NLP Summarization

summary.py

input_ids = tokenizer('summarize: ' + text.lower(), 
                      return_tensors='pt').input_ids.to(model.device)

output = model.generate(
    input_ids,
    max_length=200,
    num_beams=8,
    num_beam_groups=4,  # based on this paper, https://arxiv.org/pdf/1610.02424.pdf
    no_repeat_ngram_size=2
)

generated = tokenizer.decode(output[0], skip_special_tokens=True)
print(generated)

# >>> chalmers university of technology is a swedish university located in gothenburg. the university was founded in 1829 after he donated part of his fortune to the establishment of an "industrial school" the school is one of only three universities named after an individual, the other two being karolinska institutet and linnaeus university.
# >>> {'rouge1': Score(precision=0.9615384615384616, recall=0.37593984962406013, fmeasure=0.5405405405405406),
# >>> 'rouge2': Score(precision=0.7647058823529411, recall=0.29545454545454547, fmeasure=0.42622950819672134),
# >>> 'rougeL': Score(precision=0.9230769230769231, recall=0.3609022556390977, fmeasure=0.518918918918919)}