Nelsonlin0321/alpaca-lora-7b-tuned-on-hk-cvs-fqa

Further instruct Tuning stanford alpaca based on tloen/alpaca-lora-7b on Hong Kong 2023 Consumption Voucher Scheme Frequently Asked Questions

How to use it

from transformers import LlamaForCausalLM, LlamaTokenizer,GenerationConfig
from peft import PeftModel


device_map = "auto"

tokenizer = LlamaTokenizer.from_pretrained("decapoda-research/llama-7b-hf")
model = LlamaForCausalLM.from_pretrained(
    "decapoda-research/llama-7b-hf",
    load_in_8bit=True,
    device_map="auto",
)

### load model after fine tuned on alpaca datasets
model = PeftModel.from_pretrained(model, "Nelsonlin0321/alpaca-lora-7b-tuned-on-hk-cvs-fqa")

tokenizer = LlamaTokenizer.from_pretrained("decapoda-research/llama-7b-hf")
tokenizer.pad_token_id = 0


def generate_prompt_eval(instruction):
    template =  f"""Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{instruction}
### Response:"""
    return template

eval_generation_config = GenerationConfig(
    temperature=0.1,
    top_p=0.75,
    num_beams=4,
)


def generate_answer(instruction):
    prompt = generate_prompt_eval(instruction)
    inputs = tokenizer(prompt, return_tensors="pt")
    input_ids = inputs["input_ids"].cuda()
    generation_output = model.generate(
        input_ids=input_ids,
        generation_config=eval_generation_config,
        return_dict_in_generate=True,
        output_scores=True,
        max_new_tokens=256
    )
    for s in generation_output.sequences:
        output = tokenizer.decode(s)
        # print(output)
        print("Response:", output.split("### Response:")[1].strip())


question = "Who are eligible to be disbursed with the first-instalment voucher of $1,500 on 16 April?"

generate_answer(question)
>> Response: All eligible people who have successfully registered under 2022 CVS and met the relevant eligibility criteria will be disbursed with the first-instalment voucher of $1,500 on 16 April.