app.py

import torch
from peft import PeftModel, PeftConfig
import transformers
import gradio as gr

from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer, BloomForCausalLM, GenerationConfig
from transformers.models.opt.modeling_opt import OPTDecoderLayer

tokenizer = AutoTokenizer.from_pretrained('bigscience/bloom')

BASE_MODEL = "bigscience/bloom-3b"

LORA_WEIGHTS = "jslin09/LegalChatbot-bloom-3b"

config = PeftConfig.from_pretrained(LORA_WEIGHTS)

if torch.cuda.is_available():
    device = "cuda"
else:
    device = "cpu"

try:
    if torch.backends.mps.is_available():
        device = "mps"
except:
    pass

if device == "cuda":
    model = BloomForCausalLM.from_pretrained(
        BASE_MODEL,
        load_in_8bit=True,
        torch_dtype=torch.float16,
        device_map="auto",
    )
    model = PeftModel.from_pretrained(model, LORA_WEIGHTS, torch_dtype=torch.float16)
elif device == "mps":
    model = BloomForCausalLM.from_pretrained(
        BASE_MODEL,
        device_map={"": device},
        torch_dtype=torch.float16,
    )
    model = PeftModel.from_pretrained(
        model,
        LORA_WEIGHTS,
        device_map={"": device},
        torch_dtype=torch.float16,
    )
else:
    model = BloomForCausalLM.from_pretrained(
        BASE_MODEL, device_map={"": device},
        low_cpu_mem_usage=True
    )
    model = PeftModel.from_pretrained(
        model,
        LORA_WEIGHTS,
        device_map={"": device},
    )


def generate_prompt(instruction, input=None):
    if input:
        return f"""Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

### Instruction:
{instruction}

### Input:
{input}

### Response:"""
    else:
        return f"""Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
{instruction}

### Response:"""

def generate_prompt_tw(instruction, input=None):
    if input:
        return f"""以下是描述任務的指令，並與提供進一步上下文的輸入配對。編寫適當完成請求的回應。

### 指令:
{instruction}

### 輸入:
{input}

### 回應:"""
    else:
        return f"""以下是描述任務的指令。編寫適當完成請求的回應。

### 指令:
{instruction}

### 回應:"""


model.eval()
if torch.__version__ >= "2":
    model = torch.compile(model)


def evaluate(
    instruction,
    input=None,
    temperature=0.1,
    top_p=0.75,
    top_k=40,
    num_beams=4,
    max_new_tokens=128,
    **kwargs,
):
    prompt = generate_prompt(instruction, input) # 中文版的話，函數名稱要改用 generate_prompt_tw
    inputs = tokenizer(prompt, return_tensors="pt")
    input_ids = inputs["input_ids"].to(device)
    generation_config = GenerationConfig(
        temperature=temperature,
        top_p=top_p,
        top_k=top_k,
        num_beams=num_beams,
        do_sample=True,
        **kwargs,
    )
    with torch.no_grad():
        generation_output = model.generate(
            input_ids=input_ids,
            generation_config=generation_config,
            return_dict_in_generate=True,
            output_scores=True,
            max_new_tokens=max_new_tokens,
        )
    s = generation_output.sequences[0]
    output = tokenizer.decode(s)
#    return output.split("### Response:")[1].strip() # 中文版的話，要改為 return output.split("### 回應:")[1].strip()
    return output.split("### 回應:")[1].strip()


gr.Interface(
    fn=evaluate,
    inputs=[
        gr.components.Textbox(
            lines=2, label="Instruction", placeholder="Tell me about alpacas."
        ),
        gr.components.Textbox(lines=2, label="Input", placeholder="none"),
        gr.components.Slider(minimum=0, maximum=1, value=0.1, label="Temperature"),
        gr.components.Slider(minimum=0, maximum=1, value=0.75, label="Top p"),
        gr.components.Slider(minimum=0, maximum=100, step=1, value=40, label="Top k"),
        gr.components.Slider(minimum=1, maximum=4, step=1, value=4, label="Beams"),
        gr.components.Slider(
            minimum=1, maximum=2000, step=1, value=128, label="Max tokens"
        ),
    ],
    outputs=[
        gr.components.Textbox(
            lines=5,
            label="Output",
        )
    ],
    title="🌲 🌲 🌲 BLOOM-LoRA-LegalChatbot",
    description="BLOOM-LoRA-LegalChatbot is a 3B-parameter BLOOM model finetuned to follow instructions. It is trained on the [Stanford Alpaca](https://github.com/tatsu-lab/stanford_alpaca) dataset and my Legal QA dataset, and makes use of the Huggingface BLOOM implementation. For more information, please visit [the project's website](https://github.com/tloen/alpaca-lora).",
).launch()