A
Flan-T5 Large model fine-tuned with QLoRA (8-bit quantization) for summarizing US Congressional bills using the
BillSum dataset.
1from peft import PeftModel
2from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
3
4base_model = AutoModelForSeq2SeqLM.from_pretrained("google/flan-t5-large", load_in_8bit=True)
5model = PeftModel.from_pretrained(base_model, "juliensimon/flan-t5-large-billsum-qlora")
6tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-large")
7
8text = "Summarize: " + bill_text
9inputs = tokenizer(text, return_tensors="pt", max_length=512, truncation=True)
10outputs = model.generate(**inputs, max_new_tokens=150)
11print(tokenizer.decode(outputs[0], skip_special_tokens=True))
