经过P-Tuning训练的model用于Inference

  **Usage**

    # Load the configuration and model:
    from peft import PeftModel, PeftConfig

    peft_model_id = "Laurie/t5-large_PREFIX_TUNING_SEQ2SEQ"

    config = PeftConfig.from_pretrained(peft_model_id)
    model = AutoModelForSeq2SeqLM.from_pretrained(config.base_model_name_or_path)
    model = PeftModel.from_pretrained(model, peft_model_id)

    # Get and tokenize some text about financial news:
    inputs = tokenizer(
    "Berkshire Hathaway CEO Warren Buffett on Saturday assailed regulators, politicians and the media for confusing the public about the safety of U.S. banks and said that conditions could worsen.",
    return_tensors="pt" )

    # Put the model on a GPU and generate the predicted text sentiment:
    model.to(device)

    with torch.no_grad():
        inputs = {k: v.to(device) for k, v in inputs.items()}
        outputs = model.generate(input_ids=inputs["input_ids"], max_new_tokens=10)
        print(tokenizer.batch_decode(outputs.detach().cpu().numpy(), skip_special_tokens=True))

    # => ['negative']