MiniGPT-4/demo.py

172 lines
6.3 KiB
Python
Raw Permalink Normal View History

2023-04-16 22:04:16 +00:00
import argparse
import os
import random
import numpy as np
import torch
import torch.backends.cudnn as cudnn
import gradio as gr
2023-10-13 00:14:35 +00:00
from transformers import StoppingCriteriaList
2023-04-16 22:04:16 +00:00
from minigpt4.common.config import Config
from minigpt4.common.dist_utils import get_rank
from minigpt4.common.registry import registry
2023-10-13 00:14:35 +00:00
from minigpt4.conversation.conversation import Chat, CONV_VISION_Vicuna0, CONV_VISION_LLama2, StoppingCriteriaSub
2023-04-16 22:04:16 +00:00
# imports modules for registration
from minigpt4.datasets.builders import *
from minigpt4.models import *
from minigpt4.processors import *
from minigpt4.runners import *
from minigpt4.tasks import *
def parse_args():
parser = argparse.ArgumentParser(description="Demo")
parser.add_argument("--cfg-path", required=True, help="path to configuration file.")
parser.add_argument("--gpu-id", type=int, default=0, help="specify the gpu to load the model.")
2023-04-16 22:04:16 +00:00
parser.add_argument(
"--options",
nargs="+",
help="override some settings in the used config, the key-value pair "
"in xxx=yyy format will be merged into config file (deprecate), "
"change to --cfg-options instead.",
)
args = parser.parse_args()
return args
def setup_seeds(config):
seed = config.run_cfg.seed + get_rank()
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
cudnn.benchmark = False
cudnn.deterministic = True
# ========================================
# Model Initialization
# ========================================
2023-08-28 18:26:00 +00:00
conv_dict = {'pretrain_vicuna0': CONV_VISION_Vicuna0,
'pretrain_llama2': CONV_VISION_LLama2}
2023-04-16 22:04:16 +00:00
print('Initializing Chat')
2023-04-19 16:49:05 +00:00
args = parse_args()
cfg = Config(args)
2023-04-16 22:04:16 +00:00
model_config = cfg.model_cfg
2023-04-19 16:49:05 +00:00
model_config.device_8bit = args.gpu_id
2023-04-16 22:04:16 +00:00
model_cls = registry.get_model_class(model_config.arch)
2023-04-19 16:49:05 +00:00
model = model_cls.from_config(model_config).to('cuda:{}'.format(args.gpu_id))
2023-04-16 22:04:16 +00:00
2023-08-28 18:26:00 +00:00
CONV_VISION = conv_dict[model_config.model_type]
2023-04-16 22:04:16 +00:00
vis_processor_cfg = cfg.datasets_cfg.cc_sbu_align.vis_processor.train
vis_processor = registry.get_processor_class(vis_processor_cfg.name).from_config(vis_processor_cfg)
2023-10-13 00:14:35 +00:00
stop_words_ids = [[835], [2277, 29937]]
stop_words_ids = [torch.tensor(ids).to(device='cuda:{}'.format(args.gpu_id)) for ids in stop_words_ids]
stopping_criteria = StoppingCriteriaList([StoppingCriteriaSub(stops=stop_words_ids)])
chat = Chat(model, vis_processor, device='cuda:{}'.format(args.gpu_id), stopping_criteria=stopping_criteria)
2023-04-16 22:04:16 +00:00
print('Initialization Finished')
2023-08-28 18:26:00 +00:00
2023-04-16 22:04:16 +00:00
# ========================================
# Gradio Setting
# ========================================
2023-08-28 18:26:00 +00:00
2023-04-16 22:04:16 +00:00
def gradio_reset(chat_state, img_list):
2023-04-17 07:35:02 +00:00
if chat_state is not None:
chat_state.messages = []
if img_list is not None:
img_list = []
2023-04-16 22:04:16 +00:00
return None, gr.update(value=None, interactive=True), gr.update(placeholder='Please upload your image first', interactive=False),gr.update(value="Upload & Start Chat", interactive=True), chat_state, img_list
2023-08-28 18:26:00 +00:00
2023-04-16 22:04:16 +00:00
def upload_img(gr_img, text_input, chat_state):
if gr_img is None:
2023-04-17 07:35:02 +00:00
return None, None, gr.update(interactive=True), chat_state, None
2023-04-16 22:04:16 +00:00
chat_state = CONV_VISION.copy()
img_list = []
llm_message = chat.upload_img(gr_img, chat_state, img_list)
2023-10-13 00:14:35 +00:00
chat.encode_img(img_list)
2023-04-16 22:04:16 +00:00
return gr.update(interactive=False), gr.update(interactive=True, placeholder='Type and press Enter'), gr.update(value="Start Chatting", interactive=False), chat_state, img_list
2023-08-28 18:26:00 +00:00
2023-04-16 22:04:16 +00:00
def gradio_ask(user_message, chatbot, chat_state):
if len(user_message) == 0:
return gr.update(interactive=True, placeholder='Input should not be empty!'), chatbot, chat_state
chat.ask(user_message, chat_state)
chatbot = chatbot + [[user_message, None]]
return '', chatbot, chat_state
def gradio_answer(chatbot, chat_state, img_list, num_beams, temperature):
llm_message = chat.answer(conv=chat_state,
img_list=img_list,
num_beams=num_beams,
temperature=temperature,
max_new_tokens=300,
max_length=2000)[0]
2023-04-16 22:04:16 +00:00
chatbot[-1][1] = llm_message
return chatbot, chat_state, img_list
2023-08-28 18:26:00 +00:00
2023-04-16 22:04:16 +00:00
title = """<h1 align="center">Demo of MiniGPT-4</h1>"""
description = """<h3>This is the demo of MiniGPT-4. Upload your images and start chatting!</h3>"""
2023-04-18 09:22:28 +00:00
article = """<p><a href='https://minigpt-4.github.io'><img src='https://img.shields.io/badge/Project-Page-Green'></a></p><p><a href='https://github.com/Vision-CAIR/MiniGPT-4'><img src='https://img.shields.io/badge/Github-Code-blue'></a></p><p><a href='https://raw.githubusercontent.com/Vision-CAIR/MiniGPT-4/main/MiniGPT_4.pdf'><img src='https://img.shields.io/badge/Paper-PDF-red'></a></p>
2023-04-16 22:04:16 +00:00
"""
#TODO show examples below
with gr.Blocks() as demo:
gr.Markdown(title)
gr.Markdown(description)
gr.Markdown(article)
with gr.Row():
2023-10-13 00:14:35 +00:00
with gr.Column(scale=1):
2023-04-16 22:04:16 +00:00
image = gr.Image(type="pil")
upload_button = gr.Button(value="Upload & Start Chat", interactive=True, variant="primary")
clear = gr.Button("Restart")
num_beams = gr.Slider(
minimum=1,
maximum=10,
value=1,
2023-04-16 22:04:16 +00:00
step=1,
interactive=True,
label="beam search numbers)",
)
temperature = gr.Slider(
minimum=0.1,
maximum=2.0,
value=1.0,
step=0.1,
interactive=True,
label="Temperature",
)
2023-10-13 00:14:35 +00:00
with gr.Column(scale=2):
2023-04-16 22:04:16 +00:00
chat_state = gr.State()
img_list = gr.State()
chatbot = gr.Chatbot(label='MiniGPT-4')
text_input = gr.Textbox(label='User', placeholder='Please upload your image first', interactive=False)
upload_button.click(upload_img, [image, text_input, chat_state], [image, text_input, upload_button, chat_state, img_list])
text_input.submit(gradio_ask, [text_input, chatbot, chat_state], [text_input, chatbot, chat_state]).then(
gradio_answer, [chatbot, chat_state, img_list, num_beams, temperature], [chatbot, chat_state, img_list]
)
clear.click(gradio_reset, [chat_state, img_list], [chatbot, image, text_input, upload_button, chat_state, img_list], queue=False)
2023-04-18 09:22:28 +00:00
demo.launch(share=True, enable_queue=True)