abstractive_model.py

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Load the BART tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("EE21/BART-ToSSimplify")
model = AutoModelForSeq2SeqLM.from_pretrained("EE21/BART-ToSSimplify")

# Define a function to summarize text with minimum length constraint
def summarize_with_bart(input_text, max_summary_tokens=200, min_summary_tokens=100, do_sample=False):
    # Tokenize the input text and return input_ids as PyTorch tensors
    inputs = tokenizer(input_text, return_tensors="pt").input_ids
    # Generate the summary with minimum and maximum length constraints
    outputs = model.generate(inputs, 
                             max_length=max_summary_tokens, 
                             min_length=min_summary_tokens, 
                             do_sample=do_sample)
    
    # Decode the generated token IDs back into text
    summary = tokenizer.decode(outputs[0], skip_special_tokens=True)
    
    return summary