Update README.md

README.md

@@ -14,7 +14,23 @@ pipeline_tag: text-classification
 
 Detect implicit toxicity in Russian (details will be later :))
 
-
+```
+import torch
+from transformers import BertTokenizer, BertForSequenceClassification
+
+text = <your_text>
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model_name = 'arinakosovskaia/implicit_toxicity'
+tokenizer = BertTokenizer.from_pretrained(model_name)
+model = BertForSequenceClassification.from_pretrained(model_name).to(device)
+
+encoded_text = tokenizer.encode(text, return_tensors='pt').to(device)
+outputs = model(encoded_text)
+
+logits = outputs[0]
+prob = torch.nn.functional.softmax(logits, dim=1)[:, 1]
+prob.cpu().detach().numpy()[0]
+```
 
 ## Model Details