ready.py

#필요 패키지 설치
!pip install mxnet
!pip install gluonnlp==0.8.0
!pip install tqdm pandas
!pip install sentencepiece
!pip install transformers
!pip install torch
!pip install numpy==1.23.1

#KoBERT 깃허브에서 불러오기
!pip install 'git+https://github.com/SKTBrain/KoBERT.git#egg=kobert_tokenizer&subdirectory=kobert_hf'

!pip install langchain==0.0.125 chromadb==0.3.14 pypdf==3.7.0 tiktoken==0.3.3
!pip install openai==0.28
!pip install gradio transformers torch opencv-python-headless

# import torch
from torch import nn
import torch.nn.functional as F
import torch.optim as optim
from torch.utils.data import Dataset, DataLoader
import gluonnlp as nlp
import numpy as np
from tqdm import tqdm, tqdm_notebook
import pandas as pd

#  Hugging Face를 통한 모델 및 토크나이저 Import
from kobert_tokenizer import KoBERTTokenizer
from transformers import BertModel

from transformers import AdamW
from transformers.optimization import get_cosine_schedule_with_warmup

n_devices = torch.cuda.device_count()
print(n_devices)

for i in range(n_devices):
    print(torch.cuda.get_device_name(i))

if torch.cuda.is_available():
    device = torch.device("cuda")
    print('There are %d GPU(s) available.' % torch.cuda.device_count())
    print('We will use the GPU:', torch.cuda.get_device_name(0))
else:
    device = torch.device("cpu")
    print('No GPU available, using the CPU instead.')


# Kobert_softmax
class BERTClassifier(nn.Module):
    def __init__(self,
                 bert,
                 hidden_size=768,
                 num_classes=6,
                 dr_rate=None,
                 params=None):
        super(BERTClassifier, self).__init__()
        self.bert = bert
        self.dr_rate = dr_rate
        self.softmax = nn.Softmax(dim=1)  # Softmax로 변경
        self.classifier = nn.Sequential(
            nn.Dropout(p=0.5),
            nn.Linear(in_features=hidden_size, out_features=512),
            nn.Linear(in_features=512, out_features=num_classes),
        )

        # 정규화 레이어 추가 (Layer Normalization)
        self.layer_norm = nn.LayerNorm(768)

        # 드롭아웃
        self.dropout = nn.Dropout(p=dr_rate)

    def gen_attention_mask(self, token_ids, valid_length):
        attention_mask = torch.zeros_like(token_ids)
        for i, v in enumerate(valid_length):
            attention_mask[i][:v] = 1
        return attention_mask.float()

    def forward(self, token_ids, valid_length, segment_ids):
        attention_mask = self.gen_attention_mask(token_ids, valid_length)
        _, pooler = self.bert(input_ids=token_ids, token_type_ids=segment_ids.long(), attention_mask=attention_mask.float().to(token_ids.device))

        pooled_output = self.dropout(pooler)
        normalized_output = self.layer_norm(pooled_output)
        out = self.classifier(normalized_output)

        # LayerNorm 적용
        pooler = self.layer_norm(pooler)

        if self.dr_rate:
            pooler = self.dropout(pooler)

        logits = self.classifier(pooler)  # 분류를 위한 로짓 값 계산
        probabilities = self.softmax(logits)  # Softmax로 각 클래스의 확률 계산
        return probabilities  # 각 클래스에 대한 확률 반환

#정의한 모델 불러오기
model = BERTClassifier(bertmodel,dr_rate=0.4).to(device)
#model = BERTClassifier(bertmodel,  dr_rate=0.5).to('cpu')

# Prepare optimizer and schedule (linear warmup and decay)
no_decay = ['bias', 'LayerNorm.weight']
optimizer_grouped_parameters = [
    {'params': [p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
    {'params': [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
]
optimizer = AdamW(optimizer_grouped_parameters, lr=learning_rate)
loss_fn = nn.CrossEntropyLoss()
t_total = len(train_dataloader) * num_epochs
warmup_step = int(t_total * warmup_ratio)
scheduler = get_cosine_schedule_with_warmup(optimizer, num_warmup_steps=warmup_step, num_training_steps=t_total)
def calc_accuracy(X,Y):
    max_vals, max_indices = torch.max(X, 1)
    train_acc = (max_indices == Y).sum().data.cpu().numpy()/max_indices.size()[0]
    return train_acc
train_dataloader

model = torch.load('./model_weights_softmax(model).pth')
model.eval()

# 멜론 데이터 불러오기 

melon_data = pd.read_csv('./melon_data.csv')
melon_emotions = pd.read_csv('./melon_emotions_final.csv')
melon_emotions = pd.merge(melon_emotions, melon_data, left_on='Title', right_on='title', how='inner')
melon_emotions = melon_emotions[['singer', 'Title', 'genre','Emotions']]
melon_emotions = melon_emotions.drop_duplicates(subset='Title', keep='first')
melon_emotions['Emotions'] = melon_emotions['Emotions'].apply(lambda x: ast.literal_eval(x))

emotions = melon_emotions['Emotions'].to_list()

#gradio
!pip install --upgrade gradio
import numpy as np
import pandas as pd
import requests
from PIL import Image
import torch
from transformers import AutoProcessor, AutoModelForZeroShotImageClassification, pipeline
import gradio as gr
import openai
from sklearn.metrics.pairwise import cosine_similarity
import ast