Update app.py

app.py

@@ -30,7 +30,7 @@ def generate_text(usertitle, content, temperature, max_length, N=3):
     inputs = tokenizer(input_text, return_tensors='pt').to('cuda')
     attention_mask = torch.ones(inputs['input_ids'].shape, dtype=torch.long, device='cuda')
     generated_sequences = model.generate(inputs['input_ids'], attention_mask=attention_mask, temperature=temperature, max_length=max_length, pad_token_id=tokenizer.eos_token_id, num_return_sequences=N, do_sample=True)
-    decoded_sequences = [tokenizer.decode(g) for g in generated_sequences]
+    decoded_sequences = [tokenizer.decode(g, skip_special_tokens=True) for g in generated_sequences]
     def score(sequence):
         inputs = rm_tokenizer(sequence, return_tensors='pt', padding=True, truncation=True, max_length=512).to('cuda')
         inputs = {k: v.to('cuda') for k, v in inputs.items()}