
def perform_inference(self, doc):
text, text_length = self.batchify_input_text(doc)
# Convert to tensor and pass
text_length_tens = torch.tensor([len(text)]).to(self.device)
text_ids = self.tokenizer.convert_tokens_to_ids(text)
text_tens = torch.unsqueeze(torch.tensor(text_ids),
dim=0).to(self.device)
# Get output from model
outputs, _ = self.model.get_model_outputs(text_tens, text_length_tens)
coref, overwrite, ent, usage =\
(outputs['coref'], outputs['overwrite'], outputs['ent'], outputs['usage'])
output_dict = {
'text': text,
'coref': [coref[j][0].tolist() for j in range(text_length)],
'overwrite':
[overwrite[j][0].tolist() for j in range(text_length)],
'usage': [usage[j][0].tolist() for j in range(text_length)],
'ent':
['{:.3f}'.format(ent[j][0].item()) for j in range(text_length)],
}
return output_dict
