Spaces:
Sleeping
Sleeping
| import numpy as np | |
| import torch | |
| import requests | |
| from bs4 import BeautifulSoup | |
| if torch.cuda.is_available(): | |
| device = torch.device("cuda") | |
| 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.') | |
| import numpy as np | |
| import gradio as gr | |
| from transformers import BertTokenizer, AutoTokenizer | |
| from torch.utils.data import TensorDataset, random_split | |
| from torch.utils.data import DataLoader, RandomSampler, SequentialSampler | |
| from transformers import BertForSequenceClassification, AdamW, BertConfig | |
| import random | |
| tokenizer = AutoTokenizer.from_pretrained('armansakif/bengali-fake-news') | |
| model = BertForSequenceClassification.from_pretrained( | |
| "armansakif/bengali-fake-news", # Use the 12-layer BERT model, with an uncased vocab. | |
| num_labels = 2, # The number of output labels--2 for binary classification. | |
| # You can increase this for multi-class tasks. | |
| output_attentions = False, # Whether the model returns attentions weights. | |
| output_hidden_states = False, # Whether the model returns all hidden-states. | |
| ) | |
| # model.cuda() | |
| def extract_mainarea_content(url): | |
| response = requests.get(url) | |
| soup = BeautifulSoup(response.text, "html") | |
| mainarea_div = soup.find("div", id="mainArea") | |
| if mainarea_div: | |
| return mainarea_div.get_text() | |
| else: | |
| return "ERROR" | |
| def classify_news(news): | |
| label_list = [] | |
| input_ids = [] | |
| attention_masks = [] | |
| if "bangla.hindustantimes" in news: | |
| sent = extract_mainarea_content(news) | |
| else : | |
| sent = news | |
| label_list.append(0) | |
| encoded_dict = tokenizer.encode_plus( | |
| sent, # Sentence to encode. | |
| add_special_tokens = True, # Add '[CLS]' and '[SEP]' | |
| max_length = 512, # Pad & truncate all sentences. | |
| pad_to_max_length = True, | |
| return_attention_mask = True, # Construct attn. masks. | |
| truncation = True, | |
| return_tensors = 'pt', # Return pytorch tensors. | |
| ) | |
| input_ids.append(encoded_dict['input_ids']) | |
| attention_masks.append(encoded_dict['attention_mask']) | |
| # Convert the lists into tensors. | |
| input_ids = torch.cat(input_ids, dim=0) | |
| attention_masks = torch.cat(attention_masks, dim=0) | |
| labels = torch.tensor(label_list) | |
| testdataset = TensorDataset(input_ids, attention_masks, labels) | |
| test_dataloader = DataLoader( | |
| testdataset, # The validation samples. | |
| sampler = SequentialSampler(testdataset), # Pull out batches sequentially. | |
| batch_size = 16 # Evaluate with this batch size. | |
| ) | |
| model.eval() | |
| y_prob = [] | |
| for batch in test_dataloader: | |
| b_input_ids = batch[0].to(device) | |
| b_input_mask = batch[1].to(device) | |
| b_labels = batch[2].to(device) | |
| with torch.no_grad(): | |
| outputs = model(b_input_ids, | |
| token_type_ids=None, | |
| attention_mask=b_input_mask, | |
| labels=b_labels) | |
| loss = outputs[0] | |
| logits = outputs[1] | |
| # probability in percent code | |
| prediction_probs = torch.nn.functional.softmax(logits) | |
| y_prob.extend(prediction_probs.detach().cpu().numpy()) | |
| print(y_prob[0][0]) | |
| print(y_prob[0][1]) | |
| #------------------------------------------------------------- | |
| _, prediction = torch.max(logits, dim=1) | |
| prediction = prediction.cpu().detach().numpy() | |
| # targets = b_labels.cpu().detach().numpy() | |
| result = 'Fake News' | |
| if prediction[0] : | |
| result = 'Authentic News' | |
| print(result) | |
| labels = ['fake', 'authentic'] | |
| return {labels[i]: float(y_prob[0][i]) for i in range(2)} | |
| demo = gr.Interface( | |
| fn=classify_news, | |
| inputs=gr.Textbox(lines=10, placeholder="News here or Hindustan Times Bangla Article Link"), | |
| outputs=gr.Label(num_top_classes=2) | |
| ) | |
| demo.launch(inline=False, share=True) | |